Sam's Laser FAQ, Copyright © 1994-2022, Samuel M. Goldwasser, All Rights Reserved.
I may be contacted via the
Sci.Electronics.Repair FAQ Email Links Page.

  • Back to Sam's Laser FAQ Table of Contents.

    Commercial Stabilized HeNe Lasers

    Sub-Table of Contents

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.
  • Forward to HeNe Laser Testing, Adjustment, Repair .


    This chapter deals with stabilized Helium-Neon (HeNe) lasers that are now available or have been in the past from manufacturers like Melles Griot and Spectra-Physics. Such lasers have their frequency actively regulated either directly, or indirectly via control of intensity. Their output is generally either a single longitudinal mode (single optical frequency), a pair of adjacent longitudinal modes (two optical frequencies separated by 600 MHz to 1 GHz typical), or a pair of optical frequencies close together produced by Zeeman splitting of a single longitudinal mode (250 kHz to 8 Mhz typical) or an Acousto-Optic Modulator (AOM) generating the second frequency as a sideband of a single longitudinal mode (20 MHz typical).

    Unstabilized HeNe lasers - the more conventional type that most people are familiar with from Physics 101 - are covered in the chapter: Commercial HeNe Lasers.

    For current production lasers, the manufacturers' Web sites often provide basic specifications. A Google search is usually the easiest way to find them, but most are also linked from the chapter: Laser and Parts Sources. For older lasers, it's often difficult to obtain detailed specs so estimates based on physical size, and then testing may be the only option.

    Much more information on stabilized HeNe lasers and typical locking schemes can be found in the chapter: Stabilized Helium-Neon Lasers.

    Parameter Conversions

    Nearly all of these lasers red HeNes operating at ~633 nm. Here are the common conversions between dimensions, optical frequency, and wavelength. These are useful for small changes in the vicinity of 633 nm. Abbreviations: ppm - parts per million, ppb - parts per billion, pm - nanometer (10-9m), pm - picometer (10-12m), fm - femtometer (10-15m):

    And dimensions:

    Comparison Chart of Commercial Stabilized HeNe Lasers

    Here is a summary of the frequency stability specifications for most of the stabilized HeNe lasers that are presently commercially available, or have been in the past. The best of the mode-stabilized lasers may approach the performance of a much more complex and expensive iodine-stabilied HeNe laser (also included in this chart).

    Nearly every model of stabilized HeNe laser ever sold commercially is listed in the chart below. Many are described in separate sections of this chapter, arranged approximately in alphabetical order by manufacturer. For these, the level of detail here is probably several orders of magnitude greater than from any other source, except perhaps the operation and service manual for the laser (which with few exceptions, is generally not available in the public domain). Where there is no entry for a particular laser, a Google search using the manufacturer (with or without model number) will usually locate what little information is available. Sometimes, a research paper referencing the specific laser will even have the most information!

    Unless otherwise noted (below), these data were obtained from manufacturers' Web sites, brochures, spec sheets, or user manuals for each laser. Contributions (including stabilized HeNe lasers I've missed) and corrections are welcome. Please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    All stability values are in parts per billion (ppb). For reference, 1 ppb is approximately 474 kHz or 0.000633 pm.

                                         <--- Frequency Stability Time Scale --->
        Model                    Type/AP  Sec    Min   Hour    Day   Year    Life
      Aerotech DF170               AZ M                 ±2  ±2 (8 hr) ±20 (1 mo)
     *Aerotech LZR2000 (20)        SM M                 ±2            ±20 (1 mo)
     +Aerotech LZR3000 (20)        SM M                 ±2            ±20 (1 mo)
      Aerotech OEM05SF/105SF       SM S                 ±2  ±2 (8 hr) ±20 (1 mo)
      Aerotech OEM05SFX/105SFX     SM S                 ±2  ±2 (8 hr) ±20 (1 mo)
      Aerotech OEM1SF/110SF        SM S                 ±2  ±2 (8 hr) ±20 (1 mo)
     *Aerotech S100 (2)            SM S          ±1     ±2  ±3 (8 hrs)
      API XD Laser                 SM M     ±200 (unspecified time)
     *Axsys 150 (8)                DM M   ±2     ±6            ±20
      CDHC-Optics DH-HN250P        SM S
     *Coherent 200                 DM S          ±2 (5 min)          ±10 (long)
     *Excel 1001A/B/F              AZ M        20 (unspecified time)
      Feanor LN 10                 DM M                 ±1                   ±80
      Feanor LP 30                 AZ M                 ±2                   ±20
      Feanor LSP 30                AZ M                 ±2                   ±20
     +Frazier 100 (21)             I2 S                                     ±0.01
     *General Dynamics 150 (8)     DM M   ±2     ±6            ±20
     *HP-5500A/B/C, 5501A/B        AZ M                        ±2            ±20
     *HP/Agilent 5517 (all)        AZ M                        ±2            ±20
     *HP/Agilent 5518A, 5519A/B    AZ M                        ±2            ±20
     *HP/Agilent Z4214A (22)       AZ M          ±5 (5 min)
     *HP/Agilent N1211A (23)       AZ M          ±5 (5 min)
      JDS Uniphase 1410-1 (25)     DM S          ±0.05 (???)
      JDS Uniphase 1410-2 (25)     DM S          ±0.05 (???)
      JDS Uniphase 1420 (25)       DM S          ±0.05 (???)
      JENAer ZL 600 (10)           DM M                2.0                    20
      JENAer ZL 700 (10)           DM M                2.0                    20
      JENAer ZL 800 (10)           DM M                2.0                    20
     *Lab for Science 200          DM S  0.03   0.05   0.2     0.5
     *Lab for Science 210          DM S  0.03   0.05   0.2     0.5
     *Lab for Science 220          TZ S  0.01   0.02   0.05    0.2
     *Lab for Science 260          TM S  0.02   0.02   0.1     0.4
     *Laseangle RB-1 (3)           DM S  0.01   0.1            
      Lasertex Frequency Standard  I2 S   ±0.001 (10 s)                    ±0.025
      Lasertex HPI-3D              SM M   ±1 (short term)                    ±1
      Lasertex LL 10               DM M   ±2 (short term)                    ±30
      Lasertex LS 10               DM M   ±2 (short term)                    ±30
      Lasertex LSP 30              AZ M   ±2 (short term)                    ±20
      Laser Metric Systems SFL-1   DM M         2 (unspecified time)
      Limtek LS 10.3               GP M        20 (unspecified time)
      LINOS FS Series              DM S          ±2     ±10    ±20
      LINOS FAS Series             DM S          ±1     ±2     ±10
     *Mark-Tech 7900               DM M                        ±2 (const. temp.)
      Mark-Tech 7910 (6)           DM M                        ±2 (const. temp.)
     *Melles Griot STP-901 (4)     DM S          ±1     ±4     ±6 (8 hrs)
     *Melles Griot STP-909/911 (2) SM S          ±1     ±2     ±3 (8 hrs)
     *Melles Griot STP-910/912 (2) SM S          ±1     ±2     ±3 (8 hrs)
     *Micro-g LaCoste ML-1         DM S  0.2 (10 ms)          1.6 (const. temp.)
     *Motion X (8)                 DM M   ±2     ±6            ±20 (24 hrs)
      NEOARK NEO-262               TZ M          1 (unspecified time)
      NEOARK 430 (11)              DM S         30 (unspecified time)
      NEOARK 430-R4 IR (11)        LD S         50 (unspecified time)
      NEOARK NEO-9111   (11)       AZ S                 1     10 (1 wk)
      NEOARK NEO-92SI-NF (11)      I2 S                                     0.025
      NEOARK NEO-OL101K (11)       OL S  0.0001 (10 seconds)
      NEOARK NEO-2MSS (11)         PS S
      NEOARK NEO-5MSS (11)         PS S
     *Newport NL-1 (3)             DM S  0.01   0.1            
     *Nikon NKL-85 (14)            LD S
      Nikon NN-1                   I2 S
     *NIST ISHL                    I2 S
      NPL Hexagon (13)             ?? S               0.01                   ± 2
      NPL I2 543 nm                I2 S                                      ± 0.25
      NPL I2 633 nm                I2 S                                      ± 0.2
      Optodyne L-103               ?? M
      Optodyne L-104               ?? M
     *Optodyne L-109               DM M
     *Optra Optralite              AZ M
     *Perkin Elmer 5800            LD S
     +PLASMA LGN-212               AZ M         10 (unspecified time)
     +PLASMA LGN-302               DM S         10 (unspecified time)
     +PLASMA LGN-303               DM S         10 (unspecified time)
     +PLASMA LGN-304               DM S         10 (unspecified time)
     +Renishaw HS10                DM M       ±100 (unspecified time)
     +Renishaw ML10                DM M                 ±20          ±50        
     +Renishaw RLU10 (24)          DM M          ±10    ±50    ±50   ±100
     +Renishaw RLU20 (24)          DM M          ±1     ±2     ±10   ±100
     +Renishaw XL80                DM M       ±50 (unspecified time)
     *REO 32734                    DM S          ±2     ±2     ±4 (8 hrs)
     *REO 33099                    DM S          ±2     ±2     ±4 (8 hrs)
     *REO 39727                    DM S          ±2     ±2     ±4 (8 hrs)
     *REO R14286                   DM S          ±2     ±2     ±4 (8 hrs)
     *SIOS SL 02                   DM S          ±2     ±10    ±20
      SIOS SL 03                   DM S          ±1     ±2     ±10
      SIOS SL 04                   DM S          ±2     ±10    ±20
     *Spectra-Physics 117 (5)      DM S
     *Spectra-Physics 117A (4)     DM S          ±1     ±4     ±6 (8 hrs)
     *Spectra-Physics 117B (5)     DM S          ±1     ±4     ±6 (8 hrs)
     *Spectra-Physics 117C (5)     DM S          ±1     ±4     ±6 (8 hrs)
     *Spectra-Physics 119 (7)      LD S                        ±2
     *Spindler and Hoyer ZL-150    AZ M
      Spindler and Hoyer ZL-1150   AZ M
     *Teletrac 150-IV (8)          DM M
     *Teletrac 150 (Long) (8)      DM M   ±2     ±6            ±20 (24 hrs)
     *Teletrac 150 (Short) (8)     DM M   ±2     ±6            ±20 (24 hrs)
     *Teletrac 150 (Long) (8)      AZ M   ±2     ±6            ±20 (24 hrs)
     *Thorlabs HRS015 (19)         DM S          ±2     ±2     ±4 (8 hours)
     *Thorlabs HRS015B (19)        DM S          ±2     ±2     ±4 (8 hours)
     *Tropel 100 (15)              DM S
     *Tropel 200 (15)              DM S          ±2 (5 min)           ±10 (long)
      Uniphase 1220 (16)           DM ?
      VM-TIM LHN-212-1 (18)        AZ M                        ±10 (8 hrs)
      VM-TIM LHN-303               DM S                        ±10 (8 hrs)
      VM-TIM LHN-220SF (18)        ?? S                        ±10 (8 hrs)
     *Wavetronics WT307 (all) (17) AZ M                        ±2            ±20
      Winters 100                  I2 S                                     0.025
      Winters 200 (12)             I2 S                                     0.025
     *Zygo Axiom 2/20 (9)          DM M                 ±2     ±10   ±100
     *Zygo 7701/7702 (9)           DM M                 ±2     ±10   ±100
     *Zygo 7705 (9)                AZ M                 ±10    ±20   ±200
     *Zygo 7712/7714 (9)           DM M                 ±0.5   ±1 
     *Zygo 7722/7724 (9)           DM M                 ±0.5                 ±10

    The "*" denotes lasers that are covered in detail elsewhere in this chapter, usually including tests and photos. "+" denotes lasers that are described but not tested - yet.

    Type Legend:

    AP (Application) Legend:

    A metrology laser can generally also be used for scientific/research applications since all have very tightly controlled optical frequency. And while the converse is often (but not always) true in principle, it's not usually practical or worthwhile except for experimental purposes since metrology systems may require laser characteristics (like two-frequency) that aren't present in laboratory stabilized lasers. In addition, the optics and cabling/electronics requirements would likely make their adaptation potentially complex, if possible at all.


    1. Whether "±" is used for the ppb value depends on how they were listed in the spec sheet, research paper, or other source.

    2. The Melles Griot 25-STP-909/10/11/12 and Aerotech S100, on which they are based, are assumed to have similar performance. The frequency stability specifications listed are from the Melles Griot Web site.

    3. Specifications for the Laseangle RB-1 were obtained from the paper: "Frequency stability measurements on polarization-stabilized He-Ne lasers", T. M. Niebauer, James E. Faller, H. M. Godwin, John L. Hall, and R. L. Barger, Applied Optics, vol. 27, no. 7, 1 April 1988, pp. 1285-1289. The Newport NL-1, which evolved from the Laseangle RB-1 is assumed to have similar specifications.

    4. The Melles Griot 05-STP-901 and Spectra-Physics 117A, which is nearly identical, are assumed to have the same specifications.

    5. While the original Spectra-Physics 117 and later 117A laser use a similar laser (and plug-compatible) head, the laser tube and electronics differ so the specifications may not be the same even for frequency stabilized mode. However, the OEM Spectra-Physics 117C probably has specifications similar to the 117A if installed in a suitable enclosure. (But does not have the output polarizer.) The SP-117B is similar to the SP-117C installed in such an enclosure with polarizer and shutter.

    6. The Mark-Tech 7910 seems to be the same laser as the 7900 but with the addition of an internal optical receiver. So, the laser probably has the same specifications.

    7. The SP-119 with the Servo Option had a stability specification of ±10 ppb per day in a 1964 brochure (which probably predated the actual laser!). But this was improved to ±2 ppb in the 1966 Operator Manual. The SP-119 without the Servo Option has a stability specification of ±75 ppb per day using a constant temperature controller without optical feedback.

    8. Most of the Teletrac 150 lasers output either a single circularly polarized mode or a single linearly polarized mode oriented at 45 degrees depending on model/version. However, at least one model is an axial Zeeman HeNe laser with performance similar to that of an HP/Agilent 5517A.

      I know of only one version of the Axsys/General Dynamics 150 laser. It outputs a circularly polarized beam and uses a PIC-based controller. But other variations may be available.

      For these lasers, Teletrac went to Axsys which went to General Dynamics. This technology is now provided by Motion X (2015).

      Due to the similarity to the Teletrac lasers, Axsys info is lumped in with Teletrac.

    9. The Zygo Axiom 2/20 and 7701/02/12/14/22/24 use dual mode stabilization with an AOM to split the output into two frequency components 20 MHz apart. The Zygo 7705 is an axial Zeeman two-frequency HeNe laser similar in performance to an HP/Agilent 5517D.

    10. The JENAer ZL 600 is a two-frequency HeNe laser which uses the normal longitudinal modes of a ~240 mm long random polarized HeNe laser tube resulting in a difference frequency of about 640 MHz. The laser itself is therefore similar to one like the Spectra-Physics 117A. The ZL 700 and ZL 800 may be the same laser as the ZL 600, but as part of the ZLM 700 and ZLM 800 interferometer systems, respectively.

    11. The NEOARK MODEL-430 uses laser current for stabilization and a heater for overall cavity length control. The NEOARK NEO-430-R4 is an IR (1.523 um) stabilized HeNe laser using Lamb dip stabilization. The NEOARK NEO-OL101K requires another reference laser to which it may be locked with an offset of up to +200/-500 MHz. The NEO-2MSS and NEO-5MSS are intensity-stabilized HeNe lasers.

    12. The Winters model 200 is an offset-locked iodine stabilized HeNe laser all in one unit with a fiber-coupled output.

    13. The locking scheme for the NPL Hexagon laser is not known but may be dual mode with enhancements to improve stability. Versions operating at 632.8, 611.9, 594.1, and 543.5 nm are available. Supposedly, these lasers are now manufactured by a separate company, Hexagon Metrology, but there are no references to stabilized lasers on their Web site.

    14. The Nikon NKL-85 uses Lamb dip stabilization. Although not stated anywhere, this conclusion is based on the characteristics of the laser tube, construction of the laser head, longitudinal mode behavior, and operation of the controller. It is thus the only known commercial stabilized HeNe laser to have used this technique besides the NEOARK 430-R4 and SP-119.

    15. The Tropel 200 and Coherent 200 are similar, if not identical lasers. specifications for the Coherent 200 are shown. The Tropel 100 is the predecessor to the model 200. It uses a similar laser head but the controller us simpler.

    16. The actual model number of the Uniphase stabilized HeNe laser system probably starts with "13". The controller is labeled 1220.

    17. The Wavetronics WT307B/C/D appear to be functional clones of the HP/Agilent 5517B/C/D lasers.

    18. The VM-TIM LHN-212-1 has functional specifications similar to that of an HP/Agilent 5517A but its physical/electrical configuration is unknown. The VM-TIM LHN-220SF has an output power of 40 mW (!!) so it probably has a temperature-controlled etalon inside the laser cavity.

    19. The (HRS015 original, now obsolete) and HRS-015B are a Thorlabs design using a commercial HeNe laser tube, and not someone else's rebranded laser.

    20. The HeNe lasers in the Aerotech LZR2000 and LZR3000 are identical but the LZR2000 has a built-in optical receiver.

    21. The Frazier 100 iodine stabilized HeNe laser appears to use a laser head that is similar or identical to the one desribed briefly in the section: Iodine Stabilized HeNe Lasers and in more detail later in this chapter.

    22. The "Z4214A" is the same as the standard 5517B except for the label! :) It has identical specifications, tube assembly part number, and control PCB. It is probably for an OEM. A Web search for the text string: "Z4214A" returns mostly eBay or other sales listings, but also hits for the 5517B - including Keysight's laser head Web page - even though "Z4214A" is not there, including in any meta fields in the page source (but I didn't follow all the branches).

    23. The actual Z4203 laser tube in the N1211A is physically similar to those in the other HP/Agilent lasers, but it is optimized for high power (at the expense of the REF frequency) because there is a pair of AOMs to generate a much higher (and almost arbitrary) REF frequency.

    24. For the Renishaw RLU10 and RLU20, the stability values are for 1 minute, 1 hour, 8 hours, and 3 years. These support dual-axis with each axis fed from one of the two longitudinal modes used for stabilization. Technically, this means they differ by 0.001449 nm or around 2.3 ppm, which presumably is taken care of in the processing electronics.

    25. The existance of the JDS Uniphase 14xx lasers is only known from scans of a brochure labeled "Preliminary". Based on the absence of these ever showing up on eBay, it has been conclusively determined that they never went into wide production. :)

    Testing of Stabilized HeNe Lasers

    Testing of output power, confirmation of single mode operation, measurement of optical frequency, and possibly measurement of the split frequency (two-frequency lasers only) may be required.

    When making measurements on the output of most lasers, but particularly stabilized HeNe lasers, whether using a power meter, Scanning Fabry Perot Interferometer (SFPI), or another test instrument, care must be taken to avoid back-reflections into the laser that may, well, destabilize it. With some, even slight contamination on the surface of the output mirror, or a piece of clear tape over the output aperture will cause lock to be lost resulting in random fluctuations in output power and/or optical frequency. With most of these, no harm is generally done, but they would then more appropriately need to be called destabilized lasers. :)

    The gas fill ratio for neon of 20Ne to 22Ne affects both the absolute stability of optical frequency as well as immunity to back-reflections. Gas fills that are substantially single-isotope (including natural Ne which is approximately 9:1 for 20Ne and 22Ne, respectively) have a narrower gain bandwidth so that stabilization can be more effective, resulting in less susceptibility to drift in optical frequency. However, using an approximately equal ratio of the two isotopes may result in an increase in immunity to destabilization from back-reflections of 10 fold or more, which can be beneficial in interferometers where all back-reflections cannot be easily (or at least inexpensivly) suppressed. See U.S. Patent #6,865,211: Gas Laser and Optical System. Of course, users of these lasers have little control over the gas-fill. ;-)

    Laser Power Meter for Stabilized HeNe Lasers

    Rather than dedicate a commercial laser power meter for this rather specialized application, one can be constructed for under $20 in parts that will be every bit as accurate and more convenient to use for quick measurements. Since nearly all stabilized HeNe lasers have a power output of less than 2 mW, a fixed range of 0.000 to 1.999 mW using an inexpensive 3-1/2 digit LCD Digital Panel Meter is perfect. A 1 cm square silicon solar cell behind an ~ND2 filter in conjunction with a CMOS op-amp like an LMC6482 configured as a transimpedance amplifier provides a very linear response. (Other op-amps like the LM358 or TL082 also work but the half dozen or so I tried resulted in an offset of a few µW.) As an option, a calibrated output could be added to connect to a PC Data Acquisition System. Power is provided by a 9 V battery with a 78L05 regulator. Mine is installed in a nice little plastic box. :) See Compact Laser Power Meter for HeNe Stabilized Lasers. The swinging boom mount for the sensor was made from the antenna from a defunct radio attached to a base from who-knows-what. This is much more convenient than a Newport post for these lasers where the beam is typically only 2 or 3 inches above the mounting surface. The sensor was installed in the cap from a shampoo bottle with the ND2 filter glued in front of it. Of course, the "instrument" can be used for low power non-stabilized HeNe lasers as well! ;-)

    If the above is too complex to even contemplate :), a basic laser power meter can be constructed from a photodiode, resistor, battery, and digital voltmeter (DMM or DPM). If available, the photodiode should have a diameter at larger thans the beams from the lasers to be measured, though smaller PDs could be used with appropriate fudge factors. :). A 9 V battery is most convenient and even a nearly dead one (5 or 6 V) will be fine as the exact voltage doesn't matter. The resistor provides the calibration such that the photo-current is converted to a voltage corresponding to the output power. Wire the components in series with the photodiode back-biased by the battery and measure across the resistor. For a typical silicon photodiode, the resistor will need to be between 2.5K and 4K ohms to result in a sensitivity of 1 V/mW. Replace the resistor with a trim-pot for precise calibration if desired. The back bias assures linearity up to a few mW, sufficient for any of these stabilized lasers.

    Simple Gauss Meter for Measuring Zeeman Magnet Strength

    After repeated frustration attempting to infer the field strength of HP/Agilent magnets from questionable theory or indirect measurements, I finally bit the bullet and built a super simple but accurate gauss meter. I had obtained the sensor awhile ago but only now got up the enthusiasm to hook it up. It is a Honeywell SS49E analog Hall-effect device with a guaranteed range of at least ±650 G and good linearity, perfect for this application. All it requires is a regulated power supply of 3 to 6 VDC and a DMM. The sensitivity is between 1 and 1.7 mV/g, typical being 1.4 mV/g. The output with no field is approximately one half of the power supply voltage. To eliminate this offset and provide calibration so it reads in gauss, the extremely complex circuit consists of a voltage divider with a trim-pot for Zero match it to the output with no field, and a trim-pot for Gain so that the sensitivity becomes precisely 1 mV/g.

         |         4.7K
         |                    |
         |         + DMM -    |
     +---+---+     |     |    /
     |       |     V     |    \ Zero
     | SS49E +---/\/\/\--+--->/ 1K
     |       |    Gain        \
     +---+---+    20K         |
         |                    |
        _|_       4.7K

    The sensor was installed in a pill bottle using non-ferrous "hardware" as seen in Photo of Sam's Super Simple Gauss Meter. Power is provided to the white 2 pin header from a regulated linear 5 VDC power supply; the output to a DMM is taken via the red and black wires off to the right. The "probe" can be easily inserted into HP/Agilent or other cylindrical magnets, or used outside of them for relative measurements. This is what's called an "axial center probe" since it is forced (by the pill bottle!) to be oriented such that it reads the axial field in the center of the magnet. Any transverse component of the field - which would be due to non-uniformity of the magnetization of the Alnico material is largely ignored since the active region in the laser tube's bore is very narrow. A sample of the sensor can be seen between the circuit and pill bottle.

    Calibration was done using a carefully wound single layer 161 turn, 94 mm long electromagnet solenoid at 1 A, which works out to 21.5 g.

    The gauss meter is perfect. :) But real World magnets are not. They are actually quite terrible in terms of field uniformity. For the typical HP/Agilent magnet, only the central 1/3rd or so is even reasonably constant with the field strength tapering off toward the ends. (Some actually increase slightly near the ends.) Yet, the active discharge in these lasers is often exactly the length of the magnet! And, the field strength may not be symmetric even well inside as these Alnico magnets can have large local variations either from the way they are manufactured, or from deliberate or accidental demagnetization. For example, simply rolling a 20 penny (~3 inch) steel nail around the outside will reduce the field inside by a few percent permanently. And much more extreme effects are possible by applying and removing Alnico or rare earth magnets. More on these effects in conjunction with HP/Agilent lasers, below.

    Sam's Alnico Zeeman Magnet Chargers

    The primary means of controlling the split or REF frequency of an intact Zeeman laser like the HP/Agilent/Keysight 5517 is the strength of the magnetic field of the large cylindrical Alnico magnet surrounding the glass laser tube. (The design of the tube itself determine the possible range.) Typically, the range can be greater than 2:1 with restrictions on both the high and low ends. At the high end, "rogue" modes may be created that interfere with the locking of the laser and operation of the interferometer; at the low end, stability may suffer. Most lasers operate closer to the high limit than the low limit, so being able to boost, say, a 5517B to a 5517D by increasing the field strength is unlikely. As described elsewhere in this chapter, reducing the field is quite trivial without any fancy equipment - just some pieces or sheets of of iron or steel or even other magnets stuck to the Zeeman magnet will reduce the field with effects that may remain permanently after removal. However, to boost the field in a useful manner is almost impossible without a "device". ;-)

    Searching on eBay may turn up high price gargantuan machines for creating (or "charging") the magnetic field of Alnico, ceramic, and rare earth magnets. though most such searches return totally irrelevant results. And any that do show up are often untested (a synonym for "broken"), incomplete, or not suitable without additional effort or expense - or not at all - for modifying the field strength of the types of the magnets used in Zeeman lasers. A search on Google for "Alnico Magnet Charger" is a bit more successful with the first hit being for one of those gargantuan devices, the All Magnetics MC-8000 Series Magnetizer. It has the appearance of a dishwasher on wheels - which says something about its weight. The basic version is only $9,400 for the standard version, and there's a double strength option available. $9,400 just happens to be about the cost of a 5517 laser! So if it's a one shot (no pun) deal, just buy a new laser. ;-) But if you do by chance have access to one of those beasts, it may be suitable - at somewhere below its minimum setting. ;-)

    The magnets in commercial Zeeman lasers are almost always made of Alnico (an alloy of iron which includes aluminum, nickel, cobalt, and possibly some other metals). It turns out that Alnico is among the easiest common magnetic material to deal with in terms of changing the magnetic field. Only a very few models of Zeeman lasers use ceramic or rare earth magnets. And none of these are likely to be of more than academic interest.

    At first it never occurred to me that modifying the field of such a large magnet could be accomplished with a very simple device. Permanent magnets are, well, permanent, right? :) Based on experience with various soft magnetic materials or other magnets in contact with the Zeeman magnets, it's clear that reducing the field permanently is all too easy and often occurs accidentally. But increasing it? Digging into the actual requirements in terms of the necessary charging field magnitude for an Alnico magnet and electromagnetic solenoids, showed that it really wasn't that difficult. The only unknown was then how much energy would be required. It turns out not that much and is mostly just a matter of providing enough ampere-turns. So constructing a MacGyver-style magnet charger suitable for HP/Agilent (or other Alnico magnets of similar or smaller size) requires just a few commonly available electronic components and a spool of thick wire. The most time consuming part was winding the solenoid. It was literally only around 15 minutes to wire up a bridge rectifier and 2,000 µF capacitor bank and be able to change magnet field strength. For my initial tests, a Variac was used to adjust the voltage (and thus energy) and touching a pair of wires together discharged the capacitors into the coil with only minimal sparking, though doing this with even a relatively high current push button switch eventually resulted in the contacts tending to stick together. :( The voltage on the capacitors was increased in small steps and after each discharge the field was measured. Even at 1/4 voltage, a single pulse increased the field of a 220 G magnet by 10 G. At 1/2 voltage, the field increased to 350 G. So, it was clear that the range would be quite adequate. If boosting the magnetic field only needs to be done for one laser, nothing fancier is really required.

    My slightly more refined version will charge most bare HP/Agilent magnets to 500 G or slightly higher, which is greater than the field used in any known 5517, 5501, or N1211A laser, though I've come across a few samples of HP magnets that won't go above about 325 G. This is probably due to a different formulation - e.g., Alnico 3 instead of Alnico 5. See:

    It can, of course, also be used to reduce the field to any desired value down to 0 G or fully reverse the polarity. And as a bonus, it can magnetize screwdrivers! The total cost for this gizmo was $0.00 using all junk parts. Sorry about the ugly recycled HeNe laser power supply box. It was even uglier originally. :) The control to adjust pulse energy is a common light dimmer. This isn't ideal as there is hysteresis - a Variac would be better but more expensive (and heavier) if built-in, or a bench Variac could be used. A neon indicator shows that the unit is plugged into a live AC outlet. It is line-connected with no isolation, so this is a useful safety feature. The two yellow LEDs show when the dimmer is on and a rough indication of drive voltage. The red LED and digital panel meter display the capacitor voltage. And then there's the little silver FIRE button beneath it. ;-) The maximum energy is about 25 J at a voltage of 160 V from a pair of 1,000 µF electrolytic capacitors in parallel. An SCR discharges the capacitors into a 290 turn coil 8 inches in length wound with #14 stranded THHN building wire in 4 layers on a 2-1/4 inch I.D. cardboard form. A coil using magnet wire would be more compact but this wire was available and stranded wire is easier to wind by hand. ;-) To increase the field of HP magnets inserted with their arrow matching the arrow on the coil, the current should flow CCW when viewed from the front. The 8 inch length is to assure field uniformity in the center, though that is probably not required. At maximum energy, the field at the center on-axis inside most HP/Agilent magnets reaches between 480 and 550 G and some go higher, but there are a few exceptions as noted above. One shot usually brings the magnet to within a few percent of the target field strength.

    Rough circuit diagrams for versions using Variacs are shown in Schematics for Sam's Basic and Dial-A-Field Alnico Magnet Chargers. The schematics are called "rough" because there are other minor differences and some part values may not match. For some reason I have no record of the actual circuit that was built. And I'm not going inside to figure it out. Can you believe that? Refer to the interior photo. ;-)

    When designing these, the discharge should be critically damped or slightly over-damped to avoid a reversal of current which not only could reduce or negate any effect on the magnetic field but also put reverse voltage across the electrolytic capacitors - and that can be explosive. :( Since the damping factor will depend on what's inside the coil, a large reverse polarity ("free-wheeling") diode is present to prevent this. (The massive bridge rectifier visible in the photo was used only because it was available; Any diode rated at least 50 A and 400 V should be acceptable.) But if the discharge is too slow, it may not shut off if the pushbutton is released quickly. This was found to be true when pulsing a 100 W incandescent light bulb instead of the coil, used as a dummy load during testing. :) Its resistance when cold is around 10 ohms but becomes 100 ohms when hot, part way through the discharge. So the effective time constant can be close to seconds (2,000 µF x 100 ohms).

    The driver SCR was included to assure a solid firing pulse for the main SCR. But it is probably unnecessary if the current through the FIRE switch is comfortably above the minimum Ig for the main SCR. Even with contact bounce, the initial pulse should be much longer than needed. The next version (below) does away with the driver SCR without problems. And if recharge time is not a concern, it should be possible to decrease the charge current enough so that interrupting it is not necessary to assure SCR cutoff. Even with only 200 ohms total in series with the bridge (as in the prototype), it was a close run thing with the SCR not shutting off consistently only near maximum input. With 1K ohms as in the schematic, it's probably fine, but will depend on the SCR specifications.

    Originally, it had a small edge-view analog meter for capacitor voltage but (1) this wasn't high tech and (2) for unknown reasons the movement become sticky after awhile. Its replacement, a $1 DPM from eBay only reads to around 33 V max despite what the listing claimed. Can you believe that? :( ;-) (It's not broken, several others behave the same.) So, RCal is set so full scale is 30 which isn't as round a number as 100 (%) would be but it's just an arbitrary value anyhow. The only disadvantage of using a DPM is that it needs a separate power supply so the guts of a 12 VDC wall adapter that had mediocre regulation was stuffed inside the box. For this regulation doesn't matter as the DPM will run on 5-30 VDC. It could be powered from the SCR trigger supply but that would require some changes since the LED display draws up to 10 mA, which would be a problem especially at low output voltage. Alternatively, a 9 V battery could have been used.

    While the photo shows a bare Agilent magnet, it is possible to remove the rear-end bracket ("foot") from the complete tube assembly in any of the lasers in the small cases and then slip the magnet butt-end first into the charging coil. For "Long" tube tube assemblies, a small amount of potting material would need to be cut away for the bracket to clear, but this can be done without affecting anything else and doesn't need to be replaced. But it's very easy to accidentally slice into the high voltage cable. Then it may arc when starting and need to be re-insulated. The fatter tube assemblies in the 5517A, 5518A, and 5519A/B are out of luck without significant and messy disassembly, but the likelihood of ever needing to boost their field is rather small.

    The field uniformity at the end of the solenoid isn't as good as in the center, but is acceptable. This has been confirmed with an otherwise useless Agilent 5517C. Since only external measurements are possible with the glass tube inside the magnet, the exact field strength along the axis is not known. However, based on measurements with the sensor centered front-to-rear and against the magnet show that a similar range of adjustment is possible. Initially, the field measured 365 G. After partial demagnetization, it dropped to 135 G. And at full magnetization, was over 520 G. With the laser tube powered, the change in the split frequency was immediately apparent and there was no evidence of any damage to the tube. This is now done routinely to adjust split frequency when an increase is required. When reducing the field, only a tickle is needed - a few percent of maximum voltage on the capacitors. Multiple pulses can be used to incrementally reduce the field strength. Too much energy and the field will reverse polarity. Or if the magnet is inserted backwards during charging. After zapping a magnet, it is worth confirming that the field is still in the proper direction by bringing it near a known original magnet - they should repel when oriented the same way. The laser will not work properly if its field is the wrong polarity.

    The only annoyance of the original implementation was that the magnet had to be removed from the coil and inserted in the opposite direction to change from magnetize to demagnetize, or vice-versa. So a BIG DPDT switch was added to reverse polarity and select charge or discharge (lower right of case in the photo). ;) It's only rated 20 A but that's when switching. The contacts have very low resistance when stationary. Less hassle and cost than an H-Bridge of SCRs. It works well enough.

    Unfortunately, the magnet can't be fine tuned using the Dial-A-Field™ with the tube assembly installed in the laser. Perhaps a future coil design would solve this. :-) Or perhaps the electromagnet coil described in the next section could be used. However, for setting the field close to what is required, the laser tube can be powered while inside the charging coil. The heater voltage should be set to 5 or 6 V which is where it would be when locked. After thermal equilibrium is reached, the field can be adjusted while monitoring the Zeeman beat frequency with a scope and/or frequency counter using an HP 10780 optical receiver or biased photodiode like a Thorlabs DET110 behind a polarizer. Adjust the heater voltage slightly if mode sweep is stalled or takes too long. The effects of any change in magnetic field will be instantly obvious. The lock point would be in the center of the region of mode sweep where there is a beat. This is usually at the maximum frequency for higher-REF lasers like the 5517D, but may not be for lower ones like the 5517A or 5501B. If boosting the field, the output should also be monitored on a Scanning Fabry Perot Interferometer (SFPI) for rogue modes, which will always be present on either side of the strongest mode if the field goes too high. Below the rogue mode limit, there will be at least a short time when only the single Zeeman split mode is present. And if the field is boosted really high, the split gain curves may be pulled so far apart that there will be no beat at all. Set the field so the split frequency is slightly higher than required for the specific model laser - it can always be reduced.

    But there doesn't appear to be any reason the pill bottle Gauss meter probe described in the previous section cannot remain in place inside the magnet inside the coil when it is zapped to monitor the progress of the field adjustment. :)

    Finally (for now), a version was constructed upon the request of a major laser company for installation in a custom case with external Variac. So it's somewhat simpler than the one in the ugly box. ;-)

    The schematic for this matches the actual device fairly closely except that the polarity reversing switch will be added later if needed. And as with the boxed version, the massive bridge rectifier was used for the free-wheeling diode only because it was available; Anything rated at least 50 A and 400 V should be acceptable.

    As a side note in the trivial triviality department, it is theoretically possible to accidentally change the strength of the magnets in your expensive HiFi system loudspeakers by listening at really high volume. A rough calculation shows that for a small speaker driver using an Alnico magnet, a peak electrical power of around 1,000 watts may result in a permanent field change (about 1 millisecond before the voice coil vaporizes). :( :) Since modern speakers use more exotic ceramic or rare earth magnets with higher coercivity and a larger ratio of magnet to core area, somewhat more power will be required (like 10,000 or 100,000 or perhaps 1,000,000 W). But why take chances? Beyond preserving your hearing and relations with the neighbors, this is yet another reason to keep the decibels down. :-)

    Sam's Variable Electromagnet for Intact Zeeman Tube Assemblies

    This consists of a custom-wound coil and Arduino-based driver that provides for a range of from 0 to >500 Gauss. It was originally developed for Zeeman testing of the glass laser tubes (either removed from the magnet or new ones). However, with very minor modifications, it can also be used to test nearly intact HP/Agilent/Keysight laser tube assemblies to assess the useful range of split frequencies if the magnetic field were modified appropriately. In its current form, 3-1/2 pounds of #18 AWG magnet wire is wound on an HP magnet that serves to bias the field so that it only needs to be changed by around ±250 G to achieve the full range, This requires around 1/4 of the power to produce 500 G compared to doing it without the bias field. However, by making the magnet removable, any HP/Agilent/Keysight tube assembly could be installed by simply taking off the rear foot bracket, easily done on both Long and Short tube assemblies.

    A summary can be found in the section: Sam's Full Range Variable Zeeman Electromagnet. And complete details should you wish to replicate this can be found in Universal Bipolar PWM Driver 1 Assembly Manual.

    Sam's Optical Dropout Detector 1

    It's sometimes useful to know if the output of a laser suddenly dropped in power or went off entirely, even for an instant. This could be to monitor a Zygo 7701 or 7702 for actual dropouts due to either high mileage or poor manufacturing or to detect that a stabilized laser has lost lock. Sam's Dropout Detector 1 uses a photodiode to monitor a laser beam or portion thereof and switches from red to green if even a microsecond loss of power is detected. It's a discrete S-R flip-flop that's triggered off by the Reset button and on by detected optical power above a threshold, set via a trim-pot.

    Frequency Counter Testers for HeNe Lasers

    For testing and adjustment of Zeeman HeNe tubes and lasers, as well as characterizing subtle effects like the second order beat from longer HeNe lasers, it is desirable to be able to monitor the beam in real time. While it's certainly possible to drag out a biased photodiode detector and actual electronic frequency counter or oscilloscope, that can be annoying if it needs to be done frequently. For many years, I was setting up a Thorlabs DET110 detector wit Tektronix 465B or 2467A analog oscilloscope. Even with the instrumentation readly accessible, it became annoying to set up each time. This scheme implements a stand-alone system that is taylored to two-frequency HeNe lasers as well as lasers that generate beat frequencies in the 100 kHz to several MHz range through other mechanisms, is simpler to use, more compact and portable, and can remain assembled permanently. It is based on µMD2, which is a readout for use with all interferometer-based displacement measuring systems. See Micro Measurement Display 2 (µMD2). While µMD2 interfaces to a PC for actual measurements, there is a small OLED display on the µMD2 PCB itself that is normally only used for diagnostics or not at all. With a small enhancement to the firmware, it provides an accurate frequency counter which can either be used independently or along with actual measurements. µMD2 firmware V2.47r and above supports this option with an update rate of 10 Hz and a resolution of 10 Hz. Its accuracy depends on the Teensy 4.0 CPU clock but a correction factor (single floating point multiply) could easily be added in the firmware if needed. For two-frequency laser testing the small error is not likely to be of any consequence. See Zeeman Frequency Counter using µMD2. The entire thing is powered by a 15 VDC wall adapter, with a +5 VDC regulator added for the µMD2 PCB via a hacked Micro USB cable. The HP/Agilent 10780C does the optical detection much more conveniently than a biased photodiode since it generally doesn't need any adjustments. It automagically works with the available power from almost any Zeeman laser and provides the differential RS422 signals to µMD2. Furthermore, the sensitivity of the 10780s is so high that the beam can be sampled by something like a microscope slide. Its bandwidth is greater than 7.5 MHz due the requirement to function with a 5517D laser (split frequency of 3.4 to 4.0 MHz) at the maximum stage velocity which increases the return frequency. The HP/Agilent 10780C is spec'd very conservatively. For the most common HP/Agilent/Keysight Zeeman lasers (5517A/B/C/D, 5501A/B, N1211A), and virtually all home-built HeNe Zeeman lasers, the 10780A/B (maximum frequency around 5 MHz) would be satisfactory. And the usable range of the µMD2 input goes way beyond the ~8 MHz upper limit of any Zeeman HeNe laser so it can also be used to confirm two frequency output of Zygo 7701/7702 lasers at 20 MHz (with a Zygo 7080 optical receiver substituted for the HP/Agilent 10780C).

    The Mini Laser Mode Analyzer (mLMA1) Version 2 with its LCD display can also be used as a frequency counter. But (1) its input is single-ended so the interface circuitry would need to be matched to the specific levels of one of the outputs of the 10780 or use a different interface and the usable upper limit is around 4.0 MHz due to its much slower microcomputer. And its frequency counter display is also very small and I have no plans to change that. ;-)

    An inexpensive frequency counter or PCB could also be used. But it should be the type with an actual gated counter, not one based on down-conversion, which are sensitive to waveform shape. And the input may need to function with lower signal amplitudes than TTL provides. The one I tested is typically listed as "1Hz-50MHz Crystal Oscillator Frequency Counter Meter 5-Digital LED Display Kit", under $10 on eBay from USA sellers; under $5 from the Far East. For this application they tend to be a bit too smart for their (our) own good with auto-ranging that apparently cannot be disabled. So, as the beat frequency comes and goes, there will be instants where the gate will sample something in the kHz range and a totally bogus value is displayed in the wrong digits. If anyone can locate the firmware source code in C for these, I Could probably remedy that. The large red 7-segment LEDs also hark back to 1970s equipment. ;-) See Zeeman Frequency Counter using Low Cost Kit. A circular polarizer sheet over the display improves the contrast, especially for photos.

    The benefit provided by the 10780 (as opposed to something like a biased or even conventional amplified photodiode) with either µMD2 or a low cost frequency counter is seamless operation with almost any two-frequency Zeeman HeNe laser without adjustments of any kind.

    This rig may also be used to test for the second order beat in longer laser tubes that oscillate with at least 3 modes present during part of mode sweep. The frequency of the second order beat is typically in the hundreds of kHz range so the 10780 will respond to it. The (first order) beat between lasing modes which is in the hundreds of MHz range is well outside its bandwidth and is ignored. The orientation of random polarized tube should be adjusted so that its principle polarization axes are at 0/90 degrees so that the modes will be combined by the 10780's polarizer at 45 degrees. Linearly polarized can be at any orientation where enough optical power is passed by the 10780's polarizer.

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.

    Aerotech Stabilized HeNe Lasers

    Aerotech Syncrolase 100 and Melles Griot 05-STP-91x

    The original Syncrolase 100 was developed by Aerotech before their HeNe division was acquired by Melles Griot. The Melles Griot 05-STP-909 through 05-STP-914 are successors to the Syncrolase 100 using substantually the same technology. Both are included here.

    Description of the Aerotech Syncrolase 100

    The Syncrolase 100 is an intensity stabilized HeNe laser that combines the laser tube and locking circuitry into a single compact unit with only DC and/or HV power applied externally. Everything related to stabilization is contained in a small module called the "SFA Adapter" that screws onto the end of a substantially standard 1 or 2 mW HeNe laser head. ["SFA" stands for "Single Frequency Adapter" so "SFA Adapter" is just a wee bit redundant, but so be it. ;-)]. Included in the SFA Adapter (called the "Locking Adapter" by Melles Griot) are the polarizing optics and beam sampler, a single photodiode, and all the feedback control circuitry. An S100 laser is shown in Photo of Aerotech Syncrolase 100 With External HeNe Laser Power Supply. Except for the labeling, the Melles Griot versions are functionally identical.

    With a single photodiode sampling the beam, only the amplitude of one of the two polarized modes can be stabilized. There are two beam-splitter cubes inside the adapter. The one closest to the laser tube is a common polarizing beam-splitter cube which diverts thee unwanted polarized mode into oblivion and only passes the desired one. The second diverts about 10 percent of the desired mode to a photodiode on the locking PCB. A second photodiode soldered to the locking PCB monitoring oblivion can be used to implement dual mode stabilization as I have confirmed.

    Despite using only a single mode for feedback, the frequency stability will still be quite good once the laser tube has reached thermal equilibrium and its power has leveled off. As the tube ages and its power declines, the output power from the laser will remain constant until it approaches what's available from the HeNe laser tube. At that time, it will lose lock and may even be damaged, more below. With dual mode frequency stabilization, locking will still be possible even when the power output from the tube is very low because it is the difference of the polarized mode amplitudes that produces the error signal, not a specific value. In addition, with intensity stabilization, the frequency will drift as the tube gets weaker and the lock point moves with respect to the neon gain curve. (However, due to the shape of the gain curve for the tubes generally used with these lasers, the change in frequency will be minimal.) Nonetheles, why frequency stabilization was not implemented instead, or in addition to intensity stabilization as in the 05-STP-901, is a mystery as it would have been a very straightforward enhancement - primarily a second photodiode! Nearly everything else is already there, expecially in the newest Melles Griot version. More on this below.

    Lasers based on the Aerotech technology are now sold as the Melles Griot STP1 with specific model numbers of 25-STP-910 and 25-STP-912. [They may also be found as 05-STP-910 and 05-STP-912. Whether "05" or "25" is used simply depends on if it is considered a component (05) or system (25), and sometimes at random!] There was also an 05-STP-914, now discontinued. It had an output power similar to that of an 05-STP-910, but a larger diameter beam with lower divergence. And as of early 2016, the -910 had also apparently been discontinued, with only the higher power -912 remaining. Then a few months later, it reappeared, possibly as a result of my analysis pointing out that due to the mode profile of the shorter -910 tube, it has better absolute frequency stability with respect to output power adjustment over its life. :) Or more likely due to complaints from unhappy customers requiring them as replacements in instruments like wavemeters.

    The Melles Griot lasers are physically and functionally very similar to the Syncrolase 100 but both of these use a separate HeNe laser power supply instead of having one inside the laser head. (The Melles Griot 25-STP-909 and 25-STP-911 had the built-in power supply but have been discontinued.) While, it is not known how much the electronics differ compared to the S100 version, all indications are that very little has been changed since obtaining the technology as part of the acquisition of the HeNe laser division of Aerotech. Melles Griot calls them "frequency stabilized lasers" though the description indicates that the same amplitude stabilization technique as the Syncrolase is used (and examination of the locking adapters confirms that there is only a single photodiode). Interestingly, the latest Melles Griot locking adapter PCB has hooks for implementing full frequency stabilization, but to my knowledge, it has never been offered as a product and I'm the only one who has ever tested this feature :) More below.

    Searching for "Melles Griot 25-STP-910" or "Melles Griot STP1" should return a description and spec sheet. Here is a summary of the specifications for the Melles Griot versions:

    Note that based on the user manual, the power output range is NOT guaranteed, only the stated value (0.7 or 1.0 mW). It may be possible to order one with a guaranteed higher output but this has not been confirmed.

    And, if you'd like to order a few, the Melles Griot price in 2012 was $4,388 each for the low power version (0.5 to 0.95 mW) and $4,662 each for the high power version (0.6 to 1.4 mW). I wonder how they came up with those especially round numbers for the prices. (As of 2018, Melles Griot is no longer in the HeNe business - period. Go figure.)

    The output power is a user adjustment (a trim-pot) that sets the intensity stabilization point, and indirectly the operating frequency. In addition to versions based on output power, the Syncrolase came in two versions based on whether a pair of DC wall adapters were used to power an internal HeNe laser power supply and the locking controller, or whether the laser head had a standard Alden connector to attach to an external lab-style HeNe laser power supply, which is included in the price, along with the wall adapter for the locking controller! :) Now, why weren't the two combined, given the warning in the operation manual: "Application of power to the SFA adapter (locking controller) in excess of 5 minutes with the head de-energized may damage the SFA adapter". There's more on this below, though the Melles Griot version appears to have fixed this minor deficiency.

    One of the unique features of this system is that rather than using a resistance heater over a substantial part of the HeNe laser tube as is done in most commercial stabilized HeNe lasers, these lasers use a compact "locking adapter" attached to the end of the laser head containing a coil surrounding only the OC mirror mount stem to directly heat the metal mount via RF induction. A very simple MOSFET driver can provide over 10 W directly to the mount resulting in a very rapid response. Based on tests I've done, I estimate that at maximum RF power, it will increase the temperature of the mirror mount stem itself by greater than 1 °C per second. This is more than an order of magnitude faster than traditional resistance heaters surrounding the glass portion of the tube. A temperature sensor in close proximity to the mirror mount stem senses its temperature and is used both to switch the feedback loop on when hot enough, as well as to shut the heater off if the temperature goes too high. Warmup to fully stable operation still takes 20 or 30 minutes because the rest of the laser head has to come into thermal equilibrium as well as the mirror mount stem. But, initial locking is very quick - typically 3 to 5 minutes. And once locked, it should use less power and be more immune to ambient temperature variations, and the faster response also improves frequency stability.

    The same locking adapter may be used with any compatible laser head requiring at most minor electronic adjustments. In addition, the use of this technique allows for the possibility at least in principle of converting almost any HeNe laser tube with a suitable mode structure and cathode-end output into a stabilized laser by simply attaching the locking adapter. However, in practice, minor details like the mirror mount stem dimensions and the length of the exhaust tip-off prevent most common tubes from being used. For tubes with anode-end output, if sufficiently robust insulation could be added between the mirror mount stem (HV anode) and coil, they too would work without worrying about the exhaust tip-off.

    The DC wall adapter (either version, 1 or 2 required depending on the model of the laser) is rated 13 VDC, 1.3 A. Measurements show it to have an open circuit output of 16.5 V. The plug is 5.5 mm/2.5 mm center positive. Since there is a 7812 +12 V regulator in the controller (see the schematic below), the output of that DC adapter must be greater than about 14.5 V to assure proper regulation. So, at least once the feedback loop is closed, the input voltage should never dip below 14.5 V. I do not know the official specifications for the external HeNe laser power supply (where required), but based on the length of the tube and other typical Aerotech tubes (and the supply that comes with an OEM version of the 05-STP-910), it is probably around 1,500 V at 4 mA.

    That Melles Griot OEM version (typically found in some high-end wavelength meters as a reference) is otherwise identical to the normal one except that its output is fiber-coupled. One example is shown in Melles Griot Fiber-Coupled 05-STP-910 Stabilized HeNe Laser. There is an additional assembly that screws onto the output end of the laser (and is then glued) with a 4 position shutter which can be set to block the beam, pass it to the fiber, or divert it at right angles out the side so the laser can be set up independent of the fiber. Some versions have a fully adjustable fiber port enabling a broken or damaged fiber to be replaced and realigned relatively easily, while others are totally glued with rock-hard Epoxy at the factory with no chance of alignment in the field. This also means that even removing and reinstalling the laser head - or replacing it with a new one - is likely to affect alignment in a way that is difficult to remedy. A more recent version uses a modified laser head assembly that bolts on rather than screwing on, which is more precise and is more immune to misalignment. Usually, a replacement laser head can be installed without requiring major fiber alignment. See Melles Griot 05-STP-910-536 Reference Laser for the Agilent 86122B Multi-Wavelength Meter. It's functionally identical to the one shown above. Elongated holes in the laser head flange allow for the polarization orientation to be fine tuned. A fiber coupler attaches to the output-end of this laser. While fully adjustable, doing do can turn out to be a real treat so should be avoilded if at all possible. And as with most HeNe lasers, back-reflections can result in mode flips and loss of stabilization. With optimal alignment there tends to be some light going directly back to the laser even though the fiber ferrule is angle-polished. A "poor mans' isolator" using a Quarter WavePlate (QWP) might be able to minimize this. (A Web search will easily find information on the 86122B including specifications and operation manual.) Some versions also include a beam sampler and photodiode so that the output power can be monitored prior to the fiber. And as can be seen, these OEM systems also have the status LED and power connections brought out as twisted wire pairs:

    The HeNe laser tube in the Syncrolase 100 and 05-STP-9xx lasers is between 6 and 8 inches long, depending on version. The ballast is typically 80K ohms made up of multiple thick film resistors on a ceramic substrate potted in a rubbery ring that slips onto the anode mirror mount stem. The default current is 4 mA for all STP-9xx heads (though it may have been 4.5 mA for at least some Aerotech versions). New STP-910 tubes will remain lit down to 3 mA or less; new STP-912 tubes down to 3.5 mA or less. This dropout current tends to increase as the tube is run. Eventually, it may be necessary to turn up the power supply current to 4.5 or even 5 mA to squeeze out a few months more life from of a high mileage tube.

    A common 6 to 9 inch random polarized tube with cathode-end output (high voltage far away from the electronics!) would probably work except that the mirror mount stem needs to be a about an inch long with the exhaust tip-off cut off close to the end-cap so as not to interfere with the coupling coil assembly. Very few tubes have these characteristics, though some are close enough to be usable in a pinch. And it may be possible to add a metal sleeve over the mirror mount to extend it so the induction coil can couple to it. However, using too long a tube might result in a second longitudinal mode being present if the Output Adjustment is set so the main lasing line is too close to the neon gain center.

    For details on theory and implementation see U.S. Patent #4,819,246: Single Frequency Adapter.

    A schematic diagram of the electronics for the Syncrolase 100 can be found at Schematic of Aerotech Syncrolase 100 Controller. This may not yet be quite complete and numerous errors are possible since the PCB is tight, it is a 4 layer board, and the soldermask is almost totally opaque. It was not much fun to trace the circuit. Part numbers are not available for a half dozen components because (1) they might have been obscured and (2) there were several added parts that appear to be in the "oops" category. :-) (Those added parts relate to the overtemperature protection - more on this below.) But I bet this schematic provides infinitely more information than what's available anywhere else! :) Melles Griot has redesigned the PCB at least twice, the latest version using mostly surface mount parts. (Photo and comments on this and the other controllers below and in the next section.)

    The gate of a power MOSFET is driven by a simple oscillator, running at between 500 kHz and 1 MHz (I measured about 700 kHz on one unit). The feedback signal is summed into the gate junction from the error amp and serves to modulate the output of the induction heater to maintain lock once the operating temperature has been reached. The coil is just short of 9 turns of #24 AWG wire close wound on a 1.35 cm form. Due to the way the leads enter through the back of the form, the final turn is short changed! :) This is probably not terribly critical though.

    (The following applies directly to the original Aerotech design except as noted.)

    The RF driver consists of a Hex Schmitt trigger (MC14584BCP similar to a CMOS 40106) with one section used as the oscillator and the remaining sections paralleled to buffer its output. An RC network converts the squarewave of the oscillator to a roughly triangle waveshape at the MOSFET gate. The output of the Error Integrator feeds into the gate as well with the effect of modifying its DC offset. Since the MOSFET gate threshold is fixed, this produces a modulation effect which is a combination of amplitude and pulse width, with the net result of controlling the amount of RF power transferred to the HeNe laser tube mirror mount stem. A significant part of the capacitance in the waveshaping network is the internal input capacitance of the MOSFET gate itself, and this may exceed 1 nF. Thus, it's possible that if the MOSFET needs to be replaced, the value of the capacitor between the gate and ground (C13) may need to be adjusted as well to maintain approximately the same net capacitance and waveshape. The MOSFET gate capacitance can vary by a factor of over 2:1 between MOSFETs with the same part number, or by even more if a MOSFET with otherwise acceptable specifications is substituted. On the unit I have, it was about 1.3 nF.

    Newer versions include a ULN2003 Darlington array, possibly for driving the MOSFET in place of the HEX Schmitt Trigger. (But that is still present.) They may also use a thermistor sensor in place of the thermocouple - it looks like a 1N4148 with no markings but tests like a 10K ohm resistor at room temperature. That's much cheaper and easier to use!

    The control functions are implemented in the four sections of a TLC27L4CN quad op-amp as follows:

    When powered up, the temperature sensor is initially cool so the RF driver comes on at full power. This results in the tube expanding such that a full mode sweep cycle is order of 1 second. The result is that for the lower power versions with a shorter tube (e.g, Melles Griot 05-STP-910), the beam actually goes on and off at that rate since there are only two longitudinal modes and one polarization is blocked by the beam sampler. There is nothing wrong with your laser! :) For longer tubes there will still be a noticeable variation in beam intensity. The length of the mode sweep cycle gradually increases (and the blink rate decreases) until the mirror mount stem reaches the operating temperature (something like 80 °C) in as short a time as less than 1 minute. The feedback loop then becomes active and the SYNC LED comes on indicating that the feedback loop is closed. Lockl will then be achieved at the current mode position on the next rising slope of the output. However, since the remainder of the laser tube is still increasing in temperature due to the normal heating of the discharge and hasn't reached thermal equilibrium, the RF drive gradually decreases cooling the mirror mount stem so that the total distance between the mirrors remains constant. Eventually, when the mirror mount temperature gets to be too low, the system will switch back to continuous heating for a time based on the hysteresis of the Sync Enable Comparator. After 20 to 30 minutes, the laser tube will reach thermal equilibrium and the system will then remain locked forever. (Unfortunately, many people take this literally and leave the laser on until it dies, which is considerably sooner than forever!)

    The heating rate in most of the Melles Griot 05-STP-910s I've seen is set to be much slower so loss of lock is less likely. Usually, the laser will lock initially in 3 or 4 minutes and never lose lock after that. Thus, if the temperature set-point and heating rate is adjusted optimially, the repeated unlock-lock cycles may be avoided. For use with longer tubes, the heating rate should be set to be slower to give the rest of the tube time to partially catch up before locking.

    On the Melles Griot locking adapter, a bi-color LED is used that serves a dual function and is labeled "Stable/Overtemp". It can be off, green, or red.

    Note that the LED being green doesn't mean the output is stable, only that the feedback loop is active. For example, the laser could still lose lock due to back-reflections or be unable to lock as a result of the output level being set too high. (The latter should not be possible on a new laser if correctly set up at the factory, though it could occur with a high mileage tube that has lower output power.)

    Here are some photos:

    There are at least three major variations on the design and PCB layout of the locking adapter controller:

    There is more on the locking adapter PCBs in the next section.

    The coupling coil assembly on the first Syncrolase 100 of mine had disintegrated due to excessive temperature. (Actually the magnet wire and its insulation is in fine shape but the plastic form on which the coil was wound is no longer intact and it's not even possible to determine much about it.) I've tested the induction heating winding a test coil on a tube made from insulating plastic sheet. The effect is impressive considering the simplicity of the circuitry (see the schematic below) raising the temperature of a dummy mirror mount stem by more than 1 °C per second even with a coil that is probably far from optimal.

    I do not know for sure if the cause of the destroyed coil form was due to a part failure rather than simply a result of the laser being been left on for 7 years continuously! :) The HeNe laser power supply was indeed dead, probably due to the tube being very hard to start and impossible to run for more than a few seconds regardless of power supply or ballast resistance. So it's possible that when the tube decided it was tired of doing its thing and the power supply shorted out, the controller ended up cycling on the over-temperature condition. The Melles Griot manual does warn against running without the laser on. And, electrical tests seem to indicate that the controller is working properly.

    It's likely that the Over-Temperature (OT) adjustment was incorrect and too high all along. Since it's not something that affects normal behavior, it would be all too easy to neglect setting it properly! I've also been told by the former owner that this laser always ran very hot. If the tube fails - even if someone forgets to plug in its wall adapter! - the heater tends to be on and bad things can then happen if the OT setting is too high. Ask me how I found out. :( :) OK, I'll tell you. I acquired another Syncrolase with a good tube but that would not stabilize. I traced the problem to what I believe may have been a short in the temperature sensor and then adjusted it to operate at a reasonable temperature set-point. But I accidentally left the controller powered after turning off the laser and went away. When I returned (after lunch!), the entire assembly was too hot to touch and the platic coil form and it's cover had melted!!! Apparently, either the OT setting was way too high (it's possible someone before me messed with it) or it isn't effective or was broken.

    Interestingly, on one of those rare occasions where I was able to get the tube to remain on long enough with a lab power supply to watch a few mode sweep cycles, it is a classic FLIPPER! I suppose that the flipperitis could have happened in its old age (it is also weak - about 0.7 mW - and with brown crud in the bore), but normally the flipper or non-flipper status of a tube doesn't change over the course of its life. I do have another Aerotech laser head that would screw right on to the controller but it too is a flipper! :( :) In fact, its behavior shown in Plot of "Flipper" Aerotech OEM1R HeNe Laser Head During First Part of Warmup and the merged version in Plot of "Flipper" Aerotech OEM1R HeNe Laser Head During First Part of Warmup (Combined) looks virtually identical to that of the Syncrolase tube (over the few mode sweep cycles I could see before the tube went out). But, even more interstingly, the flipping of the tube in the plots ceases entirely and it becomes perfectly well behaved once nearly warmed up as shown in Plot of "Flipper" Aerotech OEM1R HeNe Laser Head at Transition to Normal Behavior (Combined). Perhaps that tube was intended for a Syncrolase as it in unusual in having the required long mirror mount stem and short cutoff exhaust pipe. Perhaps it was a reject due to the flipping. Or perhaps for unknown reasons, all these tubes flip when cold. Since the Syncrolase 100 would be operating well beyond this point, there's a chance that the flipping is irrelevant and it would work just fine. In fact, that one working genuine Syncrolase tube is also a flipper until it warmed up! More on this below.

    I built a replacement coil using the wire from the first dead Syncrolase on a roll of plastic. It works, though the temperature response is faster probably because the thermocouple is not in the same location as the original. So, it locks more quickly, but also loses lock more frequently during warmup but is otherwise functional. Perhaps changing the temperature set-point would correct that. It's amazing how much variability can be tolerated with this design.

    Adjusting the temperature set-point is an interesting exercise. Ideally, it should be slightly above the equilibrium temperature of the laser head with only the laser tube powered. Set too high and the laser will run excessively hot, but there will be a fewer number of lost lock events during warmup. Set too low and it may lose lock eventually when the tube equilibrium temperature exceeds the set-point temperature.

    One way to do the adjustment might be to initially set the Temp. Gain pot (R1) fully CW (for a very low temperature) and power *only* the laser head (not the controller) for at least an hour so it reaches thermal equilibrium. Then, power up the controller and slowly turn R1 CCW to slightly beyond the point where the SYNC LED goes out. Monitoring the Temperature Amp output (A1 pin 1) will indicate how effective this is. The voltage on A1 pin 1 should remain between approximately 1.5 V and 2.75 V when the laser is locked. If it goes below about 1.5 V, the feedback loop is disabled and the heater turns on full (SYNC LED OFF). This state continues until the temperature increases to the point where A1 pin 1 exceeds about 2.75 V and the feedback loop is enabled (SYNC LED ON). Better to start out with the temperature set-point adjusted too low should the over-temperature protection fail. :( :)

    I built another temporary coil for the first laser to check it out. This coil is wound on a plastic cylinder found in a junk pile that was glued to the remains of the original coil form. The Epoxy seems to stick rather well, which is a bit surprising. I didn't have any #24 AWG magnet wire, so I used #20, which just fits 9 turns in the available space. The laser works quite well now except that the speed of heating is not quite as fast, possibly due to the coil being slightly longer and larger in diameter. However, this is probably of little consequence in the grand scheme of the Universe. :) Lowering the RF frequency improved the response, though there was no resonance.

    Finally, I built a new coil form for the third laser. It has approximately the same dimensions as the original so it behaves very well. But the plastic is too think and there is very little clearance between the form and mirror mount stem. So the genuine Syncrolase laser head won't fit because its tube is too off-center. (This must have been a result of the way it was manufactured since its beam is well centered.) But my "flipper" head fits just fine and works just fine. :)

    In fact, there's really no problem using a flipper as long as the flip point is not coincident with the rising slope of the mode where lock will occur. And if it is, simply rotate the tube by 90 degrees which will swap the polarization and the rising and falling slopes during mode sweep.

    Aerotech and Melles Griot Syncrolase Operation and Testing

    Unless otherwise noted, the following applies to both the Aerotech and Melles Griot Syncrolase.

    Power for the locking adapter is 13.5 to 15.5 VDC at 2 A max. The center contact of the 5.5 mm/2.5 mm power jack is positive. On the OEM Melles Griot, version with yellow and blue wires, yellow is positive. Also on the OEM version, the bi-color status LED is brought out to red and green wires. If no LED is already present, connect a two-pin bi-color LED between the red and green wires such that positive on red results in green light. Or wire up a pair of red and green (normal) LEDs in parallel with opposite polarity.

    A laser head with an Alden connector can be powered either from a lab supply for a 1 to 2 mW HeNe laser, set to 4 mA, or a suitable HeNe laser power supply brick. The most common brick provided with the OEM systems runs on 12 VDC. Some versions use a 12 VDC brick with external dropping network so the standard wall adapters can be used. For the laser head with built-in HeNe laesr power supply, the spec is the same as for the locking adapter - 13.5 to 15.5 VDC, center positive. (Aerotech provided a pair of identical wall adapters rated 13 VDC but which actually put out much more.)

    There is only a single user adjustment - the set-point for output power.

    There is only a single indicator - the "Locked" LED (Aerotech, red) or "Stable/Overtemp" LED (Melles Griot, green/red).

    If available, use a fast responding laser power meter to monitor the output. For the Melles Griot fiber-coupled lasers, the output of the fiber can be used, though it's best to measure power in the raw beam if it is accessible to determine the health of the laser tube. On some versions, there is a four position shutter wheel in the fiber-coupler assembly. Remove the hex cap screw in the shutter and rotate the shutter by plus or minus 90 degrees - One of these positions places a mirror in the beam path to divert the beam out the side. (Of course, if nothing comes out of the fiber when the laser is powered regardless of shutter position, the fiber alignment may be bad or the fiber may be broken - or the tube may be dead.) But not all versions have a shutter wheel. Others may have a beam-sampler plate with a silicon photodiode, so that can be used to monitor relative power.

    Before applying power, make sure the laser head is securely attached to the locking adapter with the polarization references lined up. The head should be screwed in nearly all the way with the locking ring tightened against it, and the set-screw tightened. (If there is no polarization reference mark on the laser head, the orientation will need to be determined once it is powered, see below.)

    CAUTION: For the fiber-coupled lasers, even just removing and replacing the laser head may cause alignment to be compromised. On those with an adjustable fiber port, that can be fine tuned to optimize alignment. But on those with no fiber alignment adjustments, this could be bad. So, if the head is originally tightly secured, don't loosen it. If it is already loose, then it may be possible to find an orientation very close to the optimal based on the polarization reference marks where alignment is acceptable with the lock-ring tight. Additional details are left for the advanced course. :-)

    CAUTION: DO NOT allow the locking adapter to be powered if the laser is off or doesn't start for any reason. While there is supposed to be protection against an over-temperature condition, don't count on it, especially for the older Aerotech version, for which it may have been an afterthought not present in all samples. :(

    For reference here are the measured power, voltage, and dropout current for new samples of the tubes used in the STP-910 and STP-912:

    Aerotech and Melles Griot Syncrolase Controllers, Adjustments, and Test-Points

    The controller PCB in the locking adapter has gone through several revisions over the years both by Aerotech and Melles Griot. Functionally, they are all interchangeable. So a new locking adapter can be used with an original Aerotech laser head and vice-versa. However, if swapping individual parts, the coil and PCB may be a matched pair so it's best to transfer both (though in most cases, not doing this will be close enough). Really old versions used a thermocouple instead of diode as the temperature sensor and most or all versions have a "select on test" resistor to set the lock temperature (though it tends to be the same value of 6.81K ohms in most cases).

    While all three versions are functionally similar, there is some anecdotal evidence that the Melles Griot through-hole version is actually the best and most reliable. I've seen several failures of the SMT version, usually resulting in an inability to lock or loss of lock after warmup. The adjustments also appear to be more predictable allowing for better tuning of the loop behavior.

    CAUTION: There is an RF Frequency Adjust trim-pot on all versions of the control PCB. It may be possible to set this to a value which results in excessive RF power being delivered to the tube mirror mount stem. There may be other electrical fault that can result in similar behavior. The high RF power and rapid heating can cause a hairline crack in the frit seal creating a leak, ruining the tube. The typical symptom - aside from the modes changing at warp speed when locking adapter power is first applied - is that after 30 seconds or so lasing will suddenly cease. The discharge color will then typically be blue (obvious even looking through the OC mirror), and will slowly deteriorate further over time. In addition, the plastic coil form may melt, deform, and small like hot burning plastic. If the locking adapter is powered without the mirror stem inside, it may smoke after a few seconds. The DC current will also be excessive, above 2 A.

    I've heard of this happening where a company that shall not be named managed to ruin a half dozen or more new tubes by testing them one after the other in a faulty or miss-adjusted locking adapter and wondering why they stopped lasing and turned blue. :( "The tubes must have been defective.". Forensics showed that most of them had visible hairline cracks in the frit seal. And I was able to reproduce it, first by accident while testing locking adapters with a high mileage but still usable tube, and then deliberately after only two attempts using another useless tube.

    Aerotech S100 Syncrolase controller

    Closeup of Aerotech Syncrolase 100 Controller - Left Side View and Closeup of Aerotech Syncrolase 100 Controller - Right Side View show what is probably the original design. As can be seen, there are 4 trim-pots on the PCB (From left to right: Temp. Gain, RF Frequency, PD Gain, and the user-accessible Output Level Adjust), plus the one hanging in mid-air which is Temp. Limit, probably being an afterthought added after too many locking adapters self destructed).

     Trim-Pot  Function                Comments
        R1     Temperature Gain        Adjusts temperature set-point
        R?     Temperature Limit       Prevents meltdown if laser unpowered
        R10    RF Frequency            Controls power to induction heater
        P12    P Mode Photodiode Gain  Sets range of user set-point adjustment
        R?     Output Level Set-Point  User adjustment via hole in cover

    The Schematic of Aerotech Syncrolase 100 Controller applies directly to this version. Clearly, some engineering changes were needed as in addition to the floating trim-pot, several componenst are installed at peculiar angles and generally shoe-horned into place. ;-)

    Melles Griot 05-STP-9xx Syncrolase through-hole controller

    At some point after Melles Griot acquired the Syncrolase, they redesigned the PCB (still through-hole) and added a connector so the PCB or induction coil and temperature sensor assembly could be easily replaced without requiring any soldering. They also eliminated both of the temperature trim-points - these are presumably set up during initial testing and should not change when replacing laser heads. (However, it's not clear that this is entirely true.)

     Trim-Pot  Function                Comments
        P1     P Mode Photodiode Gain  Sets range of user set-point adjustment
        P2     RF Frequency            Controls power to induction heater
        R22    Output Level Set-Point  User adjustment via hole in cover

    There is also a 4 pin header with test-points (pin 1 on the right):

     Test-Point  Function
       Pin 1     P Mode (output) amplitude
       Pin 2     ??
       Pin 3     ??
       Pin 4     RF drive (digital levels, around 10 V p-p)

    Melles Griot 05-STP-91x surface mount (SMT) controllers

    Melles Griot 05-STP-91x (Syncrolase) Controller Surface Mount PCB shows closeups of one of the latest versions of the controller I have seen. It is clearly based on the older design with some parts simply being surface mount versions of the originals. But there have been changes as well. There is also at least 1 very minor variation on this layout, the only obvious difference being the substitution of a through-hole aluminum electrolytic capacitor for an SMT cap, which was perhaps not large enough. The PCB layout with the SMT cap typically has a through-hole cap soldered to the SMT pads. Reverse engineering the SMT PCB would be even more challenging than for the older one, and no, I'm not volunteering. :-) Note that even though Melles Griot no longer calls these "Syncrolase", the PCB artwork still use that designation!

    There are 4 trim-pots on this PCB:

     Trim-Pot  Function                Comments
        P1     RF Frequency            Controls power to induction heater
        P3     S Mode Photodiode Gain  This is for dual mode option, no effect
        P4     P Mode Photodiode Gain  Sets range of user set-point adjustment
        P2     Output Level Set-Point  User adjustment through hole in cover

    CAUTION: P1 is not strictly a power adjustment. It seems to balance power when heating or locked with power when cooling - which may not be 0. So it seems possible to set P1 such that RF drive may be present even if in Overtemp (cooling) mode. Thus thermal runaway could still occur. If adjusting P1, it is best to monitor the laser output on a Scanning Fabry Perot Interferometer (SFPI) to keep track of mode movement and speed, or at least the RF drive.

    As with the previous version, there are no temperature adjustments, but a resistor that appears to be selected for each unit is soldered into standoffs (visible at the top of the board in the photos, typically 6.8K ohms).

    And here is what is known about the test-points:

     Test-Point  Function
        TP1      P mode amplitude
        TP2      Output/reference level
        TP3      Temperature sensor voltage
        TP4      Output amplitude
        TP5      S mode amplitude
        TP6      RF drive (digital levels, around 10 V p-p)

    There is also a newer version of the PCB which is almost identical but does away with the optional second photodiode and its circuitry.

    Improving Over-Temperature Protection of the Syncrolase

    Of three similar original Aerotech Syncrolase 100s I've seen, 1 had already melted, one I melted, and the other is still fine. :) There are several fundamental deficiencies in the original design that can lead to fireworks:

    Another modification (or complete redesign depending on your point of view!) that would enable the Syncrolase (or any thermally-stabilized laser) to run at the minimum temperature to assure reliable operation would to have a temperature set-point that is based on the ambient temperature of the environment, not a fixed setting. In principle, this can easily be accomplished by counting mode cycles from a cold start. Since each mode cycle represents a precise change in temperature, this would enable the laser to operate at a temperature of ambient plus a constant known to be greater than the heating from the laser tube current. A microcontroller could be used for the implementation, left as an exercise for the student. :)

    This of course assumes that the ambient temperature remains relatively constant, but this is often the case with real lab environments. The Zygo metrology lasers with digital controllers compute the number of mode cycles (they call them "mode slews") needed to reach operating temperature based on the actual tube temperature when the laser is switched on, though they may still operate at the temperature required for worst case conditions.

    While the Melles Griot version of the Syncrolase is not supposed to melt down due to a fault condition like not powering the laser, it is not known if the design has actually been fundamentally improved (as I've been told) or rather that they are simply depending on careful factory set-up and the reliability of the PCB and components. :) However, I had NOT seen any evidence of thermal damage to the Melles Griot units I've tested until a unit came in with a weak end-of-life sputtering tube and could not lock. See Melles Griot 05-STP-910 Coil Meltdown. This must have gotten so hot that parts of the coil cover actually became the consistency of mollasis with the partially intact outer ring slipping down due to gravity. The coil and temperature sensor are not even recognizable. Exactly how this happened is not clear, but it may have been an actual hard failure of the locking adapter after the end-of-life tube began sputtering and continued for who knows how long before the aroma of roasting plastic or a smoke alram caught someone's attention. :( :-) However, failure to lock results in the temperature increasing until a second threshold is reached, and then it turns off the heater drive until cooled to below the normal temperature set-point. That upper limit threshold has no adjustment. It's probably just offset a fixed amount from the normal temperature set-point, but is only a few complete mode sweep cycles.

    Using the Syncrolase Controller with Other HeNe Lasers

    I acquired a second Syncrolase 100 on eBay which also had a dead tube, but with a fully intact (not melted or crisped!) controller module. So, I did some experiments with two random polarized HeNe laser tubes. These were selected mainly because they had cathode output and enough OC mirror stem accessible for coupling to the induction heater coil.

    So, while two data points may not be conclusive, it would seem that that almost any tube that can be stabilized using the conventional heating blanket technique can also be stabilized using the Syncrolase controller if its mirror mount stem will fit inside and extend far enough into the induction heater coil. Where the tip-off is not too long but interferes with the coil assembly, simply removing the plastic cover may gain enough clearance. Of course, if you happen to be friendly with the tip-off person at a HeNe laser tube manufacturer, simply ask them to pinch-off and trim the tip-off closer to the tube! :) For longer higher power tubes, the internal preamp gain would need to be reduced to allow the Output Adjust pot to lock at higher power. Of course, for such tubes, the position on the gain curve over which the output is pure single mode would be reduced.

    And flippers will work just fine, thank you. :-) As noted above, 3 of 3 Aerotech tubes from Syncrolase lasers were flippers, at least when cold!

    In fact, I have also now implemented what I'm calling an 05-STP-938 using an 05-LHR-038 tube installed in an Aerotech cylinder from an OE2R laser. (The Aerotech cylinder has threads the front bezel that conveniently mate with the locking adapter.) The 05-LHR-038 for whatever reasons has a relatively short tip-off so coupling to the induction heater is no problem. And, of course, it too is a flipper which abruptly changes state near the peak of the gain curve! As long as it is oriented so the flip makes the output go from high to low, locking occurs reliably on the smoothly rising portion of the mode sweep. While currently only operating with intensity feedback using an older through-hole adapter, it would be a simple matter to install the modified dual mode adapter discussed below to provides nearly the same output power as the 05-STP-901/SP-117A laser and similar performance but in a much smaller package using the Syncrolase technology.

    Adding True Frequency Stabilized Mode to the Syncrolase

    While Melles Griot describes the 05-STP-909/910/911/912 as a frequency stabilized HeNe laser, it really is not since only a single photodiode is used for optical feedback. Thus, adjusting the output level also changes frequency. And frequency will also drift as the tube warms up and its maximum power increases as well as when it ages and its maximum power declines. However, in principle, it would be very straightforward to convert these lasers to use dual mode stabilization. The electronics would be trivial: If an amplitude stabilized mode is no longer desired, then it's just a matter of adding a second photodiode in parallel with the first one, with the opposite polarity, and possibly a resistor to add an offset so that the balanced condition would fall within the adjustment range of the Output Level trim-pot. Otherwise, an additional op-amp for the second photodiode would be needed to feed the Error Integrator with a switch to enable it.

    In fact, the existing optics are almost set up for dual mode stabilization. There are two beam-splitter cubes in the beam path. The first is a polarizing beam-splitter that diverts the unwanted S mode polarization 90 degrees to a beam block. The second one samples a small portion of the P mode output beam for the intensity stabilization feedback. If that beam block were removed, a photodiode could be positioned to sense the S mode.

    Based on an examination the 2008 version of the SMT locking adapter controller PCB (labelled L44000-535) and my own tests, the hooks are already in place for dual mode (frequency) stabilization, but Melles Griot has either elected not to develop the product to completion, or at least has not yet released it. Even this sample of the Melles Griot 05-STP-910-536 Reference Laser for the Agilent 86122B Multi-Wavelength Meter with a manufacturing date of 2011 and this SMT control PCB, and redesigned head mounting and adapter case, still only uses a single mode for locking.

    Note the location on the underside of the Melles Griot 05-STP-91x (Syncrolase) Controller Surface Mount PCB for a second photodiode (D5), currently un-populated. The lower (on the photo) pins of both photodiode are connected, but not the upper pins, so they are not simply in parallel. In addition the wiper on trim-pot P3 (which has no effect on normal operation) connects directly to the upper pad of the un-populated photodiode while the wiper on trim-pot P4 (P mode gain) connects to the upper pad on the installed photodiode (which provides the intensity feedback). So, P3 is used to adjust the S mode gain for dual mode stabilization. The only slight problem is that for unfathomable reasons, the position of the second PD doesn't quite line up with the S mode from the polarizing BSC. But moving it a few mm would fix that. The only other change to the optics might be a filter in front of the S mode PD to equalize the S and P mode sensitivities - the diverted S mode beam is much stronger than the sampled P mode beam since most of it exits out the front of the laser.

    I have now confirmed that moving the mystery jumper near the edge of the PCB to the opposite position indeed turns on the dual mode option. It both enables light incident on a second PD installed at D5 to affect the lock point AND shifts the set-point of the Output Level trim-pot (P6) so that if P3 and P4 are adjusted properly, the locked output amplitude will be similar when switching between 1 mode (intensity, I) and 2 mode (frequency, F) operation. The first crude test was to use a flashlight to "adjust" the output level. ;-)

    So, with some relatively minor surgery, it is possible to modify the Melles Griot Syncrolasers with this version of the SMT PCB for dual mode operation and true frequency stabilization! But rather than moving the first BSC, it may be easier to use a smaller photodiode - the existing one has a sensor area about 10 times what it needs to be even accounting for normal laser tube alignment tolerances - and simply install it with its pins bent to shift the sensor area by the required distance. There is a cover over the optics that needs to have a second hole drilled for the S mode PD but this is straightforward and would help to keep the PD in position.

    And so it was done! After drilling the hole and squeezing a small photodiode (approximately 2x2 sensor area) from a barcode scanner in there, I may now have the only dual mode Syncrolase on the planet (or at least outside Melles Griot). With the jumper in the 2 mode position, both P3 and P4 have an effect and the lock set-point can be moved over a wide range. It's possible that there should really be a filter in front of the P mode PD as P3 is a rather sensitive adjustment, but the thing does work. And the Output Level trim-pot can still be used to move the lock position. (Though, strictly speaking, with these basic dual mode techniques, best frequency stability requires the P and S mode amplitudes to be equal on either side of the neon gain curve, and that occurs only at one setting.) Since its useful range can be set up to be approximately the same for both intensity and frequency stabilization, an external SPDT switch was added. See to access internal adjustments during normal operation as shown in Melles Griot 05-STP-91x (Syncrolase) Locking Adapter with Dual Mode Option Added. There is ample room to mount the switch in the cover but for this OEM version, there didn't seem to be much point. ;-).

    Whether anyone really cares may be another matter. Most users don't have a clue about the difference between intensity (1 mode) and frequency (2 mode) stabilization. Since it's easier to measure output power, intensity stability is something most users can understand and easily measure. Checking frequency stability requires more sophistication such as the use of an ultra-high precision wavemeter or comparison with a frequency reference like an iodine stabilized HeNe laser. The novice user will observe that with frequency stabilization, the output power may change significantly until the tube has reached thermal equilibrium and assume that something is wrong. With intensity stabilization, the change is very small. After full warmup, frequency drift is quite small with either technique.

    However, the competition generally implements both frequency and intensity stabilization, and additional expense is quite small, so why not do it! Hear that, Melles Griot? :) In fact, the S mode photodiode active area is way more than it needs to be even accounting for any possible differences in pointing alignment of the laser tube. Using smaller cheaper photodiodes should more than make up for the added parts cost.

    Regrettably, it seems that the latest version of the SMT controller PCB (L44001-159) has done away with all traces (no pun) of the second photodiode and its circuitry, which doesn't make any sense since they weren't causing any harm by being present. :( :) It's virtually identical in every other way but te hooks are gone.

    Relative Frequency Stability of 05-STP-910 and 05-STP-912 Lasers

    While the published stability specifications for the Melles Griot 05-STP-910 and the higher power 05-STP-912 are identical, this is almost certainly not correct. Don't tell Melles Griot. :)

    The lasing output power profiles with respect to neon gain center for the tubes used in these lasers differ dramatically. Since the STP-91x lasers are intensity (not frequency stabilized), this will directly translate into a significant difference in how the optical frequency (or wavelength) will be affected by both the output power setting (user adjustment) and power variation of the tube (due to alignment changes or how it ages with use).

    Comparison of Output Power Profiles of the Tubes used in the Melles Griot 05-STP-910 and 05-STP-912 Stabilized HeNe Lasers shows the behavior of new samples of each tube. These plots show output power versus optical frequency for one polarization of the STP-910 and STP-912 tubes over more than one half of a mode sweep cycle. This composite was created by starting with the mode sweep versus time plots for new or nearly new samples of the bare tubes used in the STP-910 (LHR-704) and STP-912 (LHR-912) lasers:

    Time during mode sweep as the cavity length changes from heating or cooling can be directly related to changes in optical frequency. (As a result of the limited extent of the neon gain bandwidth, these changes are not monotonic - there are mode hops. However, the central portion will be continuous in so far as the lasing modes of interest are concerned.) One complete cycle (red or blue) represents a change in cavity length of one wavelength (at 633 nm) and a change in optical frequency of 2 times the mode spacing of c/2L. The additional factor of 2 arises because the adjacent modes are orthogonally polarized. Note that while the profile of the mode sweep is affected by the neon gain curve, the period is NOT directly related to it. The mode sweep plots for the two tubes were normalized so that the output power scale (in percent of peak) and frequency scale (50 MHz or 0.067 pm - picometer per major division) were made the same. These were combined to form the composite.

    The cause of the asymmetry for the 05-STP-912 is not known since these should both use single isotope 20Ne. Perhaps mode competition - only really present over a significant region - is impacted differently depending on neon isotope mix. And note the small dip in the profile of the tube used in the STP-912. This was seen in two different samples that passed Melles Griot testing so it's probably not a defect in these samples. Such unsightly blemished might make locking within the small range of output power including the blips problematic. But that's another story.

    Short tubes oscillate on at most 2 longitudinal modes and the range of cavity length - and thus optical frequency - where there are 2 modes is small as with the STP-910. And only a single mode is present at all other times. However, as the tube length increases, there will be periods where up to 3 modes oscillate. These compete with each-other in such a way that the transition is much wider - as is the case with the STP-912. As a result, the curves for the STP-910 have much steeper sides than those of the STP-912. When the locking adapter is set at around 85 % for the STP-910 or 60% for the STP-912, both tubes will lock at around the same optical frequency offset from the neon gain center. However, setting to a different power, as well as power changes of the laser tube over time will shift the lock point of the STP-912 more than that of the STP-910, by a factor of up to 2 or more.

    Melles Griot has always published identical specifications for the stability of the two lasers. Unless the present STP-912 - the only stabilized HeNe laser they currently offer - uses a different design (which is highly unlikely unless they now implement dual mode frequency stabilization as deesribed in a previous section, which is even more unlikely), the STP-912 is actually not as good as the discontinued STP-910 in this regard. The specifications can be found by searching the Web for "Melles Griot Stabilized HeNe Laser".

    Also of note in those specification is one for "tuning range" which has units of MHz. The only way to tune the frequency of these lasers is to adjust the output power via the user accessible trim-pot, which has the side effect of changing the optical frequency. The range for the STP-910 is spec'd to be 50 to 600 MHz; for the STP-912 it is 400 to 600 MHz. The limits are based on (1) retaining stable lock and (2) the laser remaining single longitudinal mode. However, those limits are almost certainly reversed: The STP-910 has a much smaller range of tuning in terms of optical frequency than the STP-912 for the reasons outlined above. This can clearly be seen on the plots. ;-)

    Since the vast majority of users do not have the capability to test a laser to this level of precision and simply depend on specifications, few will ever notice. Based on the plots, setting the STP-912 to 60% of the peak value should result in an optical frequency similar to that of the STP-910 set at 85%, around 550 MHz from the neon gain center. And in this region with its higher slope, the sensitivity of optical frequency of the STP-912 to changes in laser tube power is less 1.5 times that that of the STP-910. Given the higher maximum output power of the STP-912, this will still result in greater power compared to the STP-910 but possibly not by that much. However, for a critical application like a wavemeter, it might be best to set the lock point not based on power or percent, but by actually using the wavemeter to measure a laser which has a wavelength known to many decimal places. Of course, this is also not generally possible outside of a place like NIST. ;-)

    The conclusions are that while the STP-910 has a narrower tuning range (and lower power) than the STP-912, the absolute optical frequency is likely to be more stable over its life.

    And an interesting footnote to all this: LHR-704 tubes that are extremely lively with an output power around 1.3 mW or more may have a modes that have a dip at the top and less steep sides due to mode competition, a more limited single longitudinal mode amplitude adjustment range than normal, and multi-spatial mode output where power in each mode is close to minimum (which probably little practical consequence).

    Aerotech LZR2000 and LZR3000

    The Aerotech LZR2000 and LZR3000 are single frequency stabilized HeNe lasers used for precision interferometric metrology applications such as machine tool calibration and lithographic stage positioning. The primary difference between the two models is that the LZR2000 has a built-in optical receiver while the LZR3000 requires an external optical receiver for each axis. So, the LZR2000 is more suited to single axis measurements (though either laser can be used with single or multiple axes). (With either, an interferometer is required for each axis.) It appears as though only the LZR3000 may be a current product (and the only known actual laser possibly still made by Aerotech). Even this is somewhat questionable since there are few references to it on the Aerotech Web site (and none for the LZR2000). However, The LZR2000 and Laser Feedback System Operation and Technical Manual P/N: EDU169 (V1.0), dated October, 2000, is a complete user guide for setting up the LZR2000 laser and optics to make position and velocity measurements in a single axis system and includes laser connector pinouts as well as a decent amount of general information. Like the functionally similar single-frequency lasers from Teletrac/Axsys and Mark-Tech, as well as the two-frequency lasers from HP/Agilent, Excel, and others, they use various configurations of polarizing beam-splitters, retro-reflectors, and mirrors, to implement very precise measurements based on interferometry. The configurations as well as accuracy and error considerations are the same regardless of specific manufacturer and thus the manual really applies to all of them. However, depending on implementation (and spec'smanship!) of the electronics, realizable usable resolution will vary.

    LZR2000 Laser Head Description and Specifications

    The LZR2000 is a solidly built unit with leveling feet at the front corners and center back. Power is provided by a standard IEC power cord with 8 pin DIN and DB9 connectors for signal output. A shutter wheel on the front has OFF, ON, and TARGET positions. TARGET is used for alignment and has a pinhole aperture for the laser output and cross-hair for the return beam.

    Laser head specifications

      Laser type:              Helium-Neon (HeNe), single frequency.
      Maximum output power:    1 mW.
      Warm-up time:            15 minutes maximum.
      Vacuum wavelength:       632.9907 nm
      Wavelength accuracy:     ±0.1 ppm.
      Wavelength stability     ±0.002 ppm/hour, ±0.02 ppm/month.
      Beam diameter:           5 mm.
      Beam centerline spacing: 11 mm.
      Safety classification:   Class II.
      Power requirements:      50 W at 100 to 240 VAC.
      Output signals:          λ/2 A-quad-B complimentary line driver
                                output, λ/2 A-quad-B complimentary
                                sinusoidal output (0 to ±1.5 V peak)
                                "Laser on" and "Laser Ready" (TTL).
      Dimensions (LxHxW):      15.38 x 5.50 x 4.04 in (390,6 x 139,7 x 102,4 mm)
      Weight:                  12 lb (5,5 kg).
      Operating Temperature:   15 to 40 °C.
      Relative Humidity:       0 to 90% non-condensing.
      Shock (IEC 68.2.27):     30G, 11 msec.

    LZR2000 Laser Head Connector Pinouts

    The LZR2000 has a built-in optical receiver, so the processing electronics for a single-axis system attach directly to it. The only outputs are the baseband quadrature of which there are two flavors: The SIN/COS analog signals which maintain full position resolution and the SIN/COS digital line driver signals which are limited to &Lamda;/2 (linear optics ) or &Lamda;/4 (plane mirror optics) resolution.

    8 pin DIN

      Pin    Name     Description
       1     SIN A    A-quad-B SIN output signal (high active)
       2    ~SIN A    A-quad-B SIN output signal (low active)
       3     COS A    A-quad-B COS output signal (high active)
       4    ~COS A    A-quad-B COS output signal (low active)
       5     GND      Ground
       6     VDC      +5 VDC
       7   Laser On   Signal that corresponds to the "Laser On" LED
       8    Stable    Signal that corresponds to the "Laser Ready" LED


      Pin    Name    Description
       1     SIN     A-quad-B line driver output signals (high active)
       2    ~SIN     A-quad-B line driver output signals (low active)
       3     COS     A-quad-B line driver output signals (high active)
       4    ~COS     A-quad-B line driver output signals (low active)
       5     GND     Ground
       6     VDC     +5 VDC Output Signal
       7   Laser On  Signal that corresponds to the "Laser On" LED
       8    Stable   Signal that corresponds to the "Laser Ready" LED
       9     GND     Ground

    LZR2000 Laser Head Operation

    When AC is plugged in (there is no ON switch!), the laser goes through its locking routine consisting of preheating the laser tube until it is at a high enough temperature for the thermal stabilization loop to be closed. The warmup time is less than 15 minutes, during which, the red Laser ON LED will go on and off since its intensity is proportional to the polarized mode that exits the laser. Once stabilized, both it and the green Laser Ready LED will be on solid. But under some conditions if the laser fails to stabilize, that green LED will turn red. :) More below.

    LZR2000 Laser Head Construction

    As expected, the LZR2000 is based on the Syncrolase (S100) design, but has been completely repackaged. Because of its higher class surroundings :), this wasn't conclusively determined until the wires for the induction heater coil, temperature sensor, and single mode photodiode were located. And to eliminate any possible lingering doubt, the Syncrolase patent number (4,819,246) is printed in large type on the case! The HeNe laser tube is shorter than those found in the original Aerotech S100 but is similar to the tube in the Melles Griot 05-STP-909/910. (It's possible that Aerotech had an equivalent model, but I haven't seen one.) The tube is only about 146 mm in total length with a longitudinal mode spacing of about 1.1 GHz (136 mm cavity length). Such a tube runs single mode for most of mode sweep, but this is really of no consequence since the locking circuitry will maintain it well within the SLM region even for a much longer tube that has 2 or even 3 modes at times. Exactly why the shorter tube is used is not known but probably simply because its output power is adequate for the metrology application.

    The Control PCB appears to have essentially the same circuitry as the standard S100 (with the same 5 trim-pots), but uses SMT devices for most ICs and discrete components. It also has the preamps for the optical receiver SIN, COS, and total amplitude channels. (There are 3 photodiodes.) There is also an LCD display with backup battery for a digital runtime meter. Interestingly, that meter appears to have a jumper-selectable option to only run when there is a return beam present to the optical receiver, but I can't confirm that. Perhaps the system could be rented on an hourly basis and only charged against actual use! ;-) A universal switchmode power supply provides 15 VDC for the HeNe laser power supply brick (10 to 15 VDC input, 1,500 to 2,000 V output, adjustable current) and controller PCB.

    Several photos of the LZR2000 laser head can be found in the Laser Equipment Gallery (Version 4.06 or higher) under "Aerotech HeNe Lasers".

    The LZR2000 I acquired was in excellent physical condition except that many of the screws were a bit rusted and required housekeeping services to clean or replace them before doing anything further! :) Whatever humidity was present appears to have had no effect on anything else. Its manufacturing date is 1996 though there is a second "system" sticker with a date of 1999. When first powered, the tube started sputtering after a couple minutes indicating that it probably was high mileage and the dropout current had increased. Can you believe it?! A slight increase in the tube current allowed it to run stably. The runtime meter backup battery in my sample is quite dead so I have no idea how many hours it has run. The raw output from the tube (before all optics) is almost 0.75 mW after warmup, so that's not so terrible, though probably somewhat less than when new. However, when powered up, it wasn't obvious if the controller was doing anything. The normal behavior for a standard Syncrolase is to immediately initiate a rapid ramp-up of temperature with the induction heater turned on full. With this laser, initial behavior was more like a cross between normal mode sweep and occasional hiccups or pauses. (I guess those glitches should have been a tip-off that something useful was happened!) However, allowing it to warmup for the required 15 minutes did result in the Ready LED coming on solid green. But locking did not occur and the modes then continued to come and go, gradually slowing down as though it was reaching thermal equilibrium. I assumed that either there was a controller hardware problem, or an adjustment was required, possibly due to the tube power being lower than normal. Optimistically assuming the latter, I figured that the only trim-pot that would be safe to adjust would be the one for the power output set-point. I didn't want a repeat of the core melt-down that I had with one of my S100s. Unfortunately, none of the trim-pots are labeled as to function. Although there are the same number of trim-pots associated with the locking circuitry, part numbers don't correspond to the ones on the original Syncrolase S100 PCB and even some of the part values have changed. But there was a clue: All but one of the trim-pots was very well sealed to prevent twiddling. Only the large blue multi-turn trim-pot facing forward had no Loctite™. Turning this definitely had an effect of changing the mode sweep rate dramatically at times. And, if turned too far CCW, the Ready LED turned *red*. This is not documented anywhere. It's supposed to be green or off! I assumed red meant that something really bad was about to take place. However, after some random twiddling, a remarkable thing happened. :) The laser settled down and locked, with the adjustment then having the expected effect of varying the output power. I assume the output level was simply set much too high for this tube that had weakened from long hours slaving at whatever it was doing. Thus, once the tube reached operating temperature, the controller was frantically searching for the lock point, which was impossible to achieve. Now it's set at around 350 µW, which is plenty of power for a metrology laser (especially one that will likely never be used in a real application ever again!). The laser now locks without issues from a cold start, though it sometimes requires much more than the spec'd 15 minutes. This tube's mode sweep profile has a nearly flat top and rather steep sides as shown in Mode Sweep of Aerotech LZR2000 Stabilized HeNe Laser. The top plot shows how the output would look if the laser were allowed to warm up in the normal manner which the bottom plot is of an actual run from a cold start to beyond where locking occurs. To guarantee stability with the single mode intensity stabilization technique, the lock point must be safely on one of the side regions. As such, it would probably lock reliably at well over 400 µW as the peak output power is almost 500 µW. But note the peculiar behavior in several places in the bottom plot. At several locations startup to just beyond the halfway point, there are multiple mode flips where the power in the two modes would be almost precisely equal (if the other one were shown). You can tell I'm not totally surprised, having seen flipper tubes in Aerotech Syncrolase lasers in the past, but it's still an aesthetic problem. ;-)

    A pair of anomalies are present once the laser locks. Whether these are problems (or features) associated with this specific laser, I do not know. But how can there be anything wrong with it? :) The following is from one run recorded at 60 samples per second with a data acquisition system:

    1. Over the course of the first hour or so after locking, there were two "glitches" where the output power dropped and then the stabilization controller recovered in a few seconds, with some small amount of ringing (similar to when it initially locks). No others occured in the next several hours. These are almost certainly NOT mode flips since their amplitudes differed significantly, neither amplitude was consistent with that of a mode flip, and the initial onset was not instantaneous as it would be with either a spontaneous or tube dropout-induced mode filp. The only likely explanation is for the mirror spacing to have changed. And based on their amplitude and the shape of the lasing output power curve during mode sweep, by less than 50 nm. The most likely cause is "stick-slip noise" resulting from slight movement of the tube bore within a metal spider or outer concentric glass centering structure, or of the entire tube as it expands within its mount. The ugliness is shown in Stick-Slip Events in Aerotech LZR2000 Stabilized HeNe Laser. Mechanical changes are slow compared to electrical or mode filp events. And the existence of events such as these while the structure has not yet reached thermal equilibrium, but none thereafter, is further confirmation of this hypothesis. I really should demand a replacement laser under warranty! ;-)

    2. The output power oscillates with an peak to peak amplitude of 1 or 2 percent of the total, starting out with a period of around 20 seconds just after locking and slowing down until there is little change detectable after several hours. This may result from a weak etalon effect as the distance between between one or more pairs of optic surfaces changes due to thermal expansion. Candidates include the tube OC mirror, beam sampler, and beam expander telescope - some of which are not even AR-coated. Such slightly variations are not uncommon with metrology lasers and are generally of no consequence.

    I do not have the slightest intention of investigating these much further. :)

    Aerotech Interlase 300SF Laser Interferometer with FCD300 Fringe Counter

    In the early 1990s, Aerotech may have decided to investigate the possibility of using their Syncrolase single frequency laser technology for metrology applications in addition to or in place of the LZ2000/LZ3000 lasers. The Syncrolase was much more compact and had proven performance including good stability and a fast time-to-lock. This would be a homodyne system similar in organization to Interferometer Using Single-Frequency HeNe Laser. However, the details of the fringe detection differ, using mulitple photodiodes to monitor the actual fringe pattern, rather than the usual beam-splitters and QWP. U.S. Patent #5018862: Successive Fringe Detection Position Interferometry covers the general principles. Whether this is really any better may be questionable, but it did give Aerotech a patented "feature". ;-)

    The Interlase 300SF consists of a stabilized single frequency HeNe laser with linear interferometer optics and A-quad-B detector packaged in a massive very nicely machined housing that looks something like a Syncrolase on steroids. Everything in the diagram above except for the Test Arm (moving retro-reflector) and Signal Processing is within the 300SF laser head. The laser implementation is probably functionally similar to the Syncrolase 100 using an induction heater on the cathode mirror mount stem for cavity length control. A linear interferometer consisting of a Polarizing Beam-Splitter (PBS) cube and reference Retro-Reflector (RR) comprises the built-in interferometer, which is in both the output and return beams. After passing through the interferometer, the return beam is combined with the local reference beam to create a frings pattern. Photodiodes analyze the fringe pattern in much the same way as a rotary optical encoder. The Signal Processing is done by the FCD300 Fringe Counter.

    A pair of PCBs in the laser head implement the stabilization and A-quad-B detector functions. The locking adapter PCB is about the same size as the one in the Syncrolase and appears to have similar circuitry, but the layout has been modified. The second PCB is totally unique to the 300SF.

    The 300SF laser head mates with the FCD300 Fringe Counter which is in a nice box with a HeNe laser power supply brick, DC power supply, and display consisting of a pair of ICM7217 quad BCD counters driving an 8 digit LED readout with sign. So, the value is indeed the fringe count which translates to multiples of 1/8th of the 633 nm laser wavelength, and NOT an actual displacement. This is so pathetically primitive that my original SGMD1 TTL Measurement Display is more capable. :) AND, the logic is wire-wrapped on a piece of Perf. board, not a PCB!! There is no serial number on the FCD300, so the controller box may have been a prototype. (The 300SF does have a serial numbeer.) Also with it was a hand-made breakout box to monitor the SIN, COS, and INT signals from the laser head to be used during electronic adjustments.

    While the 300SF laser is well made and could certainly be a competitive product, the FCD300 is a joke. There is no output interface to send the readings to a computer for conversion to displacement and enable environmental correction, or for other processing or logging. And with no output interface, there is no possibility of adding interpolation via post-processing, so the resolution is poor as these things go. Exactly why Aerotech would develop such a mediocre readout in the 1990s is not clear. This being decades after HP's laser interferometers have proliferated in the metrology market and other manufacturers have much more capable competing products. Perhaps it was simply an in-house test unit thrown together in preparation for the development of a more sophisticated measurement system to compete with Teletrac/Axsys or other similar systems and it got canned after the acquisition by Melles Griot.

    Now it's possible there is a successor to the FCD300 which is more sophisticated, but it's hard to believe I haven't come across one after decades of searching. :) Of course, this is the first time I've seen the 300DF and FCD300! ;-)

    Here are some photos:

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.

    Coherent/Tropel Stabilized HeNe Lasers

    Coherent Model 200 Single Frequency Stabilized HeNe Laser

    The CO-200 is a HeNe laser that operates in a single longitudinal mode. It consists of a cylindrical laser head and separate controller/power supply as shown in Coherent Model 200 Stabilized HeNe Laser. The only front panel control is the power switch on the left. The only indicator is a red "Wait"/green "Ready" lamp on the right. Although the CO-200 design dates from the late 1970s, I've seen systems with post-2000 manufacturing dates, likely for some OEM customer, possibly used in a high accuracy wavemeter as the reference laser.

    Some of these lasers are badged Tropel, which is the company that originally developed them. Coherent owned Tropel from 1972 to 1982. According to a former Tropel engineer who was involved with the optics design, the model 200 may have been the first commercial laser to use dual polarized-mode stabilization based the paper: R. Balhorn, H. Kunzmann, F. Lebowsky, "Frequency Stabilization of Internal-Mirror Helium-Neon Lasers", Appl. Opt. 11, 742 (1972).

    The HeNe laser head is powered from a standard Laser Drive 6.5 mA, 2,100 V power supply brick via a HV BNC connector. There is no special control or regulation of this supply - it's turned on by the main power switch. But some thoughtful engineer included a high resistance bleeder to discharge the HV caps in the power supply brick after power is removed. :)

    The HeNe laser tube itself is a Melles Griot (not made by Coherent!) model, labeled 05-LHR-219-158. It has similar dimemsions to an 05-LHR-120, a common 2 mW (rated) random polarized laser. But, the -158 may mean it has been specially selected to have a well behaved mode sweep cycle (not a flipper!) for this application. It may also be filled with isotopically pure (or at least enriched) gases and an AR-coated HR (to minimize back-reflections from the HR's outer surface). The tube itself puts out more than 2 mW when new - possibly up to 4 mW or even more - but the polarizing and beam sampling optics sucks up some of it. In addition, depending on the particular version, there is either a dielectric filter or polarizing filter in the end-cap. The dielectric filter cuts the output by about half but the this can be varied by 10 percent or so (though I'm not sure if this is intentional or just a byproduct of it being angled). The polarizing filter allows continuous adjustment of output power. (In both cases, the adjustment is done by loosening a set-screw and rotating the end-cap). According to the CDRH sticker, the output beam is supposed to be less than 1 mW. Given the wide swings in output power during warmup (see below), even with 50 percent attenuation, the peak output power may approach 1 mW. But regardless of the type of end-cap, only a single polarization ever exits the laser since the internal beam sampler blocks the other one.

    There is a thin film heater attached to a thick rubber jacket between the tube and laser head cylinder. A beam sampler assembly consists of a pair of Beam-Splitter Cubes (BSCs) in series and two photodiodes, each associated with one of the BSCs. The first BSC is a polarizing beam-splitter and reflects the full power of one polarized mode to its photodiode. Thus, the beam that passes through it is linearly polarized with the orthogonal orientation. The second BSC reflects 10 or 20 percent of this mode to its photodiode. So, the output beam from the laser is pure linearly polarized and has slightly less output power than one of the polarized modes of the tube. The controller monitors the lasing modes and maintain cavity length using the heater so that a pair of orthogonally polarized longitudinal modes straddle the gain curve. The beam sensor assembly can be rotated to align the photosensors with the 2 orthogonal lasing modes as this is arbitrary from tube to tube, and orientation within the cylinder, but should remain fixed for the life of the tube.

    The controller can be set up to run on various input voltages from 100 VAC to 240 VAC by changing the position of a small PCB that plugs into the AC entrance assembly, and plugging in the appropriate fuse. However, it seems that the HeNe laser power supply always runs on 115 VAC from a tap on the main power transformer so it doesn't need to be capable of 230 VAC operation, even though the one that's in there has that option - the wire for 230 VAC is not used! The output of the HeNe laser power supply is rated 2,100 V at 6.5 mA with no start delay.

    The user controls consist of one (1) power switch. There are indicators for AC power and Status. After a warmup period of 20 minutes or so for the laser head to reach operating temperature, the Status indicator will change from WAIT (red) to READY (green). Doing anything that causes lock to be lost will result in a shorter delay of a couple minutes to re-establish it.

    The internal circuitry of the controller box is relatively simple and includes a pair of LM3403 quad op-amps, a 741 op-amp, and LM311 voltage comparator, along with a TO5 power transistor on a heatsink to drive the heater.

    Here is the pinout of the circular control connector as determined by my measurements. There may be errors.

      Pins  Wire Color  Function      Comments
      1,2    Blk/Wht    Heater Power  ~22 ohms
      3,4    Blk/Red    Temp Sensor   ~880 ohms at 25 °C, ~1.2K when locked
      5,6    Blk/Blu    Photodiode A  Anode is pin 5; Approximately 250 µA max
      7,8    Blk/Grn    Photodiode B  Anode is pin 8; Approximately 50 µA max

    It would appear that the difference in sensitivities is the way it's supposed to be since this was similar on 3 heads. (However, the readings on an analog VOM for the photodiodes did differ on 2 heads I tested - I'm not sure what, if any significance, that has.) This makes sense given that the sampling is done from the main beam. The A mode is blocked entirely and thus the associated photodiode gets its full intensity. The B mode would then seem to be sampled at about 20 percent intensity. The controller and laser head are normally a matched pair and there is an adjustment inside the controller to equalize the responses.

    The response of a typical silicon photodiode is 300 to 400 µA per mW at 633 nm. But the photodiodes in the beam sampler assembly have frosted domes to minimize the effect of beam position and this also reduces the response by about 70 percent. For a typical laser tube with 2 mW maximum in each mode, 250 µA would be a reasonable response. And 50 µA would be reasonable for the sampled mode.

    U1 in the schematic has the amps for the A and B modes on pins 14 and 1, and their difference (pin 7). Monitoring the difference during warmup and adjusting the Mode A gain control (R1) so that the voltage swings from near 0 to approximately 22 V as a function of mode sweep will get close to optimal; then fine tune it once the green READY light comes so that the output power is approximately half way between the min and max near the end of mode sweep. More below.

    The heater consists of a serpentine thin file metal pattern on a rubbery backing material that wraps completely once around the tube.

    The temperature sensor extends the length of the tube and is buried within the heater backing, technology unknown.

    Here are some photos.

    There are links to the operation and service manual and schematics below.

    I picked up a controller and 3 laser heads in two separate eBay auctions for a grand total of $22.50 + shipping. The serial number on one of the heads matched that of the controller and while this head was initially hard to start, after running it for awhile on my HeNe laser test supply, it now starts normally.

    The controller originally had a dead HeNe laser power supply brick (Laser Drive 314S-2100-6.5-2, 2,100 V at 6.5 mA) which is likely the reason it was taken out of service. I replaced that with an Aerotech LSS-5(6.5) which seems to be happy enough. Using a laser power meter, one of the two modes of the laser (the one present in the output beam) could be seen cycling up and down between about 0.60 and 1.40 mW with the orientation of the beam sensor assembly adjusted for maximum peak power. Each cycle took longer and longer as the tube warmed up to operating temperature, helped along by the heater. After about 15 minutes, it would appear to try to "catch" at certain power levels but couldn't quite remain there. (This behavior may have had nothing to do with the feedback control though.) Then suddenly, after about 20 minutes, the Ready light came on and a few seconds later, it locked rock stable at 0.95 mW. :) A second laser head behaved in a similar manner but with a slightly higher final output power of 1.02 mW. No adjustments were needed inside the controller despite the fact that the second head's serial number didn't match the controller's serial number. Possibly, even better stability or slightly higher stabilized output power could be achieved with some fine tuning. (The 1.02 mW head actually had higher peak power than the 0.95 mW head. The difference is probably in part due to the photodiode sensitivities.) With the fixed filter end-caps installed, the output power dropped to around 0.50 mW. I rather suspect that these are normal power levels for this system. (This was later confirmed when a manual with detailed specifications turned up.) The third head had its cables cut but I finally scrounged a replacement control connector from a box of junk in the garage and jerry-rigged the HV BNC for testing. That laser head now works as well. It also came with an adjustable polarizer in its end-cap. With that installed on either of the other heads, the output power could be varied continuously from near 0 mW to about 1 mW.

    Note that the Ready light comes on and then the laser locks in at the proper phase of the next mode cycle. So, basically the pea brain in the controller (no actual CPU of any kind!) decides that conditions are suitable and enables the feedback loop. The final "decision" is based the cycle duration being longer than some magic number (around 1 minute). :) I've also seen the ready light come on even if the laser doesn't start and when one of the previously locked heads was plugged back in after a few minutes of cooling. In the latter case, the laser was indeed locked though it might not have been able to maintain it continuously since the tube was probably no longer really warm enough.

    There are actually two feedback loops in the controller. During warmup, the heater is driven to a fixed temperature based on the resistance between pins 3 and 4 of the Control connector. Once the period of the mode cycle exceeds a fixed time (guessing somewhere around 60 seconds), the control loop based on the difference of the photodiode outputs is enabled. The same signal that switches from the temperature feedback to mode feedback turns the Wait indicator goes off and the Ready indicator on. More on this in the next section.

    Plot of Coherent Model 200 Stabilized HeNe Laser Head During Warmup and Plot of Coherent Model 200 Stabilized HeNe Laser Head Near End of Warmup show the output power variation due to mode cycling. Note how it seems to "snap" into regulation once the time is right. :) There are roughly 90 mode cycles during warmup prior to lock. The internal optics account for the large variation in output power. The HeNe laser tube itself has a normal mode sweep of only a few percent.

    Note that if there are no mode changes detected but Mode A is higher than Mode B - for example if the laser tube is weak or the Mode A Gain is set too low - the controller will still go through the same process but READY will come on almost immediately and the lack of mode signal may drive the heater on full. While the laser head when locked normally runs almost too hot to touch, it will be scorching hot with this failure mode.

    Another Coherent 200 system I have has a fully functional controller but a fully dead laser head. It is very hard start, impossible to run, and way beyond end-of-life. So, that gave me an excuse to go inside.

    The Coherent 200 laser head can be disassembled in a reversible manner with fewer individual parts than the Spectra-Physics 117/A or the essentially identical Melles Griot 05-STP-901. However, it doesn't come apart as easily, using a press-fit for the tube/heater sandwich.

    As noted above, the tube was found to be way beyond end-of-life. If it could be convinced to start (on a lab power supply), it would not run at any reasonable current and produced no output at all. There was sputtered aluminum coating on the holes near the cathode end-cap and even through holes in the cathode can near the center of the tube. This system had obviously been left on continuously for a large number of years. It was probably not even in use for a good portion of that time, forgotten and lonely in a corner of a lab, wasting its life producing coherent stabilized photons no one was using until there were no more! :) That seems to be the destiny of so many stabilized HeNe lasers. I'll be searching for a suitable replacement tube. The original tube, a 05-LHR-219 (with or without a -158), doesn't show up in any list I've seen) but an 05-LHR-120 has nearly the same dimensions and will run on the same power supply. So, as long as one can be found that is well behaved (non-flipper, wedged HR), it will almost certainly work fine. Other random polarized laser tubes of similar length can also be adapted but may require replacing the HeNe laser power supply and coming up with a creative mounting scheme if diameter is smaller.

    An operation manual and application notes for the Coherent 200 can be found at Coherent Model 200 Operation and Service Manual. Get it while you can as Coherent has been known to complain about docs being on-line, even if for 30 year obsolete lasers! :( :)

    Schematics of Coherent Model 200 Controller

    I have reverse engineered the very nearly complete circuit diagram for the Coherent Model 200 controller. The only parts not included in the diagram are the AC line front-end and voltage selection circuitry, the power transformer for the low voltage power supply, the Laser Drive HeNe laser power supply, and the detailed wiring of some of the internal connectors. (The only internal connector on the drawing is J2, for the cable from the PCB to the laser head Control connector on the rear panel. It is labeled J2 on the PCB copper.)

    Everything is in Schematic of Coherent Model 200 Stabilized HeNe Laser. Note that most of the part numbering is totally arbitrary as there were *no* part numbers on the PCB except for the PCB connectors (and I only have J2 in the drawing). This is a late revision with PCB artwork dated 1997, though that probably only means that there was a PCB fab run in 1997, since the artwork itself was obviously hand taped. :) I guess some important customer just had to have more of these lasers made well after they would have been considered very obsolete by Coherent. :)

    The controller has two feedback loops. The Preheat Loop, which is active while the tube is warming up, drives the heater in the laser head to a fixed temperature (set by a pot). The temperature sensor in the laser head is not a common NTC thermistor, but something that increases in value with increasing temperature. It has a resistance of around 800 to 900 ohms at room temperature, but well over 1K ohms at operating temperature. The preheat loop prevents the mode feedback loop from going active until the temperature is sufficiently high. Only after this occurs, does a timer begin to look at mode changes, and switches from the preheat loop to the mode feedback loop once their period exceeds around 60 seconds. The mode feedback loop uses the difference between the orthogonally polarized A and B modes in a simple PI control loop to drive the heater. Should the laser not stabilize as evidenced by mode changes still occurring, the preheat loop will be switched back on to try again. At least, that seems to be how it's supposed to work. However, a system with a laser tube that doesn't start (or a bad HeNe laser power supply) will likely turn on READY shortly after being powered up even though it is obviously not working correctly. Well, I guess it IS quite stable - dead with a frequency of exactly 0.0000000000 Hz and an output power of exactly 0.0000000000 mW! :)

    Note: Some versions of the controller PCB lack the Temperature and Loop Gain pots (R1 and R20, respectively). I'm not surprised about the absence of R20 as it never seemed to do anything useful, but the lack of R1 either means the temperature sensor resistance is fairly consistent from one laser head to the next and a fixed value of R19 could be used, or that R19 was hand selected for each laser head.

    Adjustment of Coherent 200

    The Coherent Model 200 stabilized HeNe laser controller and laser head are a matched set and do not automagically mate with each other. There are 3 pots inside the controller. Two of them are likely to need some fine tuning when attaching a laser head that was not originally sold with the controller. Although another laser head may appear to lock, performance with respect to stability and ambient temperature range may compromised. There may also have been some subtle changes in laser head construction over the years, particularly with respect to the temperature sensor. Two late model controllers I acquired would not lock with an older laser head due to the temperature set-point being too low. But aside from component drift in the controller, which should be minimal, there is generally no need to adjust the controller over the life of a laser head.

    Here is the adjustment procedure. A multimeter (preferably an analog VOM, with a needle!) or oscilloscope is required. A 14 pin "DIP Clip" will come in handy, and a laser power meter and temperature probe are desirable but not essential. A hex wrench to set the output polarizer orientation and small flat blade screwdriver to adjust the pots will also be needed.

    This should be done from a cold start at an ambient temperature close to where the laser will typically be used. If the laser had been on, it should be turned off and allowed to cool down for a half hour minimum before proceeding.

    A printout of the Schematic of Coherent Model 200 Stabilized HeNe Laser will come in handy.


    1. Near the output end of the laser head is a slot with a hex screw positioned somewhere along its length. This allows the orientation of the output polarizer and beam sampler to be optimally set to line up the polarization axes of the laser tube. If this is still covered or sealed, there is probably no need to touch it as this will not likely change over the life of the tube. Otherwise, or just to be sure, have the proper hex wrench available to turn this screw.

    2. Remove the cover of the controller. This requires taking out two semi-recessed Philips screws in the rear and sliding the chassis out of the box. On some units, the large main filter capacitor will get stuck on the front bezel and require pressing down the PCB to free it.

    3. Connect the laser HV and control cables, and the power cord.

    4. Locate U1 and U2 (my numbering). U1 is the LM3403 furthest in the rear and U2 is the one in front of it. Also locate the labeled test points (TP-A through TP-G) near the front of the PCB.

    Mode A and B adjustment

    The balance between the two polarized modes will affect the location of the lasing line on the neon gain curve. The following sets the two mode amplitudes to be equal, which places the modes equidistant on either side of the gain curve. However, it should be possible to offset the modes if desired, if a different location or slightly more output power in Mode A (the output beam) is desired. However, it's not possible to place either mode precisely at the top of the gain curve.

    1. Attach the positive probe of the multimeter to TP-D (VC+, the midpoint reference). Attach the negative probe of the multimeter to U1-1. (The polarity doesn't matter for a DMM or scope, but for an analog multimeter, this way around will result in a positive reading.

    2. Turn the laser on. If a laser power meter is available, it can be used to monitor the output. If the Ready indicator comes on initially. wait until the Wait indicator comes on, so the laser is in Preheat mode,

    3. As the laser tube heats up and expands, the modes will sweep through the neon gain curve and the voltage on U1-1 will vary periodically from a minimum to a maximum. The laser power meter will show a similar behavior since this is Mode A which is the same mode that exits as the main beam.

    4. If the hex screw in that slot near the end of the laser head is still covered or sealed in place, this step can be skipped as the orientation of the beam sampler with respect to the tube is unlikely to have changed. Otherwise - or just to be sure - loosen slightly - but don't remove - the hex screw. Slide it back and forth in the slot to maximize the voltage swing of Mode A and tighten the hex screw. The maximum will typically be around 1 V and may occur with the screw positioned anywhere in the slot or at one end. On the one I checked, the sensitivity was actually about 1/2 V per mW of laser output power in Mode A. (Alternatively, maximizing the variation in beam power should result in a similar result.)

      Note: If adjustment of the beam sampler was necessary (or just to double check that it was set correctly), testing with a Scanning Fabry-Perot Interferometer (SFPI) would be desirable. This would allow the undesired Mode B to be virtually totally suppressed. Simply maximizing the Mode A amplitude is not nearly as precise but with care, getting to less than 1 percent of Mode B should be possible. The adjustment using the SFPI can be done at any time, even during warmup, though it's easier once the laser has locked and nothing is changing.

    5. Note the maximum reading of the Mode A voltage.

    6. Move the probe from U1-1 to U1-14.

    7. Adjust the Mode B pot (R14, next to U1), so that the maximum reading is the same as in step 9, above. (Mode B is blocked by the polarizer at the output of the laser head.) Counterclockwise rotation increases gain.

    8. Go back to U1-1 and check that it is still about the same as before. As the tube warms up, the output power will typically increase somewhat. If this is the case, readjust R14 to make Modes A and B voltage equal.

    9. Attach the multimeter between the case or TP-F (signal ground, negative) and U1-7 or TP-A (the Mode Difference Amplifier, positive) to confirm that its output is swinging from near 0 V to over 20 V along with the mode sweep. Unless the laser tube is very weak, this voltage will probably be at each extreme most of the time and will move relatively rapidly between them.

    A CO-200 that is believed to be new/NOS was found to be set so the output was around 10 percent higher than would be produced by this setting. Whether this was intentional is not known, but though slightly unbalanced, it is still safely single longitudinal mode and would provide a bit more output power. Since the absolute optical frequency is not really critical, there should be no harm in increasing the output power by 10 percent (but not much more) using R14 after locking.

    Temperature adjustment

    The HeNe laser tube and ballast resistors dissipate almost 12 W (1.8 kV at 6.5 mA). The temperature set-point must be selected such that it is slightly above what would result from the tube and ballast power alone. At an ambient temperature of 18 °C, the required temperature set-point ends up being around 40 °C, a difference of 22 °C. I do not know exactly how this is affected by a change in ambient temperature. If the difference remains constant, the head must run at 62 °C for the maximum allowable operating temperature of 40 °C (from the specifications in the Coherent manual). Such a high operating temperature seems unrealistic.

    One way to estimate the value for the temperature set-point is to power only the laser HeNe laser tube (not the heater) by disconnecting the Control cable and allow it to reach thermal equilibrium (at least 1 hour). Measure its temperature and then reconnect the Control cable and adjust the Temperature set-point to be about 5 °C higher, or so that the mode sweep goes through an additional 15 full cycles.

    The following assumes an ambient temperature of no more than 25 °C using the default value of temperature found in a new/NOS CO-200 of 48 to 50 °C:

    1. Monitor the temperature of the laser head cylinder. A non-contact IR thermometer is easiest and reads the temperature of the black cylinder fairly accurately. Or, if you have a temperature probe is available, tape it to the center of the cylinder covered by some thermal insulation. If all you have is your calibrated thumb, that will have to do. :)

    2. Watch the mode changes either with a laser power meter or with the multimeter on TP-A or U1-7.

    3. Once they slow down to a full cycle in 30 seconds or so, check the temperature of the laser head near the center, which should be the hottest location. It should be 48 to 50 °C. If it is lower, slowly turn the Temperature pot (R1, next to U2), clockwise to increase the temperature set-point. This will increase the frequency of the mode changes. If it is too high, turn R1 counterclockwise. If this causes the mode changes to cease, and the laser to lock, it may be necessary to shut it off for 5 minutes to allow the tube to cool a bit. The goal is for the laser head temperature to stabilize just slightly above the maximum ambient temperature of the Coherent Model 200 laser specifications. If no temperature probe is available, 48 °C, is quite warm but not too hot to touch and hold.

    4. Give the laser time to conclude that it can stabilize. This will occur once the full mode sweep cycle is more than about 1 minute, at which point the green Ready indicator will come on.

    5. After the green Ready indicator comes on, the laser output power on the laser power meter should quickly settle down and then remain very nearly constant approximately mid-way between the minimum and maximum. If it is not quite where you want it, adjust R14 to put it there. As noted, it can be set up to 10 percent above the mid-point without risk of losing single frequency operation.

    6. The short term fluctuations should be no more than a small fraction of 1 percent. Check the voltage on TP-G, the drive voltage to the heater. It should be relatively constant in the range 10 to 14 V.

      Note that the range of 10 to 14 V is my estimate. The Coherent manual shows a graph with the voltage at 12 V at the time of lock (which would then likely drop down to under 10 V after thermal equilibrium). But there is no description or indication of what ambient temperature was used. Perhaps some key piece of information is missing. While there's no problem adjusting the temperature so the laser locks and is stable at any given ambient temperature or a reasonable range around it like ±5 °C, I don't see any practical way the laser could be set up to operate over the entire 0 to 40 °C range spec'd in the manual without running excessively hot, especially under typical conditions (below 25 °C). It would make more sense if R2 was a sensor for ambient temperature so that the temperature set-point was an offset from ambient rather than actual temperature, but R2 looks like an ordinary resistor.

      However, a reasonable default that will work under most conditions is to adjust the temperature set-point so the voltage on TP-G at the time of lock from a cold start is around 12 V. One turn of R1 appears to change the voltage at lock by about 2 V. (CW increases the temperature and voltage.) So, assuming someone before you hasn't totally messed up the settings, power from a cold start, allow it to lock, check the voltage on TP-G, and adjust R1 by the appropriate number of turns.

      Some considerations:

      • If the heater voltage at lock starts out too close to 0 V, the laser will lose lock after awhile since the need for heater power tends to decline slightly as all parts of the laser heat reach thermal equilibrium and there's then no way to go low enough.

      • If the voltage after an hour is much above 10 V, the temperature set-point is too high and the laser will be running hotter than necessary, but will probably still remain locked.

      • If the voltage continues to fluctuate over a wide range, or remains stuck at 0 V or above 20 V even after Ready has been on for a minute or more, then the laser is unable to lock because the temperature set-point is too low or too high. In this case, the Wait and Ready indicators will probably switch back and forth at a very slow rate (minutes) - forever.

      • If the voltage continues to fluctuate over a wide range but the Ready indicator remains green, the mode balance (set by R14) is way off, or the beam sampler or one or both photodiodes or their pre-amps are faulty and the monostable that detects mode changes is not seeing any.

    7. Assuming the voltage on TP-G is in a reasonable range and the ambient temperature is similar to what it will be during use, monitor TP-G for a couple hours. The voltage on TP-G will likely decline slightly as all parts of the laser head come to thermal equilibrium, but should never be less than 3 or 4 V. If it does go close to 0 V, power off for 5 minutes and power on to reset to Preheat Mode, and turn the Temperature pot clockwise a couple turns to increase the temperature set-point. If the voltage is too high, turn the pot a couple turns counterclockwise to lower the temperature set-point. Then check TP-G after it lock. (If the laser isn't off for long enough, the Ready indicator will come on immediately and it may lock immediately. In that case, leave it off for longer and try again!)

      If the laser will be used in an environment where the ambient temperature is much different than where it was tested, readjustment may be needed. The official Coherent Adjustment Procedure (CAP) probably sets the temperature so high that this would not be required over the full spec'd temperature range of 0 to 40 °C, but that shouldn't be necessary unless the laser is to be used near in a sauna. :)

    Mode feedback gain adjustment

    1. The third pot, Loop Gain (R20, near the front of the PCB), can probably be left alone. The controllers I've checked were all set to 0 ohms. It might make a difference with a very weak tube - or it may just be filling a set of holes! But in the case where your tube is way under 1 mW for the peak output, increasing the Loop Gain might help short term stability. Otherwise, there's probably no reason to touch it unless someone before you twiddled everything.

    Finally, power off for 1/2 hour and confirm that the laser will then stabilize properly (after the warmup period) when powered back on.

    One other thing that's recommended while the case is opened is to check R39 and R40, the third and forth resistors from the right in the first row at the front of the PCB. These are the current limiting resistors for the Wait and Ready indicators, respectively, and were originally 510 and 1K ohms, both apparently 1/4 W (by size and appearance). There are other current limiting resistors inside the indicator packages, but the voltage across R39 and R40 may still be high enough to greatly exceed the 1/4 W ratings of the original resistors. If so, the PCB will probably be darkened beneath them as well. Measure the voltage across R39 and R40 when their respective indicator is lit. If either is more than 20 V and the resistor is only 1/4 W, replacement is highly desirable, especially for R40 which will be stressed possibly for years on end. :) Suitable values are 1K, at least 1/2 W for both. Yes, Ready won't be quite as bright but it will be much happier! Proper replacement will require removing the PCB but this is just five screws and several connectors. Space the new resistors off the PCB a bit to further aid in cooling. The PCB is easily damaged, so use a proper desoldering tool to remove the old resistors and clean up the holes. Or just cut the leads off at the bodies of the old resistors and solder to those.

    Replacement of HeNe Laser Tube in Coherent 200

    Some samples of the CO-200 found on eBay have hard-start but usable tubes (if you're very lucky), but most have tubes that are terminally ill and sputter with very low output power (end-of-life), or are simply dead as a glass bottle. The HeNe laser tube in the CO-200 is a Melles Griot 05-LHR-219, which runs on 2,100 V at 6.5 mA. This is physically and electrically identical (or close enough for government work!) to the common 05-LHR-120 if its modes are well behaved (not all are), a tube with a rated output power of 2 mW, though new ones may approach 4 mW.

    The first step in tube replacement is to find a suitable tube. Melles Griot probably won't even sell you a tube, and if they did, it would cost $300 to $400 or even more! Although a common type, this seems to be harder to find surplus than it would appear. Most of those that turn up on eBay seem to be the 05-LHP-120 - the polarized version - which is useless for this purpose.

    Once a suitable candidate tube has been found, it needs to be tested for non-flipper behavior. A tube that is a flipper may still be useful if the flipping is consistent, or if it disappears when the tube warms up, but a totally well behaved non-flipper is most desirable.

    The following assumes that the tube is to be replaced with the minimal required effort. The heater blanket will remain in place if possible. However, the tube may tend to stick to it and pull it along, which case a bit of maneuvering will be required to free the tube without pulling the heater out so far as to damage the wiring.

    CO-200 laser head disassembly:

    1. Remove the front bezel. This either unscrews (there may be a set-screw locking it in place) or is held on with one to three set-screws. (The front bezel contains either an angled neutral density filter, or a polarizer which can be rotated to adjust the output power.) Set it aside in a safe place.

    2. Place match marks at the joint between the cylinder and beam sample holder to be used during reassembly. This isn't essential but might as well try to get it back together same as it was originally. Gently clamp the beam sampler holder in a padded vice and rock the cylinder back and forth to break the glue bond, then continue rocking to detach the two parts. The front is soft aluminum and can be distorted if clamped too tightly or rocking is too vigerous. A cylindrical form-fitting clamp would be best. Take care not to pull on the photodiode ribbon cable when the front section comes free..

    3. Detach the cathode clip from the tip-off. Note the exact position of the tube within the cylinder so the new tube can be installed the same way. On one I checked it was around 1.7 inches. This distance must be similar upon replacement so that the anode wire is not squished or stretched.

    4. Remove the rear end-cap with the cables by unscrewing the single set-screw (possibly concealed by the yellow safety label) and setting it aside in a safe place. Gently pull the end-cap out of the cylinder. If it doesn't come freely, the anode wire is probably preventing it from moving further out. Pulling harder will probably free the clip from the mirror mount or break the wire (which can be reattached).

    5. Carefully detach the anode clip from the mirror mount (if not already free). Remove the plastic insulating sleave from the anode wire and set aside.

    6. Use a suitable "tool" to push the tube out the front of the cylinder. Take care not to stress the photodiode ribbon cable. Check for a thin metallic sleeve that fits over most of the glass portion of the tube and remove that as well. The heater blanket should remain in place.

    CO-200 laser head reassembly:

    1. If the old tube had a metallic sleave around it, transfer that to the new tube.

    2. Carefully push the replacement tube (with sleeve if present) into the laser head from the front so that it is about 1 inch beyond the location of the old tube. This will provide adequate space at the other end to reinstall the anode wire. Take care not to damage the photodiode ribbon cable or cathode wire - or lose them inside the head! Reattach the cathode clip securely so it can't come loose by accident. Having the U-shaped clip facing the center is usually best. Add a glob of RTV Silicone or Hot Melt glue if desired.

    3. Slip the plastic sleeve saved above back on the anode wire and use a suitable too to push the anode clip onto the anode mirror mount. The easiest and safest technique may be to position it alongside the mirror mount and then use a long stick to press it sideways. With the tube being an inch closer to the back of the cylinder, it should be possible to reattach the anode clip without stressing anything.

    4. While gently pushing the rear end-cap with cabling toward the cylinder so as not to stress the anode clip and possibly dislodge it, pull the tube back toward the output-end of the cylinder so that it is at the same location as the old tube. If it's too slug for Mark I fingers to do this, a suitable tool can be fabricated from a piece of thin sheet metal bent at the end to engage the narrowed section of the mirror mount. This should not affect alignment. Clean the output mirror to remove those unsightly fingerprints. ;-)

    5. At this point, it is probably a good idea to test the HeNe laser tube This can be easily done from the CO-200 controller. There is no interlock so the laser should come on when AC power is applied even without the control connector attached. In fact, it's best to do it that way as any major oops like a HV short will not destroy the controller!

    6. Replace the beam sampler holder by lining up the match marks made above rocking and/or pressing it back into position. Take care that the wires do not block the beam. On some lasers, there is a cylindrical extension on the beam sampler that fits over the mirror mount, but most do not have this feature.

    7. Since no effort was made to align the polarization axes of the tube to the beam sampler, that will be required. Go to Adjustment of Coherent 200.

    I found an old 05-LHR-121 laser head with a good tube, extracted the tube, and spent way too much time installing it in a CO-200 laser head that had a nearly dead tube. The extra time was due to cutting the photodiode ribbon cable and totally removing the heater blanket which wasn't really necessary. The simplified procedure above elimimates those additional steps, not required in most cases. But it enabled me to add six beam alignment screws. I knew that this particular tube was a flipper and expected to simply pick the proper mode polarity such that it would lock on the opposite side of the gain curve from the one that flipped. However, it turned out to only flip until it warms up for about 4 minutes or 113 half-mode cycles, then abruptly it stops flipping and becomes well behaved. I have an Aerotech tube with similar behavior, cause unknown.

    I don't think this is in what might be called original condition, but it does start right up without problems (no hard-start tube!) and has decent power (3.2 mW or more total from the tube). It locks normally with 1.2+ mW in a single mode.

    All in all though, while the effort to do a tube replacement on the CO-200 may actually be less than for the SP-117/A, it is definitely a clunkier operation with the tube just shoved into position and stuff hanging out both ends during the procedure. :( :)

    The controller that went along with this laser head also had minor problems. I had to replace the usual toated dropping resistor for the READY LED but also had to totally rebuild the READY LED assembly itself - both LEDs and their current limiting resistors were fried to a crisp. :( :)

    Strange Fiber-Coupled Interferometer Using CO-200

    This system appears to be NASA surplus, no doubt used for some really exotic purpose which is totally unknown. The only reason it is being described here is that the single frequency HeNe it uses is a Coherent 200, but any similar laser would be suitable.

    The pieces of the system I acquired consist of a CO-200 laser mounted in a semi-enclosed box connected to an interferometer block via am armored single mode fiber-optic cable, and three plastic multi-mode (light pipe) cables for sensing. Based on its design, my assumption is that there is a remote free-space "Tool" or other device with a cube-corner (retroreflector) on it that can move or change in some way.

    The CO-200 appears to be totally standard with a bezel having C-mount threads to which the adjustable fiber port for the input fiber is attached. This fiber may be "polarization maintaining" or simply single mode but there is a polarizer attached to its output end. Since the CO-200 outputs a pure single mode, even If it is not polarization maintaining, its output would still only be a single mode even if not polarization maintaining, but the amplitude would vary.

    The interesting part is the interferometer. See Diagram of Strange Fiber-Coupled Interferometer. The optical components within the solid line are all mounted inside a nicely made solid aluminum block. I do not know if this is a totally custom assembly. From appearances, it may not be. The laser beam enters from the left and is split by the NPBS into a Reference (REF) beam and Measurement (MEAS) beam. The Reference RetroReflector (RR) and NPBS form a standard Michelson interferometer. This is similar to the linear interferometers for systems from HP/Agilent and others eexcept for the beam-splitter being non-polarizating. The return beams from the Reference RR and Remote Tool interfere at the NPBS. The output of the "White Dot" fiber could be used to sense displacement in the normal way, by counting fringes or portions thereof. However, although I'm assuming that the "Remote Tool" simply uses a retro-reflector, that's just speculation. For the polarized "Green Dot" and "Red Dot" outputs to show anything interesting, it would seem that polarization changes are being sensed somehow. (Their polarizers are redundant as the PBS has already separated the polarizations.)

    However, the purpose of the two Wave Plates (WPs) is not entirely clear (as if everything else is!). The WP in the REF path below the Reference RR is made of optical grade mica. Its optical axis is aligned with vertical but does not test as either 1/4 wave or 1/2 wave. It being 1/4 wave would make the most sense so perhaps it's been damaged or is simply imperfect (which wouldn't be surprising with mica which can't really be cleaved to a precise thickness and is often installed in a mount with adjustable tilt to compensate). The WP in the MEAS path is a thin sheet of material (quartz?) sandwiched between glass plates and tests as 1/4 wave. But there is no evidence of it having been set at any specific orientation as there is no obvious alignment flat or mark, or glue residue. It is simply held in place by a retaining strip that is relatively loose and could rotate due to vibrations or even attempts at cleaning. Assuming that both WPs are supposed to be 1/4 wave and set at 45 degrees to the incoming polarization, they would serve to rotate the polarization of any reflected beams by 90 degrees. The input polarizer would then block these from being coupled to the input fiber and re-entering the laser (which can destabilize it). So, the WPs may serve as a poor-man's isolator. Alternatively, the WPs may be used to enhance whatever polarization changes are being sensed by this contraption.

    The Coherent 200 laser in the system I have is in very good health except that the polarization beam splitter at its output (inside the laser head cylinder) had become delaminated and was not passing *any* light! So, the laser appeared dead. But replacing this with one from another CO-200 resulted in like new performance, How does a prism come apart on its own?

    If anyone has more information on this system, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    Tropel Model 100 Single Frequency Stabilized HeNe Laser

    This appears to be the predecessor to the Coherent/Tropel model 200 but probably never went into large scale production as the ratio of 200s to 100s that show up on eBay is at least 10:1. The sample I had has a manufacturing date of 1980 and is badged Tropel for the controller and Coherent for the laser head. The laser head is identical to the model 200 from the outside but uses an ancient soft-seal HeNe laser tube rather than the modern Melles Griot 05-LHR-219. And the controller is much more primitive.

    So this is a bare-bones (but probably perfectly satisfactory) frequency stabilized HeNe laser. Despite the somewhat more complex controller of the Coherent/Topel 200, there is really nothing that is fundamentally different.

    Here are some photos.

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.

    Excel Precision Stabilized HeNe Lasers

    General Overview of Excel Interferometry Lasers

    While Hewlett Packard (HP, now Agilent) and Zygo are most well known for two-frequency HeNe lasers used in metrology, Excel Precision was another manufacturer of systems based on laser interferometry. (For the principles of operation and other information relating to these techniques, see the sections starting with Hewlett-Packard/Agilent/Keysight Stabilized HeNe Lasers.) It's rather doubtful Excel is still in business! The last update on their Web site is dated 1998! Excel was never as well known as HP/Agilent or Zygo depite the claim on their moldy homepage "to be the leading manufacturer of laser interferometers". Their product line was much more limited but did include the same types of system components including their own two-frequency HeNe lasers, interferometer optics, environmental compensators, and processing electronics. Laser Interferometry Tools for Precision Measurement is a summary of their product offerings. Many of the components have direct HP/Agilent counterparts. Unfortunately, beyond this, there isn't much in the way of useful information on their Web site except one page on the 1100B 6DOF Calibration System.

    Excel had only a single type of laser, the 1001, a Zeeman-split HeNe laser with a split/REF frequency between 1.5 and 3.0 MHz. But there are at least 2 different case styles. The 1001A and 1001F are about the same size as the smaller HP/Agilent lasers with a similar mounting arrangement, and have connectors and signals compatible with the 5501B and 5517, respectively. The 1001B is almost as large as the HP/Agilent 5517A (though its shape is more normal) with a similar mounting arrangement, and its connector and signals are compatible. The specifications for the 1001B appear to be the same as those of the 1001F and internally, it's virtually identical to the 1001F but with wasted space at the front. However, an Excel 1001 laser would be a drop-in replacement for an HP/Agilent laser only if selected for split frequency: 1.5 to 2.0 MHz for the 5501B (1001A) and 5517A (1001B), 1.9 to 2.4 MHz for the 5517B (1001F), or 2.4 to 3.0 MHz for the 5517C (1001F). The default beam size for the 1001s is 5 mm, versus 6 mm for the HP/Agilent lasers, but this is probably of little consequence in most applications.

    Although the Excel model 1100B 6DOF Calibration System is based on the 1001 laser technology, it may use only the internal components of the 1001 laser packaged along with additional optics and electronics in a single enclosure.

    Here are the specifications for the Excel 1001A/F lasers (mostly from the file linked above). The difference(s), if any, between the 1001A and 1001F are probably only in the connectors (5501B for the 1001A and 5517A for the 1001F) and F1/F2 orientation (rotated 90 degrees). While I haven't seen full specifications for the 1001B, the connector is 5517-compatible and everything below is probably the same except for the case size and possibly for the nominal wavelength, which is listed as 632.99136 nm on the back of one sample (though I doubt the actual wavelength of the laser is any different!). And what's inside the 1001B is essentially identical to what's inside the 1001F!

    Here are some observations/comments that apply to all the Excel 1001 lasers unless otherwise noted:

    The next three sections have more details on the 1001F, 1001A, and 1001B lasers.

    Excel 1001F Two-Frequency HeNe Laser Head

    The Excel 1001F is physically similar to the small 5517 lasers and the cable connector and electrical requirements are also compatible. However, the spec'd REF frequency for the 1001F is from 1.5 to 3 MHz, so it's not really equivalent to any of them. Perhaps, it was possible to request a particular REF frequency range via an option, rather than a different model laser. Or, a typical laser starts out close to 1.5 MHz and increases with use and normal power decline until 3.0 MHz is considered end-of-life. The nominal wavelength spec is also the same as that of the 5501A/B, 5517A/B, and 5518A - 632.99137 nm. If it is possible to specify a REF frequency range, then the 1001F could be compatible with the 5517B (1.9 to 2.4 MHz) or 5517C (2.4 to 3.0 MHz) except that the nominal wavelength is slightly different for them (632.991372 nm for the 5517B versus 632.991354 nm for the 5517C). But this discrepency of about 28 ppb may have simply been a change in the specifications by HP, and is of no real consequence anyhow. As a practical matter, the exact value of the spec'd wavelength is not important as long as it doesn't change since environmental factors have a much more significant effect than these slight discrepencies and must be factored in during tool calibration.

    The output power of my 1001F was approximately 270µW, which is a bit low compared to the original value of 335 µW) printed on the backplate but well above the spec'd minimum of 200 µW. And the output power doesn't change by more than a few percent after extended warmup indicating that the tube is relatively healthy - no gas contamination and not end-of-life. The laser was still fully functional with a REF frequency of around 2.22 MHz. After aligning the OC mirror, the output power went to 375 µW and REF to 1.98 MHz. So, it's now equivalent to a 5517B. (Specs for the 5517B: Minimum output power of 180 µW, REF frequency of 1.9 to 2.4 MHz.) However, I have no way of knowing what output power and REF were when new. It might have been deliberately detuned to achieve a higher REF at the expense of output power.

    For a summary of the specifications for the 1001F laser, see the previous section.

    Several photos of the 1001F laser can be found in the Laser Equipment Gallery (Version 3.00 or higher) under "Excel Precision HeNe Lasers".

    The two most interesting ones are:

    Excel 1001A Two-Frequency HeNe Laser Head

    The Excel 1001A is in the same case as the 1001F except that it has the 4 pin Power and Reference connectors used by the HP-5501B instead of the single 18 pin connector of the 1001F and 5517s. And while it is essentially a functional clone of the 5501B, the spec'd REF frequency range of 1.5 to 3.0 MHz exceeds its upper limit of 2.0 MHz. The interior is essentially identical to that of the 1001F. Refer to the previous section for photos, as well as the Laser Equipment Gallery (Version 4.04 or higher) under "Excel Precision HeNe Lasers".

    The unit I acquired had "junk" scribbled in Magic Marker on the case, but appears to work quite well with an output power of over 410 µW (label value is 540 µW) and REF frequency of 1.85 MHz - perfect for a 5501B clone. As with the other Excel lasers, warmup isn't as rapid as with the HP/Agilent lasers, and is constant heating followed by abrupt locking. Interestingly, the date on the backplate - which appears to be original - is 2006. This is rather peculiar given that the Excel Web site hasn't been updated since 1998 and no recent references to the Excel company can be found. :) However, the date code on a tantalum capacitor on the Control PCB - the only component with a date code visible - was 1996, which makes more sense. So perhaps it was serviced by a former Excel engineer in 2006.

    Excel 1001B Two-Frequency HeNe Laser Head

    The Excel 1001B has a mounting arrangement similar to that of the 5517A laser with the feet in the same locations, though the case itself isn't quite as large and is a more normal rectangular box rather than the funky looking trapezoidal solid. :) The cable connector and electrical requirements are also compatible. However, the spec'd REF frequency for the 1001B is from 1.5 to 3 MHz, so it's not really equivalent to a 5517A. Perhaps it was possible to request a particular REF frequency range via an option, rather than a different model laser. Or, a typical laser starts out close to 1.5 MHz and increases with use and normal power decline until 3.0 MHz is considered end-of-life. (But then life expectancy when used in a system where the upper limit for REF is 2.0 MHz would be shorter.) The nominal wavelength listed on the back of a 1001B differs from that of the 5517A - 632.99136 nm versus 632.99137 nm - but that is almost certainly rounding error. :)

    The output power of my 1001B was approximately 120 µW after full warmup, actually declining from 150 µW just after locking. This is low compared to the original value of 240 µW) printed on the backplate and the spec'd minimum of 200 µW. Other than the low output power, the laser was still fully functional with a REF frequency of around 2.60 MHz. After aligning the OC mirror, the output power increased to 315 µW and REF declined to 1.88 MHz after full warmup. The output power changed only slightly, from 330 µW just after locking. This indicates either a change in alignment or drift in the electronics, not a tired tube. So, it's equivalent to a 5517A. (Specs for the 5517A: Minimum output power of 180 µW, REF frequency of 1.5 to 2.0 MHz.) However, I have no way of knowing what output power and REF were when new. With the output power now being so much higher than what's on the label, it might have been deliberately detuned to achieve a higher REF at the expense of output power. But that doesn't make sense if it was intended as a 5517A clone.

    For a summary of the specifications for the 1001B laser, see the previous previous section.

    Several photos of the 1001B laser can be found in the Laser Equipment Gallery (Version 3.19 or higher) under "Excel Precision HeNe Lasers".

    Other than the difference in case size, everything else is virtually identical compared to the 1001F laser.

    Excel Power and Reference Cable Pinout

    The Excel Power and Reference Cable for the 1001A is similar to the HP/Agilent cable for the 5501A and 5501B lasers.

    Excel 1001A power connector

    See HP 5501A and 5501B Reference and Power Rear Panel Connectors for pin location.

       Pin      Function                              Socket View
        A       +15 VDC input                          D o  o A
        B       -15 VDC input
        C*      +5 VDC output (test-point)             C o  o B
        D       Power ground

    Excel 1001A reference signal connector

    See HP 5501 Reference and Power Rear Panel Connectors for pin location.

       Pin      Function                              Socket View
     ---------------------------------------------         A
        A*      Accessory +15 VDC fused                    o
        B*      +15 VDC return                         D o   o B
        C       Reference (difference) frequency           o
        D       Complement of pin C                        C

    Pins denoted by "*" have these assignments on the HP-5501A/B but they have not been confirmed for the Excel 1001A. Pin B (-15 VDC) is a no-connect in the 1001A.

    Excel 1001B/F Power and Reference Signal connector

    The Excel Power and Reference Cable for the 1001B/F is similar to the HP/Agilent 10791 and the two may be used interchangeably. (This is probably Excel part number 1059A but that hasn't been confirmed.) "Wire Color" is that of the power connections with ring lugs. See HP/Agilent 5517 Laser Rear Panel Connector for the physical arrangement of the pins:

       Pin   Color    Function
        A*            NC ( MEAS signal level on 5508A)
        B*            NC (~MEAS on 5518A only)
        C*            NC ( MEAS   "      "
        D*            NC (Signal Return for MEAS)
        E             ~REF (Zeeman beat signal from internal optical
        F              REF  receiver's differential line driver)
        G    Black    Ground
        H    Green    Ground
        J    Orange   +15 VDC
        K    Red      +15 VDC
        L    White    NC (-15 VDC on HP/Agilent cable)
        M             +15 VDC
       N,P*           NC (Cable Shield on HP/Agilent cable)
        R             Signal Return for REF
        S             Ground (to 4 pin BNC)
        T             +15 VDC (to 4 pin BNC)
        U*            NC (Cable Shield on HP/Agilent cable)

    * Connections to pins A,B,C,D,N,P,U are not present on the 1001B/F cable. The wire for Pin L (-15 VDC) is present, so this cable should work with an HP/Agilent 5517 laser. There is also a blue wire in the cable but it is cut off and hidden under heat-shrink and does not have continuity to the connector.

    Excel 1031A and 1031F Optical Receivers

    These are free-space and fiber-coupled optical receivers that are physically and functionally similar to the HP/Agilent 10780A/B/C and 10780F/U (except for the type of fiber. The bandwidth is not known and may be lower than their HP/Agilent counterparts since the split frequency range of Excel lasers generally not as high.

    There are a couple benefits to the 1031 over the 10780: (1) the Signal LED is duplicated on the front so it is more readily visible during alignment and (2) on the 1031F, the fiber connector insert can be removed using a flat blade screwdriver, and the entire fiber connector housing can be removed by loosening the twin set-screws in the metal frame and unscrewing it. These make the 1031F more convenient to use with a free-space beam, though a linear polarizer at 45 degrees will still be required. In addition, the photodiode area is larger than than one in the 10780F/U, so without a lens, it should be more sensitive with larger beams.

    CAUTION: The 1031s have five tantalum "gum drop" capacitors inside, some of which may go short-circuit on the first application of power or after a few power cycles. This was discovered to the detrement of several fuses when testing a batch of 1031Fs with manufacturing dates around 2005. There are two size caps, typically 10 uF at 25 V and either 1 or 5 uF at 25 V. At the very least, the 10 uF and 1 or 5 uF (depending on the specific sample) caps directly across the +15 VDC input should be replaced with aluminum electrolytic capacitors of approximately 22 uF and 5 uF at 25 V. But while you're at it, replace the others as well. Even going by the most basic rules for tantalum cap usage, the voltage derating is not sufficient - usually quoted as 50 percent. Using 22 uF for all would probably be OK also. Aluminum electrolytic capacitors are readily available that are small enough to fit without problems. So between age and voltage, these are time bombs waiting to detonate. Some do split in half and smell bad when that happens. With a higher current power supply and/or larger fuse, they may indeed explode.

    Unfortunately, while getting inside is straightforward requiring only a small Philips screwdriver and single-edge razor blade, it does end up ruining the beautiful appearance of these devices. After removing 8 screws, the plastic strips on each side must be peeled off which damages the labels underneath. And what remains of them must be slit end-to-end to allow the metal shells to be separated. :( Oh well, one can't have everything. ;-)

    Excel Part Number Reference

    Most of this information is from Excel Laser Interferometry Tools for Precision Measurement.

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.

    Frazier Stabilized HeNe Lasers

    Frazier 100 Iodine Stabilized HeNe Laser

    Frazier Instruments is (or was) a manufacturer of a variety of products, most of which have nothing directly to do with lasers. (Look at their various products.) Thus, the Model 100 ISHL is kind of strange being stuck in there with instruments mostly used for materials testing. It's also not clear if Frazier really exists now (Fall, 2012). Their Web site appears to have been put together by untrained monkeys with generally poor formatting and obvious errors that have not been corrected in eons. And the latest entry on their Frazier News Page is from 2009. There are no dates on any of the other pages that I checked. Even if Frazier is still in business, it's likely that the ISHL is only there for historical purposes and it would not actually be possible to order one. Or perhaps they have an inventory of Model 100 lasers and will be happy to sell them inexpensively. I'm tempted to ask. :)

    Having said all that, the Model 100 iodine stabilized laser consists of a massive laser head and separate controller using custom modules built into a semi-antique :) Tektronix 5000-series storage scope mainframe. This probably dates the development of this system to the 1970s or early 1980s. While not fully automatic, the controller provides a straightforward way of selecting one of 7 possible I2 absorption peaks and then locking to it. It's based very closely on the original NIST ISHL design first presented in the paper: Howard P. Layer, "A Portable Iodine Stabilized Helium-Neon Laser, "IEEE Trans. on Inst. and Meas, IM-29, pp358-361, 1980. A photo of the same or very similar laser can be found at NIST: Length - Evolution from Measurement Standard to a Fundamental Constant. The slightly shorter resonator shown in the paper uses a different two-Brewster HeNe laser tube than in the two samples I'ave seen (manufacturer unknown) and the scope plug-ins are labeled "National Bureau of Standards" instead of "Frazier Precision Instrument", but everything else appears to be identical. The Frazier controller is in a rack-mount configuration, but it can be converted to the portable upright instrument shown in the paper with a screw-driver and open-end wrench. :)

    Here are specifications for the Model 100 Iodine Stabilized Laser (mostly from the Frazier Web site with interpretation):

    That really low power output of 100 µW was bothering me as it seemed as though much more was possible even with the extra I2 cell Brewster windows from a tube capable of at least 1 mW. But the reason is far more fundamental than unavoidable losses. The problem is that for a resonator length of around 350 mm (about 14 inches), 3 to 4 longitudinal modes will oscillate unless something were done to force single mode operation. The original NIST ISHL appears to have do this with a really low OC reflectivity of 93 percent if the paper is to be believed, which raises the lasing threshold but also dramatically decreases output power. There would appear to be no fundamental reason why a normal 99% OC could not be used with the addition of an intra-cavity temperature-controlled etalon to select a single longitudinal mode. This might add some complexity to the controller though. However, the resonator I tested that appears to be identical the one in the NIST photo actually has the following physical specifications:

    So it appears as though this design does use detuning of the cavity alignment to force single longitudinal mode and miniscule output power. In tests using a PMS/REO tunable HeNe laser detuning of the cavity alignment could easily be set to force single longitudinal mode operation, with a similar drop in output power.

    Other ISHLs have similar anemic specifications for output power. Even those claiming to use an internal mirror laser tube, presumably with external iodine cell, seem to have a power output only slightly higher.

    Where greater power is desired, an "offset locked" dual mode stabilized single frequency HeNe laser is generally added to the system. The difference frequency between the two lasers is phase-locked to a crystal reference by tuning the dual mode stabilized laser, thus retaining essentially the same frequency stability as the I2 laser, but the output power can be 1 to 2 mW as with common stabilized HeNe lasers.

    Iodine Stabilized HeNe Laser Head is a photo of what is almost certainly a Frazier 100. Although there is no manufacturer label, everything is identical down to the pattern of ventilation holes in the cover. The overall appearance is unremarkable with a shutter at the front (the round black thing) and several cables coming out the back (hidden). Leveling "feet" would often be installed be installed in the cast tabs for precise alignment. Interestingly, there was an Agilent inventory sticker on the cover, so perhaps this very laser was used to certify HP/Agilent metrology lasers like the 5517A! :) In fact, the base of this laser bears a striking resemblence to the 5517A (though the dimensions don't match). It's a combination of a cast and machined assembly, clearly not made for a one-time research project.

    Iodine Stabilized HeNe Laser Head With Cover Removed shows the interior. The glow of the Melles Griot 05-LHB-290 two-Brewster HeNe laser tube can be seen along with the iodine cell within the massive 4 bar Invar resonator structure.

    Frazier Model 100 Iodine Stabilized HeNe Laser System shows the laser head on top of the controller. At present it doesn't have a working tube (though as can be seen, the tube does light up), and except for the HeNe laser power supply, the connectors on my old laser head do not mate with the controller.

    Frazier Model 100 Iodine Stabilized HeNe Laser Controller shows the front panel. The connectors on the controller are labeled as follows:

    The three modules are all discrete circuitry with mostly Burr-Brown parts.

    While the Model 100 controller doesn't have fancy auto-magical locking firmware, operation seems relatively straightforward. (This was before the era of cheap silicon!) Paraphrasing from the paper, the front panels of the scope plug-ins have a diagram showing the hyperfine I2 components available at 633 nm along with their precise vacuum wavelengths. In sweep mode, the third derivative of the laser output power is displayed on the (storage) scope screen as the cavity length - and thus laser wavelength - is scanned with a period of 1 second. The desired component can then be centered and expanded using the Sweep (span) and Bias (offset) controls, at which point the system is switched to lock mode. The complete operation summary is printed on the three plug-ins. Who needs the @!%$# manual? ;-)

    I wonder how many of these were ever made. My controller appears to be SN 15 and had a DoD (Department of Defense) inventory sticker. It's in mint condition.

    General info on iodine stabilized HeNe lasers along with additional photos can be found in the section: Iodine Stabilized HeNe Lasers.

  • Back to Commercial Stabilized HeNe Lasers Sub-Table of Contents.

    Hewlett-Packard/Agilent/Keysight Stabilized HeNe Lasers

    General Overview of HP Interferometry Lasers

    Hewlett-Packard (HP) developed and manufactured a variety of stabilized two frequency HeNe lasers for use in metrology applications including very precise measurement of displacement (relative position), velocity, straightness, and angle using the wavelength of light as the reference "yardstick". The first of these appeared around 1970 but the basic technology has remained unchagned to the present day. Around the year 1999 when HP split off the test and measurement operation to become Agilent, these lasers and associated equipment went with it. Agilent has introduced some higher performance laser models and implemented a number of enhancements to the technology, though most of them are functionally similar to those developed in the late 1980s, which are themselves based on the 1970 technology. Now, as of 2014, there has been another reorganization resulting in the new name: Keysight Technologies. What exactly they will do, if anything, besides changing the name and Web site is unknown at the present time. Since all this was developed by HP and most of the current technology is functionally (and often physically and electrically) identical, HP is the designation we'll continue to use most often. However, Agilent or HP/Agilent may be used as well especially if the particular laser or other equipment model was not available from HP. At least until Keysight starts changing stuff. :)

    While several other companies have competing product lines, all indications are that the vast majority of HeNe metrology lasers, as well as the associated optics and electronics in the explored Universe, have been made by HP or Agilent. This statistic is confirmed by some very reliable scientific evidence from multiple well funded research studies which have reached the same conclusion: Most of the metrology lasers appearing on eBay are from HP or Agilent! :-)

    The general approach to precision measurement used by all systems based on two-frequency HeNe lasers such as those from HP/Agilent is shown in Interferometer Using Two Frequency HeNe Laser. The "Signal Processing" in its simplest form is just a pair of long counters, each of which increments at the REF and MEAS difference frequency. Their difference is then the relative position or displacement. These algorithms may be implemented in dedicated digital hardware, a fast microprocessor, or combination of the two. Additional processing using a combination of hardware and software is generally used to improve the resolution to way below one wavelength of the laser light. The capabilities of these systems are quite impressive. A typical example is the HP-5501A Laser Interferometry Measurement System, which enables a position/distance resolution down to better than 10 nm (that's nanometer as in 0.000000001 meter!). And that's one of the earliest implementations. More information on interferometers based on two frequency lasers including descriptions of the optical components can be found in the section: Interferometers Using Two Frequency Lasers. What follows relates mainly to the laser technology.

    Here is a comparison of most of the basic HP two frequency metrology laser models. However, there are countless variations which are often just small tweaks in REF frequency or case style for specific OEM customers:

     (10,11)         (5,6)  Reference   Maximum    Beam
      Model  Case   Tuning  Frequency   Velocity   Diam.  Comments
      5500A  Huge :)  PZT  1.5-2.0 MHz  0.4  m/s   6 mm   (1)
      5500B   "   "    "      "     "    "    "    6 mm   (1)
      5500C   "   "    "      "     "    "    "   6,9 mm  (2)
      5501A  Small     "      "     "    "    "    "  "   (3)
      5501B   " "   Thermal   "     "    "    "    "  "   (3)
      5517A  Large     "      "     "    "    "    6 mm
      5517B  Small     "   1.9-2.4 MHz  0.5  m/s    "
      5517BL  " "      "      "     "    "    "     "
      5517C   " "      "   2.4-3.0 Mhz 0.711 m/s 6,3,9 mm
      5517D   " "      "   3.4-4.0 MHz  1.0  m/s  6,9 mm
      5517DL  " "      "      >4.4 MHz  1.3  m/s   "  "
      5517E   " "      "      >5.8 MHz  1.77 m/s   6 mm   (8,9)
      5517EL  " "      "       "    "    "   m/s   "  "   (8,9)
      5517F   " "      "      >7.0 MHz  2.15 m/s  6,9 mm  (8,9)
      5517FL  " "      "       "    "    "   m/s   "  "   (8,9)
      5517G   " "      "      >7.2 MHz  2.2  m/s   9 mm   (8,9)
      5517GL  " "      "       "    "    "   m/s   "  "   (8,9)
      Z4214A  " "      "   1.9-2.4 MHz  0.5  m/s    "     Same as 5517B
      5518A  Large     "   1.5-2.1 MHz  0.4  m/s   6 mm   S/N below 2532A02139 (4)
       " "    " "      "   1.7-2.4 MHz 0.453 m/s    "     S/N 2532A02139, above (4)
      5519A   " "      "   2.4-3.0 MHz  0.7  m/s    "     (4)
      5519B   " "      "   3.4-4.0 MHz  1.0  m/s    "     (4)
      N1211A XtrHuge   "     15-17 MHz  4.0  m/s  6,9 mm  Fiber AOM Laser (12)

    As of Summer 2014, only the 5517A, 5517B, 5517BL, 5517C, 5517CL, 5517D, 5517DL, 5517EL, 5517FL, 5517GL, 5519A/B, and N1211A are listed on the Keysight Web site as still being in general production and "orderable". There are also variations such as higher power or higher REF/split frequency for the above lasers depending on options. In most cases, the only distinction between a laser like the 5517D and the 5517DL "low heat" version is in the cover and how cooling is provided - via forced air exhausted to outside the Tool it's in rather than by convection. The major electrical, optical, and functional specifications are identical. In the general descriptions that follow, the "L" may be left off.

    The Z4214A is functionally and physically indistinguishable from a standard 5517B except for the label. It is probably for an OEM customer and is not documented anywhere except here


    1. The 5500A (or one identical to it with no model number) was the original HP metrology laser. The 5500B seems to be physically and functionally similar to the 5500A with the same tube but perhaps upgraded electronics. These are the only HP lasers with a built in (linear) interferometer so that the only required external optical component was a retro-reflector (cube-corner) on the moving tool.

    2. The 5500C is the only laser that has a set of 2 optical receivers for an external return beam built-in. Each one uses a pair of photodiodes on either side of the beam, which may be selected to be horizontal or vertical by the orientation of the front shutter plate depending on the interferometer setup. This is part of the reason for the 5500C's huge case. (They all have an internal optical receiver for monitoring the beam to confirm that the beat frequency is present and that there is adequate output power.)

    3. The 5501B is a drop-in functional replacement for the 5501A. However, since the 5501B uses thermal rather than PZT tuning, the time from power on to lock is typically 5-9 minutes instead of a few seconds. And the 5501A has an additional "Diagnostic" connector, not present on the 5501B, so if associated equipment depended on signals from the laser, substitutes would have to be provided - somehow. Going the other way - replacing a 5501B with a 5501A will also work subject to one requirement: After several hours, especially from a cold start, the 5501A may need a "Retune" cycle, normally initiated by a pushbutton on the rear panel or an external signal.

    4. The 5518A and 5519A/B have a single optical receiver built-in. And of all the HP metrology lasers, the 5519s are unique is having a built-in DC power supply so they simply plug into the wall and feed their REF and MEAS signals to the associated measurement processor/display (usually PC-based).

      Like the 5500C, the 5518A or 5519A/B can be used in the normal way (e.g., in a 5528A Laser Measurement System), but are generally intended to be set up stand-alone without any additional optical receivers in a 5529A or 5530 Dynamic Calibrator). For example, the 5519A laser head can be mounted on a cart or tripod and aimed through interferometer optics at a cube-corner (retro-reflector) or plane mirror on a tool whose motion needs to be measured precisely.

    5. The stabilization in all of these lasers is done by balancing the amplitude of the two orthogonally polarized frequency components, F1 and F2.

      • The 5500A/B/C and 5501A use a PZT to control cavity length so locking is achieved within about 10 seconds. They use tube assemblies that are totally different than those in all the other HP/Agilent lasers including the 5501B. The original patent for the 5500A/B laser tube is: U.S. Patent #3,771,066: Gas Laser and the substantially identical "Divisional Application", now U.S. Patent #3,889,207: Frequency Stabilized Gas Laser. The tubes used in the 5500C and 5501A are the same but differ from those in the 5500A/B. Specifically, the 5500A/B tube passes the waste beam out the back to be used for stabilization feedback with the PZT connection to a ring electrode surrounding it while the tube in the 5500C and 5501A blocks the waste beam and samples part of the main beam for feedback. In addition, the waveplates are separate for the 5500A/B and part of the tube assembly in the 5501A and 5500C.

        For the 5500A/B, a motor (yes motor) rotates a waveplate and polarizer in the waste beam to generate the locking signals. The 5500C and 5501A (and all subsequent lasers) samples a portion of the output beam and uses a Polarizing Beam-Splitter (PBS) to provide the locking signals corresponding to the amplitude of the F1/F2 frequency components. However, unique to the 5501A (and possibly the 5500C), the PBS is deliberately oriented so that the separation isn't perfect and a small amount of both F1 and F2 are present in each. This results in a beat frequency being generated which is used to produce the reference signal (REF) and to confirm that there is enough beam power to be usable. The other lasers sample another portion of the output beam and use a separate photodiode behind a polarizer and detector circuit for this purpose.

      • All the other lasers use a heater inside the laser tube (thermal tuning) to control cavity length so locking typically requires several minutes. Lasers with the "L" suffix may have a vented cover. (Some may spec this as an option like "CO5" rather than an "L".) Though my 5517E also has a vented cover so maybe it's really a 5517EL even if it's labeled 5517E! The beam height for the newer "L" versions also differs slightly - 80.8 mm instead of 79.5 mm. The height difference appears to be a result of the tube being mounted on a sheet metal plenum of sorts along with mica washers. All this is done to minimize heat transfer by conduction from the case - it should all go via the air stream. However, older "L" versions simply had the vented cover with no internal changes so I suppose they weren't quite as effective at this. Can you spell "kludge"? :)

        Although locking typically occurs in around 4 minutes (READY comes on solid) for most thermally-tuned lasers, some special versions may require 20 minutes. My 5517E takes about 9 or 10 minutes. But a fully stable frequency output requires 90 minutes for lasers with a non-vented cover or a vented cover but no fan. Those with a vented cover and a fan require only 45 minutes. (It's not known if the temperature set-point for the tube is lower for these compared to the non-vented variety. But if so, this would explain the "Low heat" option since power dissipation would be reduced by running at a lower temperature. However, I've never noticed any obvious differences between the warmup of the "L" and non-"L" lasers, and measurements of the temperature set-points have not found anything conclusive. And there is no difference in the temperature setpoint instructions.)

        From my observations, the REF frequency oscillates slightly immediately after locking with a period of order of minutes. The amplitude of these oscillations gradually decreases with time and eventually becomes very small. However, the laser likely still meets accuracy specifications during this time.

        An internal optical receiver samples part of the output beam and is used to generate the references signal and to confirm that there is enough beam power to be usable.

    6. All of the lasers that use thermal tuning except for the 5501B have virtually identical electronics, though the Control PCBs (called the "A3 Controller/Reference Board" in the HP/Agilent manuals) are not physically interchangeable between lasers in the small and large cases mostly due to mounting configurations. For most versions of the 5517s in the small case, the Control PCBs are functionally equivalent and may be swapped between lasers, requiring only the single temperature set-point adjustment. Newer versions (by date) of the 5517s, have had at least two major redesigns of the Control PCBs (reason unknown). These appear to be much more complex and use FPGA/FPLDs and/or DSPs in place of simple logic. The original through-hole A3 board was used with only trivial changes from the early 1980s until around 2003. This is designated the "Type I Control PCB". The one that is used in most lasers since around the end of 2003 is essentially an emulation of the this board using an FPGA and mostly surface mount parts. It is called the "Type II Control PCB". The four LEDs on the outside are the same but there are a few more for status on the PCB itself. The functions of the jumpers and even the temperature test-points and set-point pot are similar, though their locations differ slightly. In particular, the "Htr" and "Photo Diode" jumper blocks are swapped. If one accidently changes the wrong one during setup, they may think the PCB is bad. Been there, done that. :( There may also be a slightly different version of this design for some late model lasers like the 5517FL, possibly incorporating a change to the feedback loop time constant in firmware. But a much more complex Control PCB (though not the latest) only found on a small percentage of high-REF lasers like some versions of the 5517D/E/FL/GL has changed dramatically. This is called the "Type III Control PCB". It uses a SHARK DSP and Lattice FPGAs on both sides of the PCB. Its benefits are not obvious, but possibly doesn't require the temperature setpoint adjustment as it figures it out on its own. ;-)

      The 5501B is the only laser to use Pulse Width Modulation (PWM) rather than pure analog to drive the heater inside the laser tube. This was probably done to reduce power dissipation in the electronics to maintain plug compatibility with the 5501A, but does result in modulation of the optical frequency by the PWM.

    7. The usable maximum velocity will depend on the reference frequency (REF) of the specific laser. The values given are the worst case for the minimum of the specification for REF in the direction that reduces the MEAS Dopplar frequency. The approximate maximum velocity for a given laser in meters/second can be calculated using the equation: Vmax=(REF-100kHz)*3.16E-7. This is affected slightly by the measurement electronics. The maximum velocity is cut in half with the plane mirror interferometer since it doubles the Dopplar frequency shift (and doubles resolution). (Note that the value of REF really doesn't limit the velocity in the opposite direction, though the electronics will impose restrictions depending on optical receiver bandwidth and data processing implementation.)

    8. The HeNe laser tubes in ALL HP/Agilent lasers are of totally custom and uniquely interesting construction and even the earliest 5500A - developed in the late 1960s - was hard-sealed. This is done by enclosing the resonator and mirrors fully inside the glass envelope, with a window for the output beam and glass-metal seals for the electrodes. So shelf life was never a problem - with very rare exceptions, these lasers could sit idle for 25 years without degradation in tube performance. The lifetime of 50,000 hours was also amazingly long compared to other HeNe lasers back then - and longer than even most modern HeNe lasers. (Though anecdotal evidence suggests that tubes manufactured in recent years may not have the same fantastic life.)

      • The tube assemblies in the 5500A/B included the glass laser tube, magnet, and beam expander/collimator. (But unlike all the others that followed, these do NOT include the waveplates.) The 5500A/B employ PZT tuning which among other things meant that no warmup time was required - the lasers were ready to be used in about 10 seconds from a cold start. They are also unique among all HP/Agilent lasers in using the waste beam out the back of the tube for feedback.

        A diagram is shown in Internal Structure of Hewlett Packard 5500A/B Laser Tube Assembly.

      • The tube assemblies in the 5500C and 5501A are generally similar to those of the 5500A/B and also employ PZT tuning. They include the glass laser tube and magnet, beam expander/collimator, and waveplates. They are physically similar if not identical to each-other, though there were at least two possible beam diameters, so the collimators could differ. But since feedback is via a beam-sampler at the front of the tube, the waste beam is no longer needed and is actually blocked.

        A diagram is shown in Internal Structure of Hewlett Packard 5500C and 5501A Laser Tube Assemblies and a photo of an intact one in HP-5501A Laser Tube Assembly. The naked tube is shown in HP-5501A Laser Tube Removed From Magnet and Output Optics Assembly. And Major Components of HP-5501A HeNe Laser Tube for most of what's inside.

        See the sections starting with: Notes on the HP-5500 Two Frequency HeNe Laser for more details.

      • The tube assemblies in the HP-5501B and HP/Agilent-5517B/C/D including the tube and magnet, beam expander/collimator, and waveplates, are physically similar, though of course the reference frequencies differ. And the specific collimators will depend on the beam diameter option. The structure on these are all nicely machined aluminum except for the cast base metal base. These all have an internal resistance heater for tuning and require a several minute warmup period. The heater is a solid metal wire bifilar-wound on the "mirror spacing rod". Its composition is not known but may be iron or an alloy with a similar Temperature Coefficient of Resistance (TCR). It is NOT Nichrome whose TCR is too low by more than an order of magnitude.

        The magnets in the 5501B and all 5517s have the same dimensions: OD=2 inches, ID=1.5 inches, L=4 inches, fine ground outside and rough cast inside. However, their field strength varies significantly as will be discussed in depth later. But the magnet in the not very common N1211A laser (at least the half dozen or so I've disassembled) has a slightly larger ID of around 1.54 inches, probably because it is fine ground inside from the original rough cast version, reason unknown. It's unlikely that This can have any impact on performance and no one (except) me actually gets to admire the interior. ;-)

        A diagram is shown in Internal Structure of Hewlett Packard 5517B/C/D Laser Tube Assemblies and a photo of an intact one in Tube Assembly Used in HP-5517B/C/D Two-Frequency HeNe Lasers, and after a similar one is taken apart in Major Components of HP/Agilent 5517B/C/D Laser Tube Assembly. This also applies to newer HP-5501Bs. Though functionally identical, the tube assemblies in older HP-5501Bs and 5517Bs may differ very slightly. A diagram is shown in Internal Structure of Hewlett Packard 5501B Laser Tube Assembly, an intact one in Tube Assembly Used in Original HP-5501B Two-Frequency HeNe Lasers and disassembled in Major Components of Original HP-5501B Laser Tube Assembly. The differences are primarily in the beam expander, the supporting structure for the output optics, and the use of a segmented magnet instead of a single piece magnet. (However, these can show up in older 5517Bs as well.) There are at least two other minor differences Can you locate them?)

        The tube assemblies in the 5517A, 5518A, and 5519A/B lasers appear quite different, being of cast base metal and also much larger as shown in Tube Assembly Used in HP-5517A, 5518A, and 5519A/B Two-Frequency HeNe Lasers and disassembled in Major Components of HP-5517A, 5518A, and 5519A/B Tube Assembly. But the actual glass tube, magnet, waveplates, and optics themselves are similar to those in the other lasers as shown in Internal Structure of Newer Hewlett Packard 5517A, 5518A, and 5519A/B Laser Tube Assemblies. Early versions of the 5517A and 5518A tube assemblies had heat sink fins rising about 1/2 inch above the top plate which poked through the outer case. These disappeared quite early with no other apparent changes. A photo in a 1983 HP Journal article even showed a tube assembly with no fins but a cover with the required hole for the fins! Older 5517As and 5518As also had tubes and magnets similar to those in the diagram of the 5501B. See Internal Structure of Older Hewlett Packard 5517A and 5518A Laser Tube Assemblies. And unlike all tube assemblies for any of these lasers to follow, the first versions had a beam expander that was glued into a tapered hole with no adjustment possible. See Tube Assembly Used in Original HP-5517A and 5518A Two-Frequency HeNe Lasers and Internal Structure of Original Hewlett Packard 5517A, 5518A, and 5519A/B Laser Tube Assemblies.

      • The tubes in Agilent 5517E/FL/GLs (and some late model 5517DLs) are slightly shorter and have other minor physical differences. The shorter tube is needed to achieve the high REF of the 5517E/F/G, and helps for the DL. As of 2014, all Agilent lasers may use versions of the Short tube. See Internal Structure of Agilent 5517E/F/G Laser Tube Assemblies and a photo of an intact one in Tube Assembly Used in Agilent 5517E/F/G Two-Frequency HeNe Lasers. Apparently, the front portion of later ones is longer to accomodate the 9 mm beam (if used) which for these Short tubes is much longer than the 3 or 6 mm beam expanders as shown in Tube Assembly Used in Agilent 5517 Lasers with Short Tubes and 9 mm Beam Expander. So the overall length ends up being similar to that of the older tube assemblies. After being taken apart, the result is shown in Major Components of Agilent 5517E/F/G Laser Tube Assembly. The 6 mm version would have a beam expander and waveplates the same size as in the Long-HV tube lasers.

        The prototype for the 5517E, one of which I acquired ;-), used a Long-HV tube modified with a 100 mm mirror spacing rod. I was suspicious that achieving 5517E specifications would be possible at all with the normal Long-HV tube with the normal 126 mm or 127 mm mirror spacing rod. Testing with a Scanning Fabry-Perot Interferometer (SFPI) revealed that the mode spacing was around 1.5 GHz and NOT the 1.190 or 1.180 GHz of the Long-HV tube. And indeed, it even had "100 mm" hand-printed on the magnet, which I noticed only after the SFPI test when the tube assembly was removed from the chassis for cleaning and detailing. :) The back-end is unchanged with a single backing disk between the HR mirror and spring. There appear to be additional backing disks (or a spacer cylinder) between the front spring and OC mirror to make up the difference in length. The heater needed to be shorter so its resistance is also lower - around 6.1 ohms versus 8 ohms. But it is otherwise physically identical to a Long-HV tube. Since I doubt more than a handful of these were ever built - the backplate serial number is: "proto_000006" - and this could still be the only one of its kind, there is no detailed diagram, sorry. ;) Just imagine a shorter mirror spacing rod with shorter heater winding and a bunch of additional backing disks between the OC mirror and spring in Internal Structure of Hewlett Packard 5517B/C/D Laser Tube Assemblies. The laser also has the strange Type III Control PCB. It locks at around 40 µW (WOW!!) and 6.0 MHz. It is not known whether that low power is all that was ever possible, or simply the end result of a life test. This may have simply been a failed experiment. However, someone may have pointed out that an optimal redesign of the tube could provide a boost in output power and higher REF frequencies. The Short tube may have been the result.

        As of 2019, there are only 5 known FSRs for thermally-tuned laser tubes:

        • Long-LV, 132.5 mm: FSR = 1.10 GHz. Older 5517A, 5501B.
        • Long-HV, 127 mm: FSR = 1.18 GHz. Newer 5517A and 5501B, 5517B/C/D.
        • Long-HV, 126 mm: FSR = 1.19 GHz. Possibly all after around 2000.
        • Long-HV, 100 mm: FSR = 1.50 GHz. 5517E prototype.
        • Short, 102 mm: FSR = 1.48 GHz. All Short tube lasers.

        However, from "Deep Throat" evidence :), there are a few additional types of mirror spacing rods for long tubes, usually associated with high-REF lasers. But based on tests, these appear to have the same FSR as those in the list, above, so the differences may be related to the location of the discharge escape holes or bore diameter to maximize gain and output power.

        Types of HP/Agilent Thermally Tuned HeNe Laser Tubes shows diagrams of the three general varieties of glasswork. X-ray Views of Typical Long-LV (5501B), Long-HV (5517C), and Short (5517D) HP/Agilent HeNe Laser Tubes are what one would see assuming they have health insurance. :) (X-ray courtesy of James Sweet.) Can you diagnose the injury to the top tube? :) The designations Long-LV and Long-HV refer to slight differences in the tube operating voltage of some really old 5501B (and 5517B) tubes being lower than newer ones. Much much more on this below and in the sections starting with: Notes on the HP/Agilent 5517 Two Frequency HeNe Laser.

        Up until sometime after 2009, most HP/Agilent lasers used the Long-LV or Long-HV tubes, which are structurally identical with only minor changes in dimensions. These (as well as all other models before them) had mirror alignment fixed by the precision grinding of the ends of a glass "mirror spacing rod", which along with the cavity mirrors floats between a pair of springs. So aside from the entire mirror spacing rod shifting position from a physical shock (look at the X-ray), there was nothing inside that could become slightly out of whack (though dropping a laser from 25 feet onto concrete would probably pulverize the tube). However, the Short tubes found in some high REF lasers starting in the early 2000s, and probably all models after around 2014, have the HR mirror on a post that permits alignment to be adjusted even after being installed in the laser - and also for it to drift with time and use from thermal cycles or just metal creep. In some ways this is a step backwards since the alignment may need to be checked and corrected after awhile (though from tests of multiple late model lasers, this is at most a minor issue). But it was probably required due to changes to the cavity geometry of the Short tubes to maximize available output power (long-radius hemispherical versus near-hemispherical for all that preceeded it), which also makes alignment much more critical. The OC (front) mirror is mounted on a cage that permits alignment to be fine tuned during manufacture before the tube is sealed. However, alignment of the curved OC is somewhat less critical and that cage should be quite stable.

      There's no way to tell the version (e.g., 5517C) or reference frequency (e.g., 2.3 MHz) of the tube itself by inspection of the assembly or from its label. They don't have that information explicitly, only a part number. In the list below, the tube for the 5501B and 5517A/B is a Long-LV or Long-HV type unless otherwise noted:

      • 5500A: ???.
      • 5500B: ???.
      • 5500C: 05500-60033.
      • 5501A: 05501-60006.
      • 5501B: 05501-60102, 05501-69202 (Agilent, ???).
      • 5517A: 05517-60301, 05517-60501.
      • 5517B: 05517-60201 (???), 05517-68201 (6 mm), 05517-69201 (???), 05517-38201 (short tube)
      • 5517C: 05517-68218 (3 mm?), 05517-68217 (6 or 9 mm?), 05517-68240 (6 mm), 05517-68249 (3 mm?).
      • 5517D: 05517-68224, 05517-68232 (C15), 05517-68234 (C29, high REF?).
      • 5517DL: 05517-68255 (6 mm, Short tube), 05517-68256 (9 mm).
      • 5517E prototype: 5080-0183 (9 mm Long-HV tube with 100 mm mirror spacing rod).
      • 5517FL: 05517-68253 (Short tube).
      • 5518A: 05518-60301.
      • 5519A: 05519-60301.
      • N1211A: Z4203-60207 (~1 mm).

      • 5517??: 05517-68252, 05517-68263 (9 mm, short tube).

      • Z4214A: 05517-68201 (same tube as standard 5517B).

      The difference in tube part numbers for same model lasers isn't entirely clear. It may be a combination of the size of the beam optics and other special features like a particularly high REF frequency or high output power option. And to make things even more confusing, the following is information extracted from the Agilent Web site "Parts" availability for the 5517A/B/C/CL/D/DL in 2014. My interpretation is that "Long" tube PNs always begin with a "6" and "Short" tube PNs begin with a "3". In all cases, Long tubes were either listed as obsolete or available only until supplies are exhausted. There is also some question as to what these part number really apply to. The tube assemblies always seem to have PNs beginning with "68" regardless of the type of tube, Long or Short. And parameters like the beam size and to some extent, REF frequency/velocity are also determined by components in the tube assembly other than the glass tube itself (beam expander, magnet field strength, etc.). So, the PNs below must encompass those somehow even though they don't match what one sees in the actual laser! And, no, you can't plunk down $8K or so and buy a replacement tube. Agilent will only provide them if the laser is sent in for repair. I also don't know if they'd repair a laser purchased on eBay or at a garage sale. :)


      • 05517-60501: Long tube for 5517A.
      • 05517-33301: Short tube for 5517A.


      • 05517-33203: Short tube for 5517B, Option C14 (3 mm, >300 uW).
      • 05517-33202: Short tube for 5517BL (low heat laser head).
      • 05517-33202: Short tube for 5517B-C15 (low heat leak laser head).
      • 05517-33248: Short tube for 5517BL, Option H10 (high power).


      • 05517-33218: Short tube for 5517C, Option 003 (3 mm).
      • 05517-60218: Long tube for 5517C, Option 003 (3 mm).
      • 05517-33217: Short tube for 5517C, Options C01, 009, 031.
      • 05517-60217: Long tube for 5517C, Option 009 (9 mm).
      • 05517-30217: Short tube for 5517C, Option 009 (9 mm).
      • 05517-33204: Short tube for 5517CL, Option 003 (3 mm).
      • 05517-33205: Short tube for 5517CL, Option 006 (6 mm).
      • 05517-33206: Short tube for 5517CL, Option 009 (9 mm).
      • 05517-33219: Short tube for 5517C, (Standard) or with (Option N05).


      • 05517-60224: Long tube for 5517D, Option C01.
      • 05517-30224: Short tube for 5517D, Option C01.
      • 05517-60220: Long tube for 5517D.
      • 05517-30220: Short tube for 5517D.
      • 05517-60229: Long tube for 5517D, Options C06, C13, C22
      • 05517-30229: Short tube for 5517D, Options C06, C13, C22.
      • 05517-33215: Short tube for 5517D, Option C05 (3 mm, Low heat).
      • 05517-60230: Long tube for 5517D, Option C07 (low heat, >4.0 MHz).
      • 05517-33224: Short tube for 5517D, Option C19 (9 mm, low heat).
      • 05517-33263: Short tube for 5517D, Option C30 (HS, SBT).
      • 05517-33255: Short tube for 5517DL, Option 006 (>4.4 MHz).
      • 05517-33256: Short tube for 5517DL, Option 009 (>4.4 MHz).
      • 05517-33261: Short tube for 5517DL, Option 300 (>4.4 MHz).
      • 05517-33229: Short tube for 5517D, Option N06 (Low heat).

      While the 5517EL/FL/GL lasers are also listed with specifications and availability, it's not possible to order parts for them, not even screws

      And there are a few parts listed for the 5519A/B, but not tubes.

      As noted, the older 5501A and 5500C use physically identical tubes with PZT tuning. The tubes in the 5500A and 5500B are the same and functionally similar to those in the 5501A and 5500C, but the construction differs enough to make it impractical to substitute for those. None of these tubes are compatible with any of the other lasers. The chart below shows the Physical (P) and Reference frequency (R) compatibility of the most common thermally-tuned HP/Agilent lasers. (This should also apply to the low power versions designated with an "L" following the model number (e.g., 5517DL), which differ primarily in the case configuration (often only the skins). However, special options may complicate compatibility.

                      (g)                         (a) (b)
                   5   5   5   5   5   5   5   5   5   5   5   5
                   5   5   5   5   5   5   5   5   5   5   5   5
                   0   1   1   1   1   1   1   1   1   1   1   1
                   1   7   7   7   7   7   7   7   8   8   9   9
            (c,d)  B   A   B   C   D   E   F   G   A   A   A   B
        (e) 5501B  PR   R  P   P   P   P*  P*  P*   R
        (g) 5517A   R  PR                          PR  P   P   P
            5517B  P       PR  P   P   P*  P*  P*       R
            5517C  P       P   PR  P   P*  P*  P*           R
            5517D  P       P   P   PR  P*  P*  P*               R
            5517E  P*      P*  P*  P*  PR  P   P
            5517F  P*      P*  P*  P*  P   PR  P
            5517G  P*      P*  P*  P*  P   P   PR
        (a) 5518A   R  PR                          PR  P   P   P
        (b) 5518A      P    R                      P   PR  P   P
            5519A      P        R                  P   P   PR  P
            5519B      P            R              P   P   P   PR


      1. 5518A S/N below 2532A02139.

      2. 5518A S/N 2532A02139 and above.

      3. Where both "P" and "R" are present in the matrix, it should be possible to install the tube and achieve original laser performance specifications. Where only "P" is present, the tube may be installed but the performance specifications will be those associated with the replacement tube. In all cases, it is assumed that the tube assembly is transferred in its entirety since the beam expander may need to be matched to the tube type (Long-LV, Long-HV, or Short), the waveplates need to be the large version for a 9 mm beam, and the waveplate settings must be optimized for the specific tube. If transferring a tube assembly with a 9 mm beam into a chassis which previously had a 3 or 6 mm beam, the beam sampler may also need to be swapped. Also, if swapping a high-REF laser tube like the 5517F for a normal one like the 5517D, the Control PCB may also need to be swapped, and the temperature set-point will need to be adjusted appropriately for the tube type (Long or Short) in all cases.

        The "P*" (physically compatible with an asterisk) means that beam height specifications for the 5517E/F/G has changed slightly so shims may be need to be added (or transferred) to be fully compatible if installing a 5517E/F/G tube in a 5501B or 5517B/C/D case. Since the 5517E/F/G lasers appear to come with shim washers under the tube assembly feet that are easily lost and difficult to install, it may make more sense to transfer the other parts like the Control PCB into its case rather than the tube assembly into another case. If the HeNe laser power supply brick under the tube assembly is found to be faulty, the Control PCB and entire sheet metal shroud can be removed to replace it without disturbing the tube assembly. This may also apply to selected 5517DLs and other special lasers. But as long as the platform on which the laser is mounted has some height adjustment range, it probably doesn't matter.

      4. The metal casting of older 5517A tube assemblies may require minor trimming if being installed in 5519A/B lasers to provide clearance for the internal switchmode power supply.

      5. If installing a 5517 tube in a 5501B laser or vice-versa, the waveplates must be adjusted to rotate the output polarization 90 degrees which swap the axes of the F1 and F2 modes because they are interchanged in the 5517 compared to the 5501 (reason unknown). The laser will not even lock without doing this since the slopes of the split gain curves at the lock point would have the wrong sign.

      6. It is not totally known whether 5517E/F/G tubes may be swapped with other 5517 tubes, or among themselves. However, theory and experience suggests that there is no problem as long as the entire tube assembly is moved intact. (Really old Short tubes come with a series resistor to make their heater resistance similar to the others which must be kept with the tube but newer Short tubes have the same higher resistance as the Long tubes.) And there may be subtle differences in the Control PCBs used for normal and high-REF tubes like the 5517E/F/G as well as some 5517Ds. So a Control PCB from a 5517B laser may not lock reliably or at all with a 5517FL tube and there may be no proper REF signal if it does. While the 5517E/F/G case is the same size as the others, the actual tube assembly on some versions is slightly shorter. This doesn't really matter, but the raw beam characteristics (diameter and divergence) differ, so beam expanders CANNOT be swapped between Long and Short tubes. And unlike the beam expanders for Long tubes, those for Short tubes may differ in length depending on the beam diameter (3, 6, or 9 mm). This implies that the collimating lens focal length depends on beam diameter but the same expanding lens is used. Since the raw beam from a Short tube is much narrower, it may be that the focal length of the expanding lens would have needed to be too small to use the old design. So, in fact, a tube assembly with a Short tube and 9 mm optics may be similar in length to the Long tube version. A sample is shown in Tube Assembly Used in Agilent 5517 Lasers with Short Tubes and 9 mm Beam Expander. (All newer tube assemblies using Short tubes appear to be this length to be able to accomodate any size beam expander.)

      7. The 5517A (as well as the other large-body lasers (5518A and 5519A/B) being physically incompatible with the small-body lasers makes installing tubes from the latter more challenging. But with a suitable mount for the small tube assembly and brackets to secure the Control PCB, it is possible for 3 and 6 mm beam diameter lasers. (Part of a 9 mm beam may not get through the beam sampler or output aperture/turret.) The waveplates would require adjustment for any of these. So for example, the tube from a 5517C can be used to refurbish a 5519A. Even the tube from an N1211A laser could be installed with an appropriate beam expander using a similar technique with its REF frequency adjusted for a 5517A. Going the other way is not something I'd want to contemplate. :)

    9. The 5517E/F/G are not very common and used only in special and/or high performance systems. They have much higher minimum REF frequencies than all other HP/Agilent/Keysight lasers (5.8 to 7.2 MHz) except the super complex and costly N1211A. (See the section: Notes on the Agilent N1211A Fiber AOM Laser Head.) But as a consequence of the higher REF, their output power tends to be much lower, with a minimum spec of only 65 µW for most models. See the section: Agilent 5517E/F/G.

    10. Minimum output power with common options that only affect output power:

      • 5500A: 120 µW.
      • 5500B: 120 µW.
      • 5500C: 120 µW.
      • 5501A: 120 µW.
      • 5501B: 180 µW.
      • 5517A: 180 µW.
      • 5517B: 120 µW. (Some documentation lists 180 µW.) Option H10 is 400 µW and 2.2 to 2.4 MHz REF. There is/was also an option H07 which may also be 400 µW (as with the 5517C).
      • 5517BL: 120 µW.
      • 5517C: 180 µW. Option H03 is 300 µW, option H05 is 240 µW, and option H07 is 400 µW.
      • 5517D: 180 µW. Option H01 is 300 µW, option H03 is ??? µW.
      • 5517DL: 180 µW. (120 µW minimum after 3 years.) Option H01 is 300 µW.
      • 5517E: 65 µW.
      • 5517EL: 65 µW. (50 µW minimum after 3 years.)
      • 5517F: ??? µW
      • 5517FL: 65 µW. (50 µW minimum after 3 years.)
      • 5517G: ??? µW.
      • 5517GL: 168 µW. (120 µW minimum after 3 years.)
      • 5518A: 180 µW.
      • 5519A: 180 µW.
      • 5519B: 180 µW.
      • N1211A: 275 µW.
      • Z4214A: 180 µW.

      It is not known whether the 5517F and 5517G (without the "L") ever existed.

      Maximum output power for all these lasers is listed as 1 mW. The output power when new is generally lower for higher-REF lasers. So, the N1211A has the highest which may be very close to 1 mW and the 5517EL/FL has the lowest which is often less than 100 µW. The exception would be the 5517GL, which may be the same as a 5517FL optimized (or selected) for "high" power.

      Specifying a lower output power after 3 years seems to be a recent change by Agilent. Perhaps they were getting too many warranty returns due to low power. :)

    11. Additional options. Many of these are not documented anywhere. In fact, even with a laser in-hand having some of these options, it's difficult or impossible to tell what they mean! Sometimes they are for a special high power or slightly higher REF frequency. For the most part, if a given model laser (e.g., 5517C) has the desired beam diameter and acceptable output power, it will work in most tools calling for that model laser. Or, they may be a combination of other options. Only in rare instances will there be issues with something like a higher than spec'd REF frequency:

      • 5500A: ???
      • 5500B: ???
      • 5500C: ???
      • 5501A: ???
      • 5501B: ???
      • 5517A: 003, 006.
      • 5517B: 003, 006, 009, 030.
      • 5517BL: ???
      • 5517C: 003, 006, 009, 030, C01, C05, C07, N05, X01, X02.
      • 5517D:006, 009, C01, C05, C06, C07, C11, C13, C15, C15, C16, C19, C29, C30, C39, N06, N07, N13.
      • 5517DL: 006, 009, 038, 039, 300.
      • 5517E: C01.
      • 5517EL: 003, 006, 009, 030, 039.
      • 5517F: ???
      • 5517FL: 003, 006, 009, 030, 039, 300.
      • 5517G: ???
      • 5517GL: 006, 009, 030, 039.
      • 5518A: ???
      • 5519A: ???
      • 5519B: ???
      • N1211A: 002.
      • Z4214A: ???.


      • 003: 3 mm beam.
      • 006: 6 mm beam.
      • 009: 9 mm beam.
      • 030: Electrically insulated feet.
      • 038: High REF of 4.68 MHz (5517DL).
      • 039: Standard mounting.
      • A6J: ANSI/NCSL Z540-1-1994 compliant calibration.
      • C01: Long range (5519).
      • C29: High REF greater than 5.0 MHz (5517D, possibly reduced power).
      • C30: High REF greater than 5.1 MHz (5517D, possibly reduced power).
      • C39: 9 mm beam and 300 µW output (5517D).
      • STD: Standard performance.
      • X01: Combination of 006 and STD.
      • X02: Combination of 003 and STD.

      If anyone has additional info defining what these other options mean, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    12. The N1211A "Fiber AOM Laser" uses a HeNe laser tube assembly similar to those in the other thermally tuned lasers optimized for high power at the expense of the split frequency. But it adds a pair of Acousto-Optic Modulators (thus, "AOM") to shift the optical frequencies to achieve a much higher REF frequency of up to 17 MHz, and a fiber-coupled Remote Optical Combiner (1212A or 1212B depending on final beam diameter) to provide more flexibility in placement of this giant size laser assembly. There is much more in the section: Notes on the Agilent N1211A Fiber AOM Laser Head. These may also be designated Z4203 with various suffices depending on OEM and split frequency range.

    There are photos of nearly all models of these metrology lasers in the Laser Equipment Gallery under "Hewlett Packard/Agilent/Keysight HeNe Lasers". These range from the original 5500A to the 5517FL (which is similar to the 5517GL), and N1211A/Z4203. The most significant difference between the various lasers is in the Zeeman split or reference frequency (REF). A higher REF frequency enables a faster slew rate for displacement and velocity measurements. As of Winter, 2019, all the 5517s, 5519s, and Z4203s are current Agilent products. General information, descriptions, and specifications may be found by going to Keysight Technologies and searching for "laser positioning laser heads" or a specific model number like "5517C". Some of the specifications from the datasheet:

    These sound quite incredible but 1 ppm is a frequency of about 474 MHz (1/1,000,000 of 474 THz, the optical frequency corresponding to a wavelength of 633 nm). Thus 0.1 ppm is 47.4 MHz, 0.02 ppm is 9.5 MHz and 0.002 ppm is 0.95 MHz. So, still impressive, but quite reasonable for a well designed stabilized HeNe laser. However, what is somewhat unique about the 5517 and some of the other HP/Agilent lasers is that this absolute accuracy is achieved without the need for any periodic testing or adjustments, by virtue of the design of the mode sampling and locking electronics. Therefore, reasonably healthy samples of any of these lasers make excellent optical frequency/wavelength references.

    Various other versions of these lasers also exist which may have variations in REF/split frequency, beam diameter, mounting, and other specifications.

    There is also an Agilent N1211A Fiber AOM Laser Head. This uses a pair of Acousto-Optic Modulators (AOMs) to generate a much higher difference frequency - from 7.5 MHz to 17 MHz depending on version - than is possible with a Zeeman HeNe laser alone. The N1211A uses a more or less standard 5517 laser tube that has been designed and optimized for high power with a relatively low split frequency, similar to the tube assemblies described above. A polarizing beam-splitter separates the two components which are then passed through a pair of AOMs which shift them by appropriate amounts to achieve the desired difference frequency. They are then sent via a pair of polarization maintaining fiber-optic cables to an N1212A or N1212B "Remote Optical Combiner" which generates a free-space beam with a diameter of 6 mm or 9 mm, respectively. The tube assembly in the N1211A resembles the one used in the "small" 5517 lasers (but with a different part number), happily locks using a 5517 Control PCB, has a beam approximately 1 mm in diameter, a REF frequency similar to that of a 5517A or perhaps even lower, works in a normal interferometer setup, and has a mounting arrangement that is just incompatible enough to be difficult to adapt to a small laser body (though it can be done). But the vast majority of HP/Agilent lasers are standard products.

    More info may be found in the section: Notes on the Agilent N1211A Fiber AOM Laser Head. Everything else below deals only with the normal HP/Agilent Zeeman-split HeNe lasers.

    With respect to selecting among the various laser models, if your application has no need for the higher REF frequency (often called the split frequency), there is no advantage to getting a laser like a 5517D as opposed to a 5517B. In fact, the lasers with a lower REF frequency tend to have higher output power and thus may be easier to set up and align especially in multiple-axis configurations. They also tend to be less expensive on the surplus market, though the Agilent price isn't that much different. The only disadvantage of a laser with higher output power is that there can be enough of a detected MEAS signal due to slight angular misalignment of interferometer optics like the 10706A to result in a reading even if the beam to the tool or whatever whose position or velocity is to be measured is blocked or misaligned. The interferometer cube contains a polarizing beamsplitter and if the F1 and F2 orientation are not precisely aligned with the polarizer, there will be a small amount of F1 mixed with F2 and vice-versa even without the reflection from the mirror on tool. With a 400 µW laser and single axis, the required angular accuracy to avoid a false MEAS signal is well under 1 degree with the optical receiver threshold at its default most sensitive setting. And even if the alignment is perfect, polarizing beam splitter and AR coatings are not perfect so there can still be residual mixing. None of this matters once the return beam is aligned since the MEAS signal will be much stronger than the bogus one, but it can be confusing. Increasing the threshold should eliminate the issue.

    And a note about that impressive spec'd lifetime of 50,000 hours - about 6-1/4 years of continuous use. HP lasers used to last a long long time and it wasn't unusual to find an HP laser running fine after 8 years. But I rather suspect this is no longer the case. I've seen many late model (2004 to 2006) Agilent 5517s that were going down hill well before 6 years including at least one that was essentially dead after less than 3 years. These were standard 5517Bs or 5517Cs pulled from semiconductor fabs, either because they failed in normal use, or because they were rejected during preventive maintenance due to low power or the REF frequency going out of spec (which is usually related to the power decline). Thus, even a late manufacturing date is no longer assurance of a healthy laser. Nor would even a close inspection of the HeNe laser tube, as the they appear identical except for the Agilent label - perhaps that's enough! So if you are buying these things new, it probably pays to go for the extended warranty if that is an option. :)

    HP/Agilent Laser Principles of Operation

    All the HP/Agilent laser heads use what is known as an "axial Zeeman-split HeNe laser". These produce a single beam with two orthogonally polarized components differing in optical frequency by between 1.5 and 4 MHz for most models. (Orthogonal here means that the two components are perpendicular to each other and has only one of the two frequencies present.) The beam is single spatial mode (TEM00) though the outer tail may be clipped by the beam expander. The two polarized components are actually a pair of lasing lines operating on the same longitudinal mode of the laser cavity but slightly shifted due to the Zeeman effect resulting from the laser tube operating inside an axial magnetic field. However, it's not strictly accurate to think of the longitudinal mode oscillating on its own and then somehow being split into two sub-modes by the magnetic field. Rather, it is the Doppler-broadened neon gain curve itself that splits and the two halves are actually separated by an amount proportional to the magnetic field, typically 100s of Mhz. Lasing on one of these gain curves will result in right-hand circular polarization and lasing on the other will result in leff-hand circular polarization. Then, any given longitudinal mode will see some amount of gain from each of the split gain curves, and if above threshold, will lase there. But due to mode pulling from its respective gain curve, the actual location of the lasing line will be shifted slightly in opposite directions, and that is what gives rise to the separation resulting in the Zeeman split frequency. This also means that the location where the two Zeeman split modes are equal in amplitude is NOT at the peak of the neon gain curve but part way down as shown in Axial Zeeman Split HeNe Laser Mode Behavior, a simplified diagram like one probably found in a textbook. (More below.) A lasing mode will exist where a cavity mode coincides with a location on the Zeeman-split and shifted neon gain that has a gain above the lasing threshold. While the cavity may be long enough for more than one longitudinal mode to be present (as in the diagram), only when the a longitudinal mode is approximately centered on the Zeeman-split neon gain curve will it be able to create a Zeeman-split mode pair and beat signal in the frequency range that matters (a few MHz rather than the many 100s of MHz between different longitudinal modes).

    Note that in this diagram and the others depicting Zeeman-split mode behavior, the magnitude of both normal mode pulling and the mode pulling that produces the split frequency is greatly exaggerated. For example, even for a 5517D with a split/REF frequency of 3.4 to 4.0 MHz, the spacing between F1 and F2 would be only about 0.3% of the longitudinal mode spacing of 1.2 GHz! Without the plots showing F1 and F2 spaced more than 2 orders of magnitude further apart than they really are, they would merge into one line on any reasonable size diagram!

    Waveplates at the output of the HeNe laser tube convert the left and right-hand circularly polarized Zeeman split modes to linearly polarized modes that are orthogonal and aligned with the horizontal and vertical axes of the laser. These two modes usually differ in optical frequency by between 1.5 and 4 Mhz (depending on the specific laser). (Some recent versions of the 5517 may actually go to 7 MHz or more.) The X and Y polarizations are sent down different paths in the metrology application. One is generally a reference length and the other is the dimension to be measured or tracked. (It's the change in path length difference that matters so they could both move if desired.) Rather than creating an interference pattern that changes slowly, the two beams are combined together in a detector that outputs a difference (or heterodyne) signal. If the relative distance between the two beam paths changes by one half wavelength of the laser (about 632.8 nm but accurate to many significant digits!), the phase of the difference signal will change by 360 degrees. The laser also generates an electrical signal from beating the signals together internally. This constant reference is compared to the detector signal and an electronics package measaures the phase shift continuously and uses it to determine the distance traveled.

    A moderately powerful cylindrical permanent magnet does the Zeeman splitting resulting in a pair of circularly polarized outputs at two very slightly different frequencies. The field strength inside the magnet near the center ranges from less than 200 gauss (G) to over 400 G depending on the model laser and type of tube. (For any given tube design, the REF frequency is roughly proportional to field strength up to a point.) F1 is designated the lower frequency and F2 is designated the higher frequency. For the 5501A/B, F1 is vertical (perpendicular to the laser base) while F2 is horizontal (parallel to the laser base). For the 5517A/B/C/D/E/F/G, 5518A, and 5519A/B, F1 is horizontal (parallel to the laser base) while F2 is vertical (perpendicular to the laser base). (Exactly why HP switched orientations between the two model series is not clear as there is no benefit to one over the other and it just causes confusion, or perhaps that was the intent!) The difference between F1 and F2 ranges from 1.5 to 4 MHz for most of the HP/Agilent lasers depending on the model (as listed above) and also the specific sample of the laser. The cavity length of the early HP lasers (5500A/B/C and 5501A) is PZT-controlled using feedback based on the amplitude of the two modes. For the 5501B and all later lasers like the various 5517s, cavity length is adjusted by a heating coil wrapped bifilar-style around the bore inside the tube. In all cases, the feedback is used to maintain the position of the lasing lines symmetric on the Zeeman split neon gain curves as shown in Axial Zeeman Split HeNe Laser Mode Behavior. A Quarter WavePlate (QWP) converts the circular polarized output to orthogonal horizontal and vertical polarized components which are used externally. F1 is reflected from whatever is being measured or tested (e.g., disk drive servo writer or wafer stepper) and F2 is reflected from a fixed reference. The difference frequencies (F1-F2) and (F1-F2)+dF1 are then analyzed to determine precise position, velocity, or whatever. The end result is identical in terms of sensitivity to position changes compared to the common single frequency (or homodyne) interferometer, but the two frequency approach has lower noise and greater stability, and is therefore potentially more accurate.

    Interestingly, the actual beat or reference frequency (REF) does NOT need to be super stable over the long term. Rather, it is the difference between REF and the return (MEAS) signals that matters and that only depends on the motion of the target reflector, the optical frequency of the meausrement beam, and the speed of light. Thus, although the optical frequency needs to be known to high precision (±0.1 ppm for the standard lasers; ±0.02 ppm for those calibrated to MIL STD-45662), the exact beat frequency of each laser is not precisely controlled or even precisely measured and recorded or used anywhere in the calculations. This is one reason why the listings above include only a range of values. Any given sample will operate somewhere within that range during its expected life, but the exact value is somewhat random depending on the specific characteristics of the tube/magnet assembly and the specific place on the neon gain curve that the lasing line is parked. In fact, REF tends to increase over the life of the laser as the gain and thus output power decline with use. As long as REF remains within the spec'd range for the particular model laser, then the system in which it is installed will work properly. What exactly a machine will do if REF goes out of range is implementation dependent. But one reason for a laser such as this to be replaced is for REF to approach or exceed the high end of the spec'd frequency range, though in many cases, a REF which greatly exceeds the spec'd upper limit can be tolerated.

    While one might think that locking the difference frequency to a crystal reference would be superior - and the technique is patented - it's not clear that this would be better and might actually be worse. The difference frequency relative to the mode position can change for any number of reasons. In fact, the REF frequency of HP/Agilent lasers tends to slowly vary by a small amount (typically a fraction of 1 percent) even after locking with the period of the cycle increasing as the tube assembly approaches thermal equilibrium. The cause is probably back-reflections from internal optics and the resulting etalon effect. Despite this, because the amplitude of the two modes is forced to be equal to keep the modes centered on the split neon gain curves, the optical frequency ends up being very stable.

    All of the HP lasers use conventional dual polarization mode stabilization to lock the lasing lines to the split neon gain curve. However, the two signals are not from adjacent longitudinal modes as with most common laboratory stabilized HeNe lasers, but are the two Zeeman split sub-modes differing in frequency by a few MHz instead of many 100s of MHz. In fact, both are the same cavity mode but shifted slightly higher and lower than would be predicted by c/2*L. One twist on the implementation is that the 5501B and all later lasers (those below it on the chart) use a Liquid Crystal Device (LCD) polarization selector to alternately sample the horizontal and vertical polarized modes. The LCD consists of a large area single pixel LCD (!!) with a linear polarizer bonded to it. Applying a voltage to the LCD rotates the polarization of the sampled beam by 90 degrees prior to it passing through the polarizer. This sensed output is fed to a subtracting ample-and-hold to compare the amplitudes of the two polarized components in the error amp driving the heater. This is radically different than the polarizing beamsplitter and dual photodiodes used in most other dual polarization mode stabilized lasers including the 5500C and 5501A. The LCD approach does have a sort of elegance as well as practical benefits: Since the same optical path and photodiode are used for both polarization modes, the sensitivity is identical, so the mode balance should be perfect (assuming the LCD polarization rotation is 90 degrees). Since the intent is to park the modes symmetrically on the split neon gain curve, this is ideal and thus requires no offset adjustment over the life of the laser as the output power of the tube declines. And, the LCD and associated electronics may in fact be cheaper than a high quality polarizing beam splitter. However, it also creates some artifacts as a result of the digital switching, resulting in small cyclical variations in optical frequency over a period of 2 or 3 seconds. These are of no consequence for most metrology applications, but do detract from the elegance of these lasers.

    In fact, the thermally tuned lasers have only one adjustment associated with stabilization, and that is for the temperature setpoint at which the controller switches from pre-heating to optical locking. The resistance change of the actual heater coil is used to sense temperature and there are variations from one tube to the next. But this is an extremely non-critical setting and won't affect accuracy, only possibly the temperature range over which the laser will remain locked. (5517s with the Type III PCB have no hardware adjustments, though it's possible that the firmware provides access via RS232. More below.)

    One oddity with respect to the thermally tuned laser tubes is the patent reference that appears on the label of all newer ones at least: "Licensed by Patlex Corporation Under Patent No, 4,704,583". The title of this patent is: "Light Amplifiers Employing Collisions to Produce a Population Inversion", filed in 1977 but not granted until November of 1987. The most curious thing is that there appears to be very little of relevance in the patent other than its association with laser action! Nothing in the patent diagrams or text has any obvious connection to the tube assembly design. In fact, the exact same text exists on other more mundane things like a Carl Zeiss-badged Siemens LGK 7634, a bog standard 2 mW random polarized HeNe laser head. I've heard that Patlex is actually a bunch of lawyers and I bet they made out or are making out quite well. :)

    Higher Order Spatial Modes HP/Agilent Lasers

    In geniune HP/Agilent lasers, I had assumed that conditions are such that when locked (READY on solid), only the single set of Zeeman-split modes can oscillate (with no "rogue" modes). Rogue modes can potentially result in problems in the interferometer. This is shown for the general case in Avoiding Unwanted "Rogue" Modes in Zeeman-Split HeNe Lasers. Careful design of the laser tube cavity length (which determines the FSR) must balance output power (longer is better) and avoiding the production of rogue modes (shorter is better). In addition, the magnetic field must be chosen to produce the desired shift of the neon gain curves without being so large as to permit rogue modes to oscillate on the tails, as well as decreasing output power below the spec'd (or useful) value by reducing the height of the gain curve at the center of their overlap. By using cavity length control to force the mode position to be centered, the actual optical wavelength/frequency can be set very accurately. Using the MHz range beat signals makes signal processing straightforward and is more immune to noise than baseband optical signals.

    However, in 2013, I obtained a supposedly new/NOS Agilent 5517C which did have very low level rogue modes after locking, approximately equal in amplitude on either side of the main Zeeman-split mode. Their polarizations were aligned with the principle axes. While the amplitude of each was less than 1 percent of the total power, this does seem to be "impure" for Agilent! ;-) Initially, I assumed them to be rogue longitudinal modes. But based on their position on the SFPI display, they would be around 0.5 GHz distant from the main modes on the right and 0.7 GHz from them on the left. While the distance between them is consistent with the longitudinal mode spacing of the 5517C tube of 1.180 GHz (or close enough for government work!), their position is inconsistent with TEM00 longitudinal modes. Thus, they are probably a pair of longitudinal modes of a higher order spatial mode, which would appear with a mode spacing indistinguishable from the TEM00 longitudinal modes, but a shift in position on the SFPI display, as shown in Rogue Spatial Modes in Agilent 5517 Laser. As can be seen, their amplitude is not very large, unless it is compared to zero! And there is no evidence of rogue TEM00 longitudinal modes, which if present, would be offset from the Zeeman-split modes by 1.18 GHz. Also unlike TEM00 rogue modes, their amplitude is never more than a few percent of that of the main modes during mode sweep. So, it's not like they don't quite disappear when the laser is locked, they are simply there. As possible further confirmation, when sampled with a 1 mm aperture, they seem to become smaller relative to the main mode, or even disappear entirely. But as with rogue longitudinal modes, they also disappear when shims are added to reduce the magnetic field and REF frequency, though more of a reduction was needed than expected for their size-down well into the 5517B range. The cosmic significance of this, if any is as yet unclear. :) But based on the behavior of these modelets, there is no real possibility that they are an SFPI artifact, and another 5517C with similar output power tests 100 percent pure SLM.

    As a practical matter, even if a true "defect", this would be unlikely to result in any performance issues since neither mode will be Zeeman-split and any anomolous beat signal(s) would be way outside the measurement pass-band. No one else outside of Agilent would likely ever notice the anomaly! And most likely not even anyone inside Agilent. ;-)

    Why Does Keysight Still Use a HeNe Laser?

    One might ask why a "lowly" HeNe laser is still used more than 40 years after the two-frequency interferometer was first developed, and the design hasn't changed in any fundamental way in all that time. The answer is simple: It's still arguably the best approach and is likely to remain so for the foreseeable future. While designing a two-frequency interferometer using another type of laser would be possible, the Zeeman-split HeNe laser has certain fundamental characteristics that make it ideal and affordable - well the latter in a relative sort of way! :) One alternative might be a Diode Pumped Solid State (DPSS) laser. While a solid state alternative could be developed using a pair of DFB diode lasers or DPSS lasers offset-locked to each-other or a single laser with an AOM bifringence to generate the second frequency, matching the stability of the HeNe laser would be a challenge for the basic reason that the gain curve for most solid state laser mediums is at least two orders of magnitude wider than for HeNe, and that would make it much more difficult to maintain the precise location of the lock point - and thus wavelength "yardstick" - constant. An external reference like a Fabry-Perot resonator would probably be required. Furthermore, at the production volumes of metrology lasers, the manufacturing cost would be extremely high given the precision crystals and optics that would be required, as well as their mounting, alignment, and testing. It's sometimes hard to beat good old gas laser technology!

    HP/Agilent Metrology Lasers through the Ages

    Here is a rough chronology of the HeNe laser tubes used in HP/Agilent metrology lasers. Click on the model number for a diagram of the internal structure. Note that the diagrams do not distinguish among variations in tube design such as OC mirror reflectivity and magnet field strength required to achieve the specific REF frequency for each model laser. These are more or less to scale, subject to limited resolution graphics and a bit of artistic license. ;-)

    The duplication in laser models is due to switching over to a different type tube. For example, around 1990, the tube used in the 5501B and 5517A/B was modified to have a slightly shorter distance between mirrors and a higher tube voltage (thus "Long-LV" and "Long-HV"). It is speculated that the reduced mirror spacing was required to eliminate "rogue modes" in the locked output of higher REF frequency lasers (5517C/D), and the higher tube voltage came about as a result of reducing the diameter of the stepped bore, which increased gain and output power, even though the active discharge is actually shorter in these tubes. All lasers built after that used the new design (though it may have been tweaked slightly as the cavity spacing changed from 127 to 126 mm) until the Short tube was introduced around 2003 for the 5517E/FL/G:. Special versions of the 5517D with a high REF frequency (up to 5 MHz or a bit more) were then available as options and these used the Short tube. Later, the standard 5517D and 5517B/C migrated to the Short tube. The exact transition date is unknown, only that by mid-2013, both a standard 5517B and 5517D were found with Short tubes. Thus, now there may only be a single size tube - the Short. But even if all lasers going forward use a Short tube, it is likely that there is more than one version differing in the OC mirror reflectivity to accommodate the required range of REF frequencies for models from 5517B to 5517G, and the N1211A. So all Short tubes are not created equal. Much more on all this below.

    There is little known about the N1211A "Fiber AOM Laser" beyond that it uses a thermally tuned tube optimized for high output power at the expense of REF frequency. The N1211A has now switched over to using a special version of the Short tube which has both the required higher OC reflectivity optimized for high power as well as a slightly lower divergence (reasons unknown). But the collimator (which was overly complex) has changed to an actual beam expander with no adjustments at all. There is ample space inside the Short tube glass envelope to accomodate a longer cavity length if that was required to achieve higher power at the low REF. It could be at least as high as the 126 mm length of the Long-HV tube (possibly with a centering "spider" added to minimize the chance of fracture from side-ways shock). But there is no evidence to suggest that anything but the standard ~100 mm cavity has ever been used in the Short tube.

    Explanation of Axial Zeeman HeNe Laser Behavior

    I've yet to see a treatment of the axial Zeeman effect that isn't either total hand waving or pages of hairy math! I can hand-wave with the best of them, but would welcome something more satisfying. :) All fundamental physical phenomena have fundamentally simple explanations that at least capture the spirit, if not all the details). (OK, well, except perhaps for high energy physics where hypothetical particles and extra dimensions are created at the pleasure of high energy physicists who need to attract equally high energy grant money.) I'll even accept that an axial magnetic field splits the neon gain curve and results in the separate shifted gain curves for left and right circularly polarized modes. But why do the two lasing lines move apart?

    There are two possible "simple" causes of the lasing frequency shifts resulting from Zeeman splitting: Mode pulling (which tends to attract each lasing mode towards its respective gain center) and magnetically-induced birefringence of the plasma (which results in the effective cavity length differing for each lasing mode's polarization).

    Several papers attributes the phenomenon entirely to the birefringence of the plasma for right and left circularly polarized photons. (See for example: T. Baer, F. V. Kowalski, and J. L. Hall, "Frequency stabilization of a 0.633-µm He-Ne longitudinal Zeeman laser", Applied Optics, vol. 19, no. 18, 15 September 1980). While birefringence may be a factor, those researchers were testing at at very low magnetic fields and their results don't explain some observations for lasers like those from HP/Agilent. In addition to cavity loss (or cavity Q) not entering into their equations at all, one fundamental result - that the split frequency is a minimum where the lasing lines are centered on the split gain curves - is exactly opposite from reality for most HP/Agilent and similar lasers! Finally, a Zeeman beat can often be observed with HP/Agilent tubes at very low magnetic fields. Where I test these tubes, there are a variety of magnets a few feet away stuck to a steel cabinet. The residual field from them - probably a few G or even less - is sufficient to result in a beat frequency from a typical HP/Agilent tube chirping through the audio range. Same with just the residual magnetism from a <50 G magnet left on the steel centering strips inside the tube. And unlike the signal with high fields, the one in this very low field regime is generally not sinusoidal. The beat frequency will typically peak at 10s or 100s of kHz at a slightly higher field (e.g., 25 G) and then decline and disappear as the field is increased until a beat reappears in the normal range of higher field strength at perhaps 100 G. That leaves mode pulling in higher field regime. (See the section: Axial Zeeman Experiments Using Variable Magnetic Field. Note that mirror coatings in common HeNe laser tubes may have enough asymmetry (resulting in birefringence or whatever) that they typically do not produce a beat frequency until well over 100 G. And tubes called "flippers" which have unstable polarization produce a Zeeman beat at lower magnetic fields. HP/Agilent tubes generally behave like the quintessential totaly unstable flipper where even the axes of polarization appear to change at random.)

    Though I've now lowered the bogosity quotient to 0.1 (down from a much higher value until quite recently), some aspects of the following explanation may still be totally without any basis in fundamental physics. But it has the attractive property that using some reasonable assumptions and not-so-hairy math, it is able to predict the approximate behavior of real HP/Agilent two-frequency lasers. And there is also support for the mode pulling being involved in the book: "Gas Lasers", by Charles Geoffrey Blythe Garrett, McGraw-Hill advanced physics monograph, 1967. Yes, most of the serious work on this stuff was done in the first decade after the invention of the laser. That is, until what's here. ;-)

    The mechanism for the shift of the Zeeman modes away from the cavity modes is a type of mode pulling - or at least can be modeled that way. There is no need to invoke esoteric effects like plasma or mirror birefringence (though these may alter the results in subtle ways, particularly at low magnetic fields). Mode pulling essentially shifts the positions of the longitudinal modes of a Fabry-Perot laser slightly away from the locations determined by c/2L of the linear cavity, toward the gain center. The basic mode pulling equation for a normal (non-Zeeman) laser (with truly galactic-size gobs of assumptions!) is:

      FS = FSR * ---------
                  GB + CB


    For high-R mirrors with equal reflectivity, Finesse = π*sqrt(R)/(1-R). Where one of the mirrors is HR as is the case with HeNe lasers, Finesse will actually be close to 2*π*sqrt(R)/(1-R). This equation predicts the shift of an adjacent longitudinal mode 1 FSR away from a mode centered on the peak of the neon gain curve toward the peak. As will be seen later, this is where the magnetic field comes into play. What the equation does show is that for GB >> CB, FS is proportional to CB, or equivalently, the lower the finesse of the cavity, the more frequency shift will be present. But at least this should get us into the ball park.

    A factor of 2 should be tossed in since what we're interested in is not the shift of a single mode, but the change in distance between the two modes due to each of them shifting in opposite directions. However, this can arguably be partially offset by the fact that the modes are positioned only part of the way down down the gain curves due to the Zeeman splitting magnetic field, not at 1 FSR away and farther down where the mode pulling effect would be greatest. So, it's somewhere in between. (Don't worry, much of this hand waving will go away shortly.) So, we'll use a fixed value in place of FSR selected and for want of something better, let's select it so the REF frequency is reasonable for the 5517B tube whose mirror reflectivity is known.

    So, here are 3 laser tubes with OC mirrors whose reflectivity has been measured (dissected conventional Melles Griot 05-LHR-006 barcode scanner tube, HP-5501A tube, and HP-5517B tube), and several others where the reflectivities can be estimated based on difference frequency (REF) specifications (5517A/C/D/E/F/G). There is also a Melles Griot 05-LHR-007 (same as the Spectra-Physics 007), which is about the shortest commercially available barcode scanner tube. The conventional tubes were installed in an HP magnet (model unknown or not remembered) but not locked to the mid-point between the split neon gain curves:

    Conventional tubes:

                 Cavity     Cavity   OC Mirror   Cavity    <--- F2-F1 (REF) --->
      Tube Type  Length      FSR    Reflectance  Finesse   Predicted    Measured
       LHR-006*  139 mm   1.078 GHz    99.0%       625     1.34 MHz   1.2-1.6 MHz
       LHR-007   110 mm   1.360 GHz     " "         "      1.70 MHz   1.5-1.7 MHz

    HP/Agilent lasers (constant magnetic field):

                  Cavity    Cavity    OC Mirror   Cavity    <--- F2-F1 (REF) --->
      Tube Type   Length     FSR     Reflectance  Finesse   Predicted   Specified
       5501A*     130 mm   1.153 GHz    98.74%     495.5    1.74 MHz   1.5-2.0 MHz
       5501B      127 mm   1.180 GHz    98.8%      520.4    1.70 MHz   1.5-2.0 MHz
       5517A       "  "     " "   "     98.8%      520.4    1.70 MHz    "   "  MHz
       5517B*      "  "     " "   "     98.5%      415.7    2.13 MHz   1.9-2.4 MHz
       5517C*      "  "     " "   "     98.0%      311.0    2.71 MHz   2.4-3.0 MHz
       5517D       "  "     " "   "     97.5%      248.2    3.65 MHz   3.4-4.0 MHz
       5517E     101.6 mm  1.475 GHz    96.8%      193.2    5.94 MHz    >5.8   MHz
       5517F       "  "     " "   "     96.2%      162.2    7.07 MHz    >7.0   MHz
       5517G       "  "     " "   "     96.1%      157.9    7.26 MHz    >7.2   MHz

    "*" denotes tubes where the OC mirror reflectance was measured.

    (Really old 5501B and even some original 5517 lasers used a tube with a mirror spacing of 132.5 mm which thus had a correspondingly smaller FSR. There may also be some small variation in the cavity length even for later tubes (126 versus 127 mm). This discussion will apply to those as well with only very slight modifications, left as an exercise for the student. More on this below.)

    Note that the predicted REF values here assume a constant (though not specifed) magnetic field, but F2-F1 is roughly proportional to field strength. And based on measurements of many HP/Agilent lasers, the magnetic field may differ by up to nearly a factor of 2 depending on model and specific sample. (At first, I assumed they were all the same!) Higher REF frequency lasers like the 5517F (nearly the highest split frequency laser available) on average have stronger magnetic fields than lower REF frequency lasers like the 5517A, but there are some notable exceptions. (See the section: HP/Agilent 5517 Laser Construction.) So, a selection process must be involved in mating magnet and tube to achieve a specific split frequency. Nonetheless, the mirror reflectivities for higher split frequency lasers in the chart are almost certainly way too low and would likely result in no lasing at all for these short tubes. Thus the actual mirrors would have somewhat higher reflectivities since their magnetic field strengths tend to be higher as well. For example, the 5517G mirrors may be easily over 97%. While still low for a normal laser (where 99% or more would be typical), it is a much more realistic value for the short cavity of the 5517G where maximum output power is not the objective.

    OK, so I kind of picked the reflectivities for the 5517A/C/D/E/F/G mirrors to make the results reasonable. :) With the increasing cavity loss, the output power of lasers with higher REF frequencies will tend to be lower, but the reflectivites listed may simply be too low to lase at all or with useful power on a tube of this length. However, the actual change in discharge length going from the 127 mm to the 101.6 mm cavity is small, perhaps at most 10 mm, so not that much gain is lost. And the bore of the short tubes appears to be narrower and uniform (as opoposed to the stepped bore of the long tubes. This would also result in higher gain, possibly totally making up for the loss in discharge length. Eventually, I will measure the reflectivity of OCs for other HP/Agilent lasers. But I don't have any tubes that I'm willing to take to bits at the present time, partially due to (1) the physical and emotional trauma that would result and (2) the fact that I haven't located the special chants and incantations required for metrology laser tube sacrifice. :) If anyone has done this, or has certifiable 5517 tube bits or a 5517 tube that's already cracked or broken they'd be willing to contribute to the cause, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    Of course, all these nice results based on numerous assumptions may be wishful coincidence, but they are close to what is observed and don't require delving into esoteric plasma physics. Whew! :)

    However, since we do know that the magnetic field is what splits the neon gain curves and moves them apart in proportion to the field strength and this affects the split frequency roughly via a proportionality constant (up to a point where the output power goes to zero!), the magnetic field must be added to the equation. Measurements suggest that the magnets used in the various 5517A/B/C/D/E lasers and 5501B tend to have a higher strength for higher split frequencies, but not always. So, mirror reflectivity alone is probably not used for split frequency selection. And there may in fact be some "mix and match" going on mating tubes with magnets to achieve the desired split (REF) frequency for the specific laser model. For example, the spec'd REF frequency for the 5517B is 1.9 to 2.4 MHz. The completed laser must start out with a REF that is low enough to anticipate the effects of normal tube aging where the beat frequency tends to increase as well as operation at high ambient temperature where the REF frequency tends to decrease (due to higher tube pressure). So, new 5517Bs typically have a REF frequency between 2.1 and 2.3 MHz.

    If we assume that the split frequency is proportional to the magnetic field, we can modify the super simplified mode pulling equation as follows:

      DF = B * ZSS * k * ---------
                          GB + CB


    Note that the product of H, ZSS, and k replaces the FSR (mode spacing) in the previous equation. Assuming that CB is small compared to the GB, this simplifies to:

                     MHz             CB
       DF = B * 2.8(-----) * 1.17 * ----
                    Gauss            GB

    (MHz/Gauss are just there to signify the units, not variables. B is in Gauss.) The value 1.17 for k in the equation above results from the desire for the predicted REF frequency of the 5517B (whose mirror reflectivity and magnetic field have both been determined) to be the nominal value for the 5517B, 2.20 to 2.30 MHz. :) It's not always needed and k = 1 sometimes works fine. :( :) An adjustment may simply be needed because the magnetic field is not uniform along the axis of the HeNe laser tube. Thus, the result might be called a "phenomenological equation" - based on physical principles but adjusted to fit actual observations. :-) Garrett's equivalent of "k" is 1.00. But it turns out that this can be dealt with as well. :) More below.

    Given that: FSR = c/(2L), finesse = π * 2 * sqrt[R/(1-R)], and simplifying the squareroot term to 2 * sqrt(R)/(1+R) for R close to 1, the result is:

                     MHz                c * (1-R)
       DF = B * 2.8(-----) * kk * ----------------------
                    Gauss          GB * π * L * sqrt(R)


    And if the Garrett equation is boiled down to its fundamentals, the result is similar. (You really don't want to see the original!) So in the end, it comes down to the difference frequency being proportional to the product of B and CB multiplied by a constant. Or in more detail (at least to the first order), the DF is proportional to the magnetic field multiplied by the OC transmission and divided by to the cavity length. With a constant thrown, the "proportional" becomes an "approximately equal". ;-)

    As was shown above, the difference frequency is (approximately) inversely proportional to the neon gain bandwidth (GB). All of the calculations so far have assumed the GB to be 1.5 GHz, which is used rather than 1.6 GHz because the GB will be narrowed slightly for each of the split gain curves since they each only have 1/2 the original gain. This helps to boost the difference frequency but also decreases the power output at the lock point! Other factors being equal, the GB for a shorter tube with less gain will be even narrower, further increasing the difference frequency, thus making it easier to obtain the required higher ones with lower magnetic fields and/or higher OC reflectivities. And, since GB gets narrower as the overall gain declines with use, this could also explain at least in part why the split frequency tends to increase over the life of the tube. How all these factors interact is not something I'd want to contemplate, let alone calculate! :)

    However, there would appear to be something even more signficant going on. Looking at the bottom diagram in Normal and Zeeman-Split HeNe Laser Mode Power Curves, which is based on actual data, in order for there to be no unwanted "rogue" modes oscillating, it is seen that the effective GB (lasing output power curve which is basically the neon gain curve above threshold) must not be much greater than the FSR of the laser tube, which For the 5517B (long tube), is 1.180 GHz. And it must not extend to or beyond the FSR on either side of the lock point. Or to put it another way, the maximum total shift cannot be greater than 2*(FSR-GB/2) or both the center split mode and the adjacent modes will be above threshold. Thus, the GB cannot be 1.6 or even 1.5 GHz in this case because then there would enough gain for modes on either side of the split lasing mode to lase. In the diagram, the GB is about 1.28 GHz, which is consistent with an actual measurement in SFPI Display of Lasing Mode Power Envelope of Horizontal Polarized Output of Healthy HP/Agilent 5517B Laser. The maximum total shift in this case would be 1.08 GHz. Conventional HeNe laser tubes do not seem to exhibit this effect, or at least not to the same extent.

    This brings up the issue again of whether the gases in HP/Agilent tubes are isotopically pure since using only 20Ne or 22Ne would reduce the GB significantly. The shift in the gain center going from one isotope to the other is about 1 GHz! Using only a single isotope could easily account for this narrowing.

    Based on some numbers obtained from industry sources, the HP/Agilent mix is skewed toward Ne20. The specific conditions under which these numbers were obtained are not known, but they seem reasonable:

                    Vacuum     Absolute  Relative
      Isotope     Wavelength  Frequency  Frequency   Offset
        Mix          (nm)       (THz)      (MHz)     (ppm)       Comments
     Pure 22Ne    632.990084  473.613198    +905    +1.911
     50% 20Ne     632.990752  473.612698    +405    +0.855
     85% 20Ne     632.991219  473.612348     +55    +0.116
     90.48% 20Ne  632.991293  473.612293       0     0      Natural Isotope Ratio
     95% 20Ne     632.991353  473.612248     -43    -0.091  5517C/D/E/F/G, 5519A/B
     96.5% 20Ne   632.991372  473.612234     -59    -0.124  5501B, 5517A/B, 5518A
     Pure 20Ne    632.991420  473.612198     -95    -0.200

    (Another stable isotope of neon is 21Ne, about 0.27%, which is ignored and lumped in with 22Ne.)

    However, we don't know the conditions under which these estimates were made. The optical frequency is quite sensitive to the He and Ne pressures. If these differed even by a small amount, there would be a significant change in optical frequency. The optical frequency is also modestly sensitive to the temperature. If these values are at a temperature like 25 °C rather than much higher internal temperature at which these lasers operate, that would also represent a modest difference. There is more on this in the section HP/Agilent Laser Wavelength/Optical Frequency, below.

    But if these numbers are accurate, the HP/Agilent mix is skewed a bit toward 20Ne compared to the naturally occuring mixture. And if in the interest of cost reduction, HP switched to a mixture with less 20Ne, that would explain the change in wavelength in the early to mid-90s - from 96.5 to 95 percent pure 20Ne. Note that even using the natural mix might be better than a more equal ratio since it would narrow the neon gain bandwidth. This would have two advantages: It would increase the REF frequency for a given magnetic field and increase the magnetic field at which rogue modes would appear. There would probably not be any dramatic decrease of output power (as there might be with a common multi-longitudinal HeNe laser) becuase only the split Zeeman mode will be present when locked.

    But it turns out that the narrowing actually only occurs at the high magnetic fields present in HP/Agilent lasers. With no magnetic field, the GB of a healthy tube is between 1.5 and 1.6 GHz as expected. So, isotopic purity may not be fundamental to the line narrowing, with mode competition at the lock point being more important.

    Using a Scanning Fabry-Perot Interferometer (SFPI), The GB of a healthy short 5517 tube was measured with a magnet and naked. Don't ask where I found a healthy bare tube! I can only say that no lasers needed to be sacrificed this time. :) Gain and GB will increase as the tube warms up from its bore discharge alone, but for these tests, 8 V was applied to the internal heater to bring it up to approximately the temperature at which the laser would normally operate:

    So the high magnetic field in conjunction with mode competition narrows the GB. Whether the narrower GB is really what should be used in the equation isn't known for sure. But it would appear to make sense assuming the mode pulling for the RCP and LCP modes is independent. However, since the effective GB decreases with with increasing magnetic field, the beat frequency may increase faster then would be accounted for simply by the increase in magnetic field and this may help to explain at least in part why theory and practice do not agree - yet!

    As noted, the GB narrowing appears to be much more significant in HP/Agilent tubes compared to conventional tubes. Testing of a Melles Griot 05-LHR-007 barcode scanner tube using an SFPI shows a GB of around 1.6 GHz without a field and and at best goes down to 1.5 GHz inside a 5517C magnet. Further, the skewing in the lasing output power profile due to mode competition is much less severe with HP/Agilent tubes. My hypothesis is that conventional tubes are filled for maximum power at a ratio of round 1:1 with 20Ne:22Ne while HP/Agilent tubes are filled with nearly pure 20Ne to narrow the gain bandwidth.

    Another phenomenon that isn't addressed at all is the variation in beat frequency during mode sweep. All of these simplified mode pulling equations use only the distance from gain center but not the gain profile, so position of the split lasing mode doesn't matter as long as it is between the gain curves. But we know that depending on the magnetic field, the beat frequency may be a maximum or minimum when the split mode is centered between the gain curves - or it may not change much at all.

    And then, there is a second term in the Garrett formulation that has been totally ignored in anything discussed so far and will continue to be ignored for the forseeable future!

    But here goes:

                 2.8 MHz       CB
       DF = H * --------- * ----------
                    G        1.28 GHz

    (As above, H is in gauss.) This is precisely the simplified form of the Garrett equation. So, all the bogosity terms must have canceled out. ;-)

    Plugging in numbers for the reference 5517B:

                     2.8 MHz     2.84 MHz  
       DF = 362 G * --------- * ---------- = 2.25 MHz
                        G        1.28 GHz  

    Now, this isn't 100.0000000 percent accurate. :) The resulting DF becomes 2.25 MHz to make it work for the reference 5517B. But it was still close, previously 2.21 MHz, and the actual DF for that specific tube in that specific magnet is not known.

    The strengths of the magnetic fields for samples of most of the long tube HP lasers (5501A/B and 5517A/B/C/D) have been measured directly. Those of the Short tube Agilent lasers (5517E/F/G) have been estimated by measuring the fringe field at the center of the magnet exterior relative to the fringe field of a 5517E, which was the only laser that had a label (363g) with the field strength. Then the interior field was kind of guessed based on measurements of magnets from the long tube lasers. This is the only option available for determining the interior field on intact tube assemblies. Since the consistency among the various magnets on the ratio of outside to inside is not very good, the actual fields may be quite different, most likely lower. More info on the measurements may be found in the section: HP/Agilent 5517 Laser Construction.) But for the few tubes that have been removed from the magnet, the field in the interior could be measured inside. But even so, the field may vary significantly along the length of the tube, so the value used is a sort of more or less average. :) I built a simple gauss meter specifically for making these measurements. See the section: Simple Gauss Meter for Measuring Zeeman Magnet Strength.

    Adding the magnetic field into the equation for the 05-LHR-006 and 05-LHR-007 barcode scanner tubes, and what may be a special version Hughes 3121H (unmarked) somewhat longer tube produces results that are quite reasonable. While the OC mirror reflectivity for these tubes has not been measured, 99% is usually a good value for Short tubes. The two entries for the LHR-007 are in 5517D and 5517A magnets; the LHR-006, LHR-640, and 3121H are in a 5517C magnet. The range of predicted REF values corresponds to a GB between the 1.6 and 1.28 GHz.

    Conventional tubes (measured magnetic field):

      Tube    Magnet  Cavity    Cavity  OC Mirror  Cavity  <---- F2-F1 (REF) ---->
      Type    Field   Length     FSR     Reflect.  Finesse   Predicted     Actual
     LHR-007  250 G   110 mm  1.360 GHz   99.0%      625   0.95-1.19 MHz  1.12 MHz
      "   "   380 G    "  "     "    "     " "        "    1.45-1.81 MHz  1.66 MHz
     LHR-640  300 G   118 mm  1.272 GHz    " "        "    1.07-1.33 MHz  1.25 MHz
      "   "   350 G    "  "     "    "     " "        "    1.25-1.56 MHz  1.50 MHz
     LHR-006  350 G   139 mm  1.078 GHz    " "        "    1.15-1.43 MHz  1.25 MHz
     3121H    350 G   192 mm  0.781 GHz    " "        "    0.76-0.96 MHz  0.70 MHz

    For the 3121H, the measured REF is slightly out of range low but the cause may be that for this longer tube, the magnet covered only about 2/3rds of the active bore.

    Note that due to the gas-fill and cavity length, some if not all of these tubes will produce "rogue modes" even when the Zeeman-split mode is centered between the gain curves as it would be when locked in an HP/Agilent laser. (Rogue modes are non-Zeeman lasing lines on the tails of the gain curves.) For example, the 05-LHR-640 in the 350 G 5517C magnet had a pair of rogue modes, each about 5 percent of the total power. Reducing the magnetic field to approximately 300 G eliminated them. Only the 05-LHR-007 with its shorter cavity length may be immune.

    The 05-LHR-640 in the chart above was in like-new condition with a power output of around 1.25 mW, well above its 0.5 mW spec. A high mileage 05-LHR-640 with a power output of only 0.6 mW was also tested in the same magnet. It had a split frequency of around 1.75 MHz - well above the predicted range - probably due to the reduced gain and/or increased cavity loss. However, interestingly, rogue modes were still present, and possibly even a bit larger relative to the Zeeman mode amplitude than for the healthy sample.

    For the HP/Agilent laser tubes, if we use the measured magnetic field where available, or else the estimated magnetic field, the values for mirror reflectivity become more realistic:

    HP/Agilent lasers (measured or estimated magnetic field):

     Tube    Magnet  Cavity    Cavity   OC Mirror  Cavity   <--- F2-F1 (REF) ---> 
     Type    Field   Length     FSR      Reflect.  Finesse  Predicted   Specified
     5501A*+ 371 G   130 mm   1.153 GHz   98.74%    495.5   1.89 MHz  1.5-2.0 MHz
     5501B+  256 G   127 mm   1.180 GHz   98.5%     415.7   1.59 MHz   "   "  MHz
     5517A*  259 G    "  "     " "   "    98.5%     415.7   1.60 MHz   "   "  MHz
     5517B*+ 362 G    "  "     " "   "    98.5%     415.7   2.24 MHz  1.9-2.4 MHz
     5517C*+ 328 G    "  "     " "   "    98.0%     311.0   2.71 MHz  2.4-3.0 MHz
     5517D+  380 G    "  "     " "   "    97.7%     270.0   3.65 MHz  3.4-4.0 MHz
     5517EL* 445 G  101.6 mm  1.475 GHz   97.8%     282.4   5.82 MHz   >5.8   MHz
     5517E+  363 G    "  "     " "   "    97.0%     206.3   5.90 MHz   >5.8   MHz
     5517F   470 G    "  "     " "   "    97.5%     171.4   7.00 MHz   >7.0   MHz
     5517G   485 G    "  "     " "   "    97.5%     171.4   7.22 MHz   >7.2   MHz

    The cavity length of 127 mm (~5") for the 5517A/B/C/D correspond to that of the most common Long-HV tubes. The "*" denote lasers where the mirror reflectivity has been measured (5517EL). That 445 G may be excessive (as well as for the other entires above 400 G), but unfortunately, I do not have the magnet to test. The "+" denote lasers where the axial magnetic field of at least one sample has been measured or is believed to be known from the label (5517E). Where interior measurements were not possible because the tube was not removed, the average value of the magnetic field predicted from external measurements (which can be quite unreliable) was used along with mirror reflectivities selected (e.g., fudged) to make the resulting REF frequency be a reasonable value.

    The measured values correspond to the field inside on-axis in the center of the magnet. However, the field of an ideal uniformly magnetized cylinder is also not constant. Peak values may be 10 to 15 percent higher halfway from the center to the ends compared to the value at the center, with the field declining to exactly 0 G at the very ends. See Field Along Central Axis of Ideal Magnet used in HP/Agilent Laser. This plot is labeled "Ideal" because the fields of actual HP/Agilent can depart significantly from these curves. Most have little or no center dip and are not at all symmetric end-to-end.

    To further complicate matters, tests of the REF frequency of original laser assemblies and the same tube reinstalled in the same magnet after having been extracted show that the field can decrease due to trauma from the tube removal process itself, possibly by as much as 15 percent or more. The decrease may be greater for a segmented magnet if the segments are separated at any time (as they would be to simplify removal of the tube), as well as for stronger magnets (e.g., from 5517Ds) which will tend to have a greater decrease, those with more non-uniform fields where the field distribution can change non-uniformally, and for a particularly uncooperative potting job leading to a higher amount of trauma. So the original field strength is not known for any of the lasers listed above except for the 5517E, where the field was labeled but not measured since it was not disassembled, and a single 5517C whose locked REF had been recorded prior to the tube-ectomy. That one declined by just under 10 percent. The REF frequency of other tube-magnet combinations had clearly declined but the exact amount is not known. Hopefully the Law of Large Numbers will prevail: Averaging all these uncertainties should produce an accurate result! ;-)

    So, it looks like the identical tube is used for the 5501B, 5517A, and 5517B, with the higher magnetic field boosting REF for the latter. But the 5517C and 5517D require lower mirror reflectivities to be consistent with the measured field strength (and this has been confirmed for the 5517C at least). To keep things simple, a cavity length of 127 mm has been used for all thermally tuned lasers. But really old ones had a cavity length of 132.5 mm rather than 127 mm (for those built after around 1990). They also had a wider bore. Shortening the cavity increases the REF frequency and also increases the rogue mode limit. Reducing the bore diameter increases gain and thus output power (as well as increasing the operating voltage). So, taken together, these may have been a set of design refinements. But a late model (2006) 5517C tube was found to have a cavity length of only 126 mm. And based on SFPI tests, an even newer 5517D as well. A 1 mm reduction in the very precisely fabricated mirror spacing rod is definitely a design change, not a process variation. Its significance, if any, is not known, but this was also probably fine tuning.

    As a practical matter, all these different tube types would not be necessary (based on the mirror reflectivity estimates, above) but it's almost certain there were more than one since the range of magnetic fields is not sufficient for both the 5517A and 5517D. However, with careful control of tube design, it's quite possible for there to be only two types of tubes - long and short. All tubes of each type would then be identical - including mirror reflectivities - with REF set solely by magnetic field strength. Perhaps Agilent has an adjustable field magnetizer like mine. (Shown in Sam's Magneto-Matic Dial-A-Field™ Alnico Magnet Charger. See the section Sam's Alnico Zeeman Magnet Chargers for more info on this device.) Insert tube assembly and set the field strength to obtain the desired laser specifications. ;-) Then, we could have the following:

    HP/Agilent lasers (Long-HV and Short tubes, selected magnetic field):

     Tube   Magnet  Cavity     Cavity  OC Mirror  Cavity   <--- F2-F1 (REF) --->
     Type   Field   Length      FSR     Reflect.  Finesse  Predicted   Specified
     5517A  240 G   127 mm   1.180 GHz   98.3%     366.4   1.71 MHz   1.5-2.0 MHz
     5517B  310 G    "  "     " "   "     " "       " "    2.21 MHz   1.9-2.4 MHz
     5517C  370 G    "  "     " "   "     " "       " "    2.64 MHz   2.4-3.0 MHz
     5517D  210 G  101.6 mm  1.475 GHz   97.0%     206.3   3.76 MHz   3.4-4.0 MHz
     5517E  330 G    "  "     " "   "     " "       " "    5.90 MHz    >5.8   MHz
     5517F  395 G    "  "     " "   "     " "       " "    7.07 MHz    >7.0   MHz
     5517G  405 G    "  "     " "   "     " "       " "    7.25 MHz    >7.2   MHz

    The mirror reflectivities were selected to be as high as possible without requiring magnetic fields so strong that they introduce problems of their own. Though Alnico magnets can have a field strength "at the poles" of over 1,500 g, nothing close to this may be achievable inside an actual cylindrical magnet. (In fact, measurements of some of these magnets do show more than 1,000 G at the poles even though the interior field is less than half that.) But more fundamentally, the neon gain curves will be spread so far apart that rogue modes may be generated if the magnetic field exceeds about 385 G for long tubes and 475 G for Short tubes. And the limit at which the output power goes to zero because the neon gain curves are spread so far apart that there is no longer any overlap may not be far above 400 g. Thus, the listed fields could be too strong for the 5517F/G.

    It would be possible to select the mirror reflectivity for long tubes such that the 5517D could use either type tube. But to keep the field below the "rogue mode limit", would have required a mirror reflectivity of under 98%.

    These same tubes are also used in other common HP/Agilent lasers. The 5518A would use the 5517A or 5517B tube depending on serial number, the 5519A would use the 5517C tube, and the 5519B would use the 5517D tube. There are some (not very common) options mostly for the 5517D where a higher REF frequency is specified. These are probably accommodated with magnetic field adjustments, or for very high values, the use of a Short tube.

    As if this isn't confusing enough, as of 2013, it appears as though *all* 5517 laser tubes are of the short variety. This is confirmed by my testing of a 5517B from 2013 with a Short tube, Type II Control PCB, and VMI PS 504 HeNe laser power supply. So this probably replaces the long tube for 5517Bs by using a combination of higher OC mirror reflectivity and different magnetic field. And the Agilent Web site shows Long tubes for all 5517A/B/C/D lasers as either obsolete or only available until supplies are exhausted.

    HP/Agilent lasers (Short tubes with two different OC mirrors, selected magnetic field):

     Tube   Magnet  Cavity     Cavity  OC Mirror  Cavity   <--- F2-F1 (REF) --->
     Type   Field   Length      FSR     Reflect.  Finesse  Predicted   Specified
     5517A  240 G  101.6 mm  1.475 GHz   98.8%     520.4   1.71 MHz   1.5-2.0 MHz
     5517B  310 G    "  "     " "   "     " "       " "    2.21 MHz   1.9-2.4 MHz
     5517C  370 G    "  "     " "   "     " "       " "    2.64 MHz   2.4-3.0 MHz
     5517D  210 G    "  "     " "   "    97.0%     206.3   3.65 MHz   3.4-4.0 MHz
     5517E  330 G    "  "     " "   "     " "       " "    5.90 MHz    >5.8   MHz
     5517F  395 G    "  "     " "   "     " "       " "    7.08 MHz    >7.0   MHz
     5517G  405 G    "  "     " "   "     " "       " "    7.25 MHz    >7.2   MHz

    (Note that Agilent no longer offers the 5517A, so that entry is not needed, though its absence would not change anything else.) Based on Deep Throat evidence :), it appears as though Agilent did indeed use two versions of the Short tube to cover the common laser models. One for 5517A/5501B through 5517C and the other for the 5517C and 5517D. The 5517C is included for both because it may use either version, perhaps based required output power. However, the higher REF lasers like the 5517DL and above (REF above 4 MHz) used several different tubes. Perhaps they were fine tuning the parameters as it would seem that a single model would suffice.

    One might ask if it would be possible to use a single OC mirror reflectivity and vary the magnetic field to adjust REF. Aside from the output power being limited for lasers like the 5517B due to the low mirror reflectivity, a very low magnetic field might result in less stability of the REF frequency. For example, if all tubes had a 97% OC, the magnetic field for the 5517A would be under 100 G. However, other companies have had Zeeman lasers with relatively puny magnetic fields, so perhaps it is the output power or some other more obscure reason. Or perhaps they now do! :)

    Finally, there is one other Agilent laser, the N1211A "Fiber AOM Laser", which is probably designed for higher power rather than a specific range of REF frequencies since the REF frequency doesn't really matter the way it is being used: A pair of Acousto-Optic Modulators (AOMs) are employed to shift the optical frequencies of the F1 and F2 components to generate a much higher difference frequency (REF) between F1 and F2. The original REF from the tube adds into that but is small in comparison. And it's value is nearly irrelevant because it can be compensated for by selecting the AOM frequency shifts. Since REF can be lower, the mirror reflectivity could be higher to boost power, which could approach or even exceed 1 mW in a new laser. Measurements of the external magnetic field of one N1211A laser suggested that it has a magnet like the one in the 5517C and that could indeed work out assuming an OC mirror reflectivity of around 99% as in the first entry below:

    Agilent N1211A lasers (Long-HV tubes, selected magnetic field):

     Tube     Magnet  Cavity    Cavity   OC Mirror  Cavity   <--- F2-F1 (REF) ---> 
     Type     Field   Length     FSR      Reflect.  Finesse   Predicted  Measured
     N1211A   350 G   126 mm   1.190 GHz   99.0%     625.2    1.45 MHz   1.60 MHz
      " " *+  370 G    "  "     " "   "    99.05%    658.2    1.46 MHz      NA
      " "     350 G    "  "     " "   "    98.9%     568.0    1.60 MHz   1.60 MHz
      " "     250 G    "  "     " "   "    98.5%     415.7    1.56 MHz   1.60 MHz

    That one sample had been removed from service with an output power of only around 700 µW, so its REF frequency may have already crept up some. Or, if the mirror reflectivity were really only 98.9%, then the predicted REF frequency would be 1.60 MHz. However, since the correlation between external and actual internal magnetic fields is rather poor, it could simply be a 5517A tube selected for high power installed in a 5517A magnet, which would be quite likely. Why create yet another tube type for what must have been a very limited production run of these very specialized lasers? But they would need to be selected for high power. However, I've now disassembled a different one of these tube assemblies that had been broken before I received it. In fact, the OC reflectivity for that sample is actually 99.05%, very close to 99%, which may have been the actual target reflectivity. But I wouldn't be at all surprised if some of these tubes used the lower field with the 98.5% OCs - same as 5517As and several other standard 5517 tubes. Unfortunately, since it was pre-broken, I could not measure the actual REF frequency. The mirror spacing rod was also found to be 126 mm as expected, like newer 5517 lasers. Its magnetic field was found to be around 370 G in the center and 1/2 the way to one end, and somewhat lower 1/2 the way to the other end. I did take care to minimize trauma to the magnet during the discombobulation process (you don't want to know the details), so its field should be close to the original value.

    Speculating that Agilent/Keysight has switched to using the Short tube for the N1211A, they could either go with a similar mirror reflectivity and lower field, or increase the reflectivity which may be a bit more optimal in terms of power output for the shorter cavity:

    Agilent N1211A lasers (Short tubes, selected magnetic field):

     Tube     Magnet  Cavity    Cavity   OC Mirror  Cavity   <--- F2-F1 (REF) ---> 
     Type     Field   Length     FSR      Reflect.  Finesse   Predicted  Measured
     N1211A   295 G  101.6 mm  1.475 GHz   99.0%     625.2    1.52 MHz      NA
      " "     370 G    "  "     " "   "    99.2%     782.2    1.52 MHz      NA

    Agilent switched over to using Short tubes for N1211As as early as 2012 though some N1211A tube assemblies with manufacturing dates as late as 2015 still use the Long-HV tubes. I suppose there was old stock to use up. :)

    Mirror tests

    I've come across several boxes of Agilent HeNe cavity mirrors undoubtedly intended for use in these type of laser tubes, almost certainly "Long tube" lasers. These were mostly only photos of the label but I do have a few boxes in-hand and was able to measure the diameter and RoC which match mirrors from dissected 5517 tubes within my margin of error. The first were labeled "2% HeNe cc/pl" with a manufacturing date of 2003. The "2%" refers to the mirror transmission, or equivalently, a reflectance of 98%. (I actually measured them to be 2.1%T/97.9%R but we'll go with the labeled values.) There was no further description so they could have been for any of the Agilent lasers. But the 5517C would probably be the most likely. However, it's possible that Agilent was actually adjusting the reflectivities to reduce the number of different tube types. These same mirrors would work easily for 5517As, 5517Bs, and 5517Cs, with somewhat lower magnetic fields than in the table, above. But using them for a "Long tube" 5517D would require a very strong magnetic field, possibly resulting in rogue modes:

     Tube   Magnet  Cavity     Cavity  OC Mirror  Cavity   <--- F2-F1 (REF) --->
     Type   Field   Length      FSR     Reflect.  Finesse  Predicted   Specified
     5517A  210 G   127 mm   1.180 GHz   98.0%     311.0   1.74 MHz   1.5-2.0 MHz
     5517B  275 G    "  "     " "   "     " "       " "    2.28 MHz   1.9-2.4 MHz
     5517C  330 G    "  "     " "   "     " "       " "    2.73 MHz   2.4-3.0 MHz
     5517D  420 G    "  "     " "   "     " "       " "    3.48 MHz   3.4-4.0 MHz

    Using them for "Short" tubes would have no problems with 5517Ds, but would require a magnetic field lower than used in any other HP/Agilent lasers (though found in Zeeman lasers from other manufacturers). (However, their RoC would not be correct for use in "Short tube" lasers.) It is not known if there is a problem with Zeeman stability or other reason why lower fields are not found in HP/Agilent lasers:

     Tube   Magnet  Cavity     Cavity  OC Mirror  Cavity   <--- F2-F1 (REF) --->
     Type   Field   Length      FSR     Reflect.  Finesse  Predicted   Specified
     5517A  150 G  101.6 mm  1.475 GHz   98.0%     311.0   1.78 MHz   1.5-2.0 MHz
     5517B  190 G    "  "     " "   "     " "       " "    2.26 MHz   1.9-2.4 MHz
     5517C  230 G    "  "     " "   "     " "       " "    2.73 MHz   2.4-3.0 MHz
     5517D  310 G    "  "     " "   "     " "       " "    3.69 MHz   3.4-4.0 MHz

    A similar box of mirrors has a part number with "5518" in it so they may have been intended for 5518A lasers. It listed the same 136 mm RoC but a transmission of 1.6% (98.4%R). This is enough greater than what I measured (1.5%T, 98.5%R) for the 5517A (which should be similar to the early versions of the 5518A) to suggest that Agilent may have been trying to boost output at the expense of REF frequency. And another one listed 2.1% (97.9%R), also with a "5518" in the part number, so that could have been for later versions of the 5518A where the REF frequency is spec'd to be 1.7 to 2.4 MHz, or even for 5519A/Bs (2.4 to 3.0/3.4 to 4.0 MHz). And yet another labeled "3%" (97%R), also with 5518 in the part number, but would more likely be used in a "long tube" 5517D or 5519A/B and is perfect for any of the "Short tube" lasers and the default in the chart, above. But now we are into absolutely totally pure speculation. :)

    One thing is certain: The wide variation in mirror reflectance/transmission confirms the overall scenario whereby cavity loss is a key parameter in designing HP/lasers to meet REF frequency specifications. And the range found so far agrees with my predictions.

    Some of these mirrors have a flat on one side, purpose unknown. At first I thought it might simply be the edge of a larger plate from which these were cut with the optics version of a plug cutter. But it could be for indexing or simply to prevent the mirrors from rotating in their mounts, or to provide a path for the discharge to pass by the mirror if the anode were behind it (possible in the short-tubes). Some of the mirrors were also thinner than the normal ~4 mm. It's possible some or all of these were prototypes or trial runs, I'm not sure how much significance to attach to these details. :)

    Axial Zeeman summary

    1. An axial magnetic field splits the neon gain curve into two parts due to the Zeeman effect, shifting one up in optical frequency and the other down in optical frequency.

    2. A cavity mode that intersects either gain curve will result in lasing if the gain is above the lasing threshold.

    3. The resulting lasing mode will be Right Circularly Polarized (RCP) from one gain curve and Left Circularly Polarized (LCP) from the other.

    4. Mode pulling shifts the position of the lasing mode(s) toward their respective gain centers as a function of the magnitude of the Zeeman Shift of the gain curves, and the gain bandwidth and cavity bandwidth (or equivalently, the cavity Q). The shift is approximately proportional to magnetic field times cavity bandwidth, and inversely proportional to gain bandwidth.

    5. Where a cavity mode intersects both gain curves above threshold so both RCP and LCP lasing modes are present, mode pulling in opposite directions of each one moves them apart slightly creating a "split lasing mode". A beat frequency equal to the distance between them will be detected by a sufficiently fast photodiode behind a linear polarizer. The beat frequency is also called the "split" frequency or REF (for REFerence) frequency.

      For the short HP/Agilent tube, there can only be a single split lasing mode present when the laser has locked (READY on solid) centered between the two gain curves, though a second RCP or LCP mode may be present at times during mode sweep before the laser locks. For longer tubes, there may be additional RCP or LCP modes or even multiple split lasing modes. There may then be additional beats present due to the split modes and with RCP or LCP modes.

    6. The precise beat frequency and its amplitude will vary based on how mode pulling affects the position of each of the components of the split lasing mode. This will depend on the separation of the gain curves and how mode competition changes their effective shape.

      For an HP/Agilent tubes with its strong magnetic field, the maximum beat frequency usually occurs when the split lasing mode is centered. With weaker magnetic fields, the minimum frequency may be at that location. This is also probably related to the profile of the neon gain curves and how mode competition changes their effective shape. The maximum beat frequency for HP/Agilent lasers varies from 1.5 MHz to over 7 MHz depending on the model.

    7. To produce the desired beat frequency while retaining useful output power without generating rogue modes requires a careful choreography between several key design parameters including tube length, mirror reflectivity, magnet strength. The 5517D (long tube) and 5517G (Short tube) appear to be at the hairy edge for all of these.

    To reiterate, the only direct consequence of the Zeeman effect is to split the discharge spectral lines and shift their position up and down in optical frequency. Since the neon gain curves are related to the emission spectra, they then also get shifted. Lasing which takes place will then be RCP or LCP depending on which gain curve applies. The Zeeman effect does NOT produce the beat frequency directly. A spectrometer or interferometer with sufficiently high resolution would show a split of the spectral lines regardless of whether the gas is in a laser or not. Thus the Zeeman effect also has nothing directly to do with creating the lasing lines except that both LCP and RCP components will be present if the gain curves overlap at the cavity mode location. The lasing optical frequencies - the RCP and LCP lines - are then spread apart slightly by mode pulling depending on where they are relative to their respective gain centers affecting both the difference frequency and its magnitude.

    For example, in a typical HP/Agilent 5517B laser, the Zeeman shift of the two neon gain curves may total around 1 GHz. But mode pulling shifts the RCP and LCP components of a single split lasing mode apart by only 2.2 MHz (1 part in 455). This is the difference (or beat or split or REF) frequency. Increasing the magnetic field would increase the Zeeman shift and corresponding difference frequency proportionally until the point where the neon gain curves are so far apart that the gain for a mode centered between them is below threshold and there is no lasing at all.

    HP-5517C HeNe Laser Tube Mode Sweep Versus Magnetic Field has several plots for very weak to normal field strengths. More on this can be found in the section: Axial Zeeman Experiments Using Variable Magnetic Field. Until the numrical discrpencies have been resolved, consider this as a qualitative demonstration of the effect of changing the magnetic field, but don't take the exact field values too seriously.

    Unresolved issues:

    Now back to your regularly scheduled programming. :)

    Observation of Zeeman Splitting of HeNe Laser Modes

    There are various ways of confirming that a Zeeman-split lasing line is actually present for the axial (as well as transverse) Zeeman HeNe laser. However, direct methods like spectroscopy would be extremely difficult if not impossible due to the very small spacing of the Zeeman-split lasing line. While your Physics 101 Zeeman lab was able to demonstrate the Zeeman effect with a strong magnet, sodium or mercury lamp, and spectrascope, the effects on the lasing line are 2 or 3 orders of magnitude smaller and would not be detectable even with the finest optical spectrum analyzer. The spectral emission lines will split and separate with a sensitivity of 2 or 3 MHz/Gauss while the Zeeman-split lasing line of a typical HeNe laser tube will only have a sensitivity of 5 or 10 kHz/Gauss. Here are two options:

    The first is by far the easiest and generally sufficient except for extremely dedicated academic types. :)

    There is a YouTube video claiming to show Zeeman splitting of the HeNe lasing line using stone knives and bear skins, well almost. :) However, while what is shown is due to the Zeeman effect, it is NOT the split lasing line, but rather rogue lasing modes on the tails of the split gain curves. See Longitudinal Zeeman splitting with a HeNe laser.

    Rogue modes originate when the laser cavity is so long that both the (normally desired) Zeeman-split lasing mode AND 1 or more normal modes fit within the gain bandwidth of the split neon gain curve as shown in: Avoiding Unwanted "Rogue" Modes in Zeeman-Split HeNe Lasers. Where the tube is too long and thus its longitudinal mode spacing (also known as FSR) is too small, there will be significant gain one or more mode spacings away from the desired Zeeman-split lasing mode. Where this is above the lasing threshold, additional rogue lasing modes will appear. And with a large enough magnetic field, the central split mode may be totally non-existent (though that is unlikely with the field from a home-built electromagnet).

    The laser tube in the video appears to be 9-10 inches long, with a mode spacing of around 700 MHz. Thus the split rings in the video are actually the rogue outer modes in the second diagram, above, about 1.4 GHz apart. Like the central Zeeman-split mode, they would also be left and right circular polarized. I estimate the FSR of the etalon in the video to be around 10 Ghz, consistent with the split being around 10 percent of the ring spacing. It all makes sense! Too bad. :( :)

    Estimating the Rogue Mode Limit

    The approximate equation for avoiding rogue modes is:

     FSR > 0.5(GB + ZS) = 0.5(GB + 2.8MHz*B)
     B < ----------
          2.8 MHz

    Because of the difference in the numerator, B at the rogue mode limit increases proportionally faster than FSR.

    So for the HP/Agilent Long-HV tubes with an FSR of 1.18 and the Short tube with an FSR of 1.5 GHz, B at the rogue mode limit will be around 390 G and 614 G, respectively. This assumes an effective GB of 1.28 GHz, which may be narrower than reality, so maximum B may be lower. And for a typical barcode scanner tube with an FSR of 1.04 GHz and GB of 1.5 GHz, max B would be only 207 G. (The GB is assumed to be wider for the barcode scanner tube because it is probably filled with equal isotopes of 20Ne and 22Ne to maximize power, while the HP/Agilent tubes are filled with nearly pure 20Ne in part to maximize REF.)

    However, as an added complication, the effective GB is also dependent on the cavity losses. Increased losses reduce the height of the gain curve and thus the GB at the lasing threshold decreases. Tubes like the 5517D with high %T OC mirrors may avoid rogue modes at a greater magnetic field than tubes like 5517As or N1211As. The same applies to high mileage tubes whose gain has declined. So instead of rogue modes appearing at 390 G, they may remain well behaved at 450 G or even 500 G. B at the rogue mode limit for mid-life N1221A and 5517D tubes was approximately 445 G and 475 G, respectively.

    Factors Affecting the REF/Split Frequency

    All of these are discussed in great length elsewhere in this chapter, but here's a quick summary of the most significant factors affecting the REF/split frequency at the optimum lock point (centered between the split neon gain curves):

        Factor                Mechanism                 Effect
        Magnetic Field        Spreading of Gain Curves  Increases
        Gain/Power            Width of Gain Curves      Decreases
        Cavity Length         Cavity Bandwidth          Decreases
        Mirror Transmission   Cavity Loss               Increases
        Diffraction Losses    Cavity Loss               Increases
        Mirror Misalignment   Cavity Loss               Increases
        Internal Temperature  Gain?                     Decreases

    HP/Agilent Laser Output Power and REF Frequency versus Use

    With common HeNe lasers, output power may decline slightly with thousands of hours of use and then drop precipitously near end-of-life. However, with some, power will actually increase initially due to the tube being overfilled. And in many cases, there may be a decline in output power mostly due to misalignment of the mirrors that occurs after many power cycles. This is fully recoverable with realignment. There is of course, no REF frequency for such vanilla (or cherry!) flavored HeNe lasers. :)

    I do not have any full life data for HP/Agilent lasers. But in general, the trend is for output power to decline, usually accompanied by an increase in REF frequency, especially near end-of-life. The gas-fill He:Ne ratio and pressure do change with use affecting gain. And, optics can degrade increasing cavity loss. These will result in lower output power and higher REF. The strength of the permanent magnetic field also affects output power and REF frequency, but if it were to change at all, it would most likely decrease, resulting in an opposite effect, most notably reduced REF. Until recently most HP/Agilent lasers used laser tubes (designated "Long-LV" and "Long-HV") where mirror alignment is permanently fixed by glass-glass contact, so changes in alignment cannot be the cause for either power decline or REF frequency increase (except possibly if subjected to extreme abuse). However, some "high REF" Agilent lasers (5517E/EL/F/FL/G/GL and special 5517D/DLs) and most lasers manufactured after around 2012 use a tube (designated "Short") where the alignment of the rear (HR) mirror can change on its own, but there is not yet enough data to determine how common that might be. Of the dozen or so Short tubes that's I've tested, most had near optimal alignment, even those from high mileage lasers. However, a new/NOS 5517D tube and a N1211A tube had power increase by 25 percent or more after alignment. Both of these may have been originally rejected for low power caused by careless adjustment during manufacturing, not drift. The one electronic adjustment in these lasers - the tube temperature set-point - has a modest effect, but rarely changes on its own. Cavity length is fixed for any given laser tube but is included here as one of the relevant parameters. And while there may be a very small effect from external losses in the optics due to changes in beam expander alignment and contamination over the years, it's almost always insignificant.

    As noted, the REF frequency of HP/Agilent/Keysight lasers tends to increase with laser run-time and may exceed the upper limit of the specifications for the particular laser before failure due to some other cause like low output power. Whether this results in problems depends on how excessive REF is handled by the measurement electronics. A new laser will generally have a REF frequency 10 to 20 percent above the minimum spec. This is likely due to normal tolerances in manufacturing, to anticipate a slight decrease in the Zeeman magnetic field from aging, and also to avoid problems from ferrous materials in the vicinity of the laser, which can also reduce the field by 5 percent or more.

    Here's how the output power and REF frequency are affected by various physical parameters:

        Parameter       Output        REF
       (Increases)       Power     Frequency   Comments
          Gain         Increases   Decreases   Declines with use
       Cavity Loss     Decreases   Increases   Mostly due to mirror transmission
      Magnetic Field   Decreases   Increases   May be higher for higher REF
       Temperature     Increases   Decreases   Set by electronic adjustment
      Cavity Length    Unchanged   Decreases   100, 102, 126/127 mm
      External Loss    Decreases   Unchanged   Windows, beam expander, WPs, etc.

    Of course, these aren't independent. For example, effective gain increases with temperature and tubes with a longer cavity also generally have higher gain due to a longer discharge length.

    The following table shows how the output power and REF/split frequency changes with use for some recent vintage (2004-2006) Agilent 5517B lasers:

      Laser  Label PWR  Warm PWR  W-PWR/  Label REF  Warm REF  W-REF/
       ID       (µW)      (µW)    L-PWR     (MHz)     (MHz)    L-REF
        1       355       395      1.11      2.27      2.17     0.96
        2       580       525      0.91      2.29      2.39     1.04
        5       560       510      0.91      2.29      2.29     1.00
        6       607       525      0.86      2.22      2.27     1.02
        7       580       516      0.89      2.22      2.32     1.05
        9       680       530      0.78      2.20      2.47     1.12
       12       251       222      0.88      2.19      2.17     0.99
       13       560       305      0.54      2.23      2.75     1.23
       16       654       519      0.79      2.23      2.51     1.13
       17       660       574      0.87      2.24      2.35     1.05
       18       628       387      0.62      2.24      2.67     1.19
       20       625       543      0.89      2.24      2.36     1.05
       21       642       436      0.68      2.26      2.44     1.08
       23       580       500      0.86      2.22      2.35     1.06
       25       628       536      0.85      2.25      2.31     1.03
       27       692       637      0.84      2.28      2.35     1.05
       28       568       495      0.87      2.26      2.36     1.04
       29       562       473      0.84      2.30      2.43     1.06
       30       564       512      0.91      2.27      2.30     1.01
       31       457       375      0.82      2.29      2.49     1.09

    (The missing laser IDs did not have label values for output power and/or REF Frequency.)

    Though the actual number of on-time hours is not known, these lasers were all labeled by Agilent with the original output power and REF frequency. ID #1 is believed to either be grossly mislabeled, or an unused spare, which thus has like-new performance. All the others have probably been run for thousands of hours. There appears to be no correlation between the manufacturing date and relative performance compared to the label values, so the date is not listed here. And for this reason, it is believed that these lasers were replaced after a specific number of on-time hours or for performance reasons, not when a fab shut down. Several have REF frequencies that have climbed to near or beyond the upper limit for the 5517B (2.4 MHz) and may pulled from service for that reason. (However, they would now meet all specifications for the 5517C!) But on some others, while the output power has declined significantly, the REF frequency is unchanged or even lower than the labeled value. So, it's possible that there is indeed more than one mechanism accounting for the changes in output power and REF frequency. With gain and cavity loss being critical, each of these will degrade at different rates.

    HP/Agilent Laser Split Frequency and Output Power Versus Tube Temperature

    These measurements were motivated by the apparently random changes in behavior from one power cycle to the next on many HP/Agilent lasers. While the controller is supposed to drive the heater so the tube reaches a fixed set-point temperature before switching to optical feedback, this may not work as well as would be assumed. In addition to ambient conditions, it's possible that other factors converge to make this less than totally consistent. These may include how long the tube takes to start, slight resistance in the two pin heater connector, and the phase of Mars. ;-) So, I decided to measure the output power and split frequency with the laser tube run open-loop with respect to temperature control, by controlling it manually with a variable DC power supply. The temperature range was from ambient (a cold start) with only the bore discharge contributing to power dissipation until thermal equilibrium was reached after a 1+ hour warmup. Then power was applied to the heater in small increments until a maximum of about 7.5 VDC. During this time, the maximum split frequency for each mode sweep cycle (displayed on a frequency counter) was recorded by hand and the output power was monitored by a PC Data Acquisition System (DAQ). Due to lack of sufficient hands and eyeballs, as well as the sample rate of the frequency counter, it's possible that some values of the split frequency are in error, especially near the beginning where it's changing rapidly. Thus, the ripples and wiggles in the plots may simply be artifacts of inaccurate recording. The scribbled data and DAQ plots were then stitched together sort of the way dinosaur DNA was reconstructed for Jurassic Park. ;-) The "Mode Number" is the number of full mode sweep cycles relative to thermal equilibrium (no heater drive). See HP/Agilent 5517B Split Frequency and Output Power Versus Mode Number. Decreasing REF frequency is consistent with increasing output power, though this change is larger than I had expected. The reason the slope of the output power curve below Mode Number 0 is somewhat higher is because it was from initial power-on and there could be processes involved in addition to internal heating. The range above thermal equilibrium from Mode Numbers 0 to 46 goes from well below to well above the normal operating point, which for this specific laser, has a REF frequency in between 2.37 and 2.40 MHz, so around Mode Number 26.

    After reconnecting the Control PCB so the laser would lock normally, I used a hair dryer to confirm that heating the overall tube also affected the split frequency and output power with the cavity length maintained constant by the feedback loop. However, the correlation between split frequency and output power was not quite the same as when using the internal heater. So, the temperature and pressure of the gas inside the tube is a factor but not the whole story. But both the split frequency and output power changes could be caused by increased Doppler broadening at higher temperatures. This would both reduce the mode pulling effect thought to cause the Zeeman mode splitting (and thus reduce the split frequency), and increase the gain at the "valley" between the Zeeman-split gain curves (and thus increase the output power).

    The typical 5517 laser goes through a total of about 70 mode sweep cycles from power-on until READY starts flashing. But because the heater resistance is only sampled every 25 seconds or so, the temperature can overshoot by a large amount. The controller then supposedly backs off and zeros in on the set-point. Supposedly.

    HP/Agilent Laser Output Power Versus Magnetic Field

    Normally, output power is not a variable that's specified directly during manufacturing, only that it meets the minimum specification of the model (and any options) for the laser. And the magnetic field is set at the factory to achieve the desired REF frequency. While there is some range over which it is acceptable (e.g., 1.9 to 2.4 MHz for the 5517B), the target value will generally be more or less centered to accommodate tube aging from use (which increases REF) and the magnet loosing strength from handling or storage in proximity to other magnets (which reduces REF. For example, just an unrestrained screwdriver being attracted to and whacking the magnet may reduce the field, and thus REF, slightly. So, for most normal applications, you get what you get. ;-)

    But how does the strength of the magnetic field affect the output power when locked? The answer is: "It's complicated". :) There are three regions with respect to magnetic field strength:

    1. No field or very low field: In this situation, there is no mode splitting and the lasing mode will be drawing from the one population of excited neon atoms. In addition, the mirrors in HP/Agilent tubes are designed to have virtually no asymmetry (birefringence, etc.). Although a standard Scanning Fabry-Perot Interferometer (SFPI) display of the longitudinal modes will be stable, the mode structure with respect to polarization flipping with little or no magnetic field is generally quite chaotic.

    2. Modest field: The neon gain curve has split and spread apart somewhat. There is a split lasing mode so it's two components can draw from separate populations of excited neon atoms. The neon gain curves in the overlap region are also relatively flat so power is a maximum at the lock point.

    3. Strong field: The neon gain curve has split and spread apart by a large amount. There is a split lasing mode so it's two components can draw from separate populations of excited neon atoms. The neon gain curves in the overlap region are highly sloped so power dips at the lock point. If the field is increased too far, the locked output power goes to exactly zero. :( :) This may actually show up (without increasing the field) on high mileage tubes with a dip in power at the lock point whose gain has declined - typically 5517Ds or other lasers with even higher REF frequencies.

    Therefore, the maximum locked power is probably going to be present with a modest magnetic field. Based on a single data-point so far :), it may be 5 to 10 percent higher than with little or no field. Thus if high stage slew rate is not critical, setting the magnetic field to maximize power would be a valid approach - if 5 to 10 percent higher power matters. ;-) This usually means reducing the field, which can be accoplished in a variety of simple ways. However, as a general guideline, for HP/Agilent lasers, keeping REF above at least 1 MHz or so is probably prudent both to assure stable operation and be compatible with processing electronics.

    Notes on the Waveplates in HP/Agilent HeNe Laser Tube Assemblies

    The raw output of the tube in a HeNe Zeeman laser has the F1/F2 optical frequencies but they are not useful in a two-frequency interferometer in their original form. When these are discussed in textbooks and research papers, it is generally stated that the magnetic field splits the original mode into two modes that are left and right circularly polarized. A Quarter WavePlate (QWP) is then used to both convert them to orthogonal linear polarized modes and to align them with the axes of the laser as required for use in an interferometer. The reason is that a QWP takes a left-hand circular polarized input and produces a linear polarized output at +45° with respect to its fast axis, and takes a right-hand circular polarized input and produces a linear polarized output at -45° with respect to its fast axis. The input orientation doesn't matter. So, the QWP could be oriented to align the output linear polarizations without regard to the input.

    But HP lasers from day 1 (the original 5500A around 1969 even before it had an official model number) and Agilent/Keysight lasers to the present have all had both a QWP and HWP, with the basic design unchanged over more than 40 years. Further, both waveplates (WPs) are in mounts that allow the tilt of each one to be adjusted around one of its principle axes. So why both a QWP and HWP and why the tilt?

    Real HeNe laser tubes exhibit some random amount of birefringence both from the fine structure of the mirror coatings as well as from unavoidable geometric asymmetry in their construction. Without a magnetic field or explicit polarization control measures such as a Brewster window or plate, these tend to lock the polarization of the longitudinal modes to a fixed orientation about the tubes optical axis, and 90 degrees from it. In a HeNe laser with an axial magnetic field such as one from HP/Agilent, the asymmetry will result in the Zeeman modes being slightly elliptically polarized. And since the magnetic field is far from uniform or even perfectly axial in these lasers, it will affect the polarization as well. Thus the orientation of the QWP will matter and only specific ones will convert the elliptically polarized Zeeman modes to orthogonal linearly polarized modes. But in general, they won't be aligned with the system's Horizontal (H) and Vertical (V) axes, so the HWP is then required to rotate them to match. And if the degree of ellipticity of the raw Zeeman modes isn't the same (ratio of major to minor axes), even the combination of QWP and HWP may not be able to make both of the output modes purely linear and they may not even be quite orthogonal.

    In principle, a simple test for raw F1/F2 modes that are not circular or linearly polarized modes that are not orthogonal would be to direct the beam to a high speed detector WITHOUT a polarizer in front of it. Then there should be absolutely no AC response - only the DC term representing the total optical power. A null result would be confirmation. But in dealing with very small numbers buried amongst very large numbers, it doesn't take much for there to be some response. However, this can only be done consistently if the laser is locked. In a test of one tube where the raw beam was split off (so that locking would occur normally), there was a beat at the REF frequency without a polarizer of a few percent of the value with a polarizer at 45 degrees. In an attempt to determine if the symmetry or other characteristics of the magnetic field influenced this, various magnets and non-magnetic pieces ofe ferrous material were manipulated in the vicinity of the magnet. Even using an amplified detector (Thorlabs DET55), the change in amplitude was too small to be conclusive. The REF frequency was also changing, and that also affects the response due to the limited bandwidth of the detector. Of course, any effect on the magnetic field from outside the magnet may simply be too small to be detectable (other than REF frequency). In another attempt, the position of the lock point on the split gain curve was shifted on both sides away from the center. No change was detected either.

    The adjustable tilt allows the exact retardation of each waveplate to be altered slightly. While I have been skeptical they are there simply to be able to use cheapo waveplates that might not always be exactly 1/4 or 1/2 wave, this explanation may actually be correct since the accuracy of the retardaion is critical to producing F1 and F2 modes that are purely linear and precisely orthogonally as required for the metrology applications. The WPs are made from what looks like optical-grade mica whose discrete layers preclude the ability to select the exact retardation by controlling thickness. But even expensive waveplates might not be close enough and would require tilt adjustment anyhow. And whether mica waveplates were originally selected based on low cost or zero order or temperature stability or being very thin to avoid significantly shifting the beam when tilted or being what the designers had laying around is not known either. However, based on waveplate theory (yes, there is such a thing!), selecting inexpensive waveplates that can be fine-tuned does make sense. Note that an exact QWP requires an exact HWP. But if the first waveplate isn't exactly quarter wave, this can be compensated for by adjusting the second waveplate to not be quite halfwave or vice-versa!.

    It's often possible to swap the entire waveplate assembly between HP/Agilent tubes of similar type (e.g., 5517 or 5501B) and achieve acceptable performance without any adjustments, as long as they are installed with the same orientation and direction. Therefore, it would appear that the differences tend to be small.

    One characteristic that appears to be common to most or all HP/Agilent lasers is that the mode purity in the H axis is better than that in the V axis. This could indicate that H is the one that optimized during initial adjustment. If everything were symmetric, it would be assumed that optimizing H should also optimize V. The fact that this is not true implies an asymmetry, and it may be very simple: The output window of the tube is set at a slight angle. It's only a few degrees, but that may be enough to introduce a sufficient polarization preference to force both circularly polarized modes out of the tube to be slightly elliptical with respect to the same axis. Waveplates alone would not be able to compensate for that asymmetry. It would require something like a similar angled plate at right angles to the output window of the tube mounted before the waveplates. The degradation in performance - assuming can even be detacted - was probably never considered significant enough to worry about.

    However, this does suggest a simple way to compensate for asymmetries that might be present: Adding an angled plate to equalize the circularity of the two components coming out of the tube.

    I did tests of waveplate sets from three HP-5517 lasers using a linearly polarized HeNe laser. The results are as follows:

                         <--------- Orientation ---------->
        ID#    Laser     Input     QWP       HWP     Output
         1   5517C 9 mm   +20°     +20°     +32.5°    +45°
         2   5517B 6 mm   -20°     -20°     +12.5°    +45°
         3   5517B 6 mm   +20°     +20°     +32.5°    +45°

    (The specific type of 5517 laser or its beam diameter makes no difference.) Angles are with respect to vertical when viewed from the front of the laser. The accuracy of my measurements on orientation is within ±2° for input and ±1° for output, though the latter at +45° is probably quite precise based on theory.)

    The Input is the orientation required for the polarized HeNe laser to produce a pure linearly polarized beam at the output and thus also of the orientation of the optical axes of the QWP. Only if aligned with an optic axis of the QWP will the polarization remain linear, a requirement for these tests. As expected, the output orientation is the same in all cases since the desired output will be rotated by 45° to align with the H and V axes. This is a result of the conversion from circular to linear polarization by the QWP, at 45° with respect to its optical axes. The orientation of the HWP was inferred from the transfer function from input to output.

    The significance of the +20° or -20° is not obvious. For pure circularly polarized inputs, this would put the linear modes out of the QWP at +65° or +25° (and multiples modulo 90°). The tubes are installed in the magnet with any physical asymmetries approximately aligned with the H or V axes (0° and 90°). If the axes of elliptical modes line up with the physical asymmetries, the outputs from the QWP at +20° or -20° would still be somewhat elliptical and not useful - the QWP would need to be at 0° or 90°. If the modes are pure circular, then it's just arbitrary.

    Now, it's quite possible that the orientation of the QWP was chosen at random or based on some jig and not actually determined for the specific tube unless it was found that the adjusting the HWP alone would not meet specifications. However, I have found that it may be necessary to iteratively tweak the QWP and HWP to achieve best purity of the F1/F2 modes - the same result could not be obtained by adjusting only the HWP. Indeed, with a genuine original HP/Agilent tube that has its waveplates optimally adjusted, the purity of the F1/F2 modes is often nearly perfect. So, perhaps they start at some fixed orientation for the QWP and go from there.

    In a test of another waveplate assembly, it was found that aligning the optic axis of the QWP vertical and the optic axis of the HWP to be -22.5° (CCW) for a 5517 or +22.5° for a 5501B (CW) was close enough to allow the laser to lock, but fine adjustment was required to optimized orthogonality for use in an interferometer.

    With the Zeeman tube producing a pure two-frequency left and right circularly polarized beam, all of these waveplate assemblies would result in pure orthogonal linearly polarized outputs oriented along the H and V axes. This was confirmed by placing a true (non-HP) QWP in the linearly polarized HeNe laser's beam at +45 or -45 degrees to produce pure right and left circularly polarized inputs to the HP WPs. The results were linearly polarized outputs oriented along the H and V axes. In neither WP assembly, was there any indication that the tilt of either WP was adjusted for other than pure 1/4 or 1/2 wave as the extinction when set for optimal linearly polarized outputs was nearly perfect.

    Excel 1001A/B/F metrology lasers, which are functional clones of HP/Agilent lasers, only have a QWP on a similar tilt-rotate mount, and that may be nearly as good as the HP approach (though it's also possible the tubes and magnets used by Excel exhibit less birefringence). On two sample lasers (a 1001B and 1001F), it was impossible to achieve quite the same degree of orthogonality with the single waveplate as in some HP/Agilent lasers. But it was close, and other variables like laser power and beam overlap also affect the measurements. And for that matter, using only a QWP with HP/Agilent tubes also works well enough in some cases. So perhaps Excel concluded that the improvement resulting from the use of a HWP was not worth the effort and cost. It may simply be that HP/Agilent has retained both of them out of caution - why mess with something that works! Or both waveplates are still there to provide temptation for would-be twiddlers!

    However, I believe I do have a possible explanation of sorts with respect to the HWP: HeNe laser tubes used in Zeeman lasers still have some residual polarization preference, even if specifically designed for this purpose. By orienting the QWP at a specific angle (which may not coincide with the normal polarization axis), the circular mode purity can be optimized. But that angle isn't the same for all tubes. When a conventional tube is installed in a laser like those made by Excel Precision, it can be oriented optimally since nearly all modern tubes are physically rotationally symmetric. That cannot be done with HP/Agilent/Keysight tubes where it is desired to have the terminals in the same place all the time, so the HWP is there to correct for whatever the optimal angle turns out to be.

    For more, see the section: Adjusting the Waveplates in HP/Agilent Lasers.

    Open questions:

    1. Why do HP/Agilent/Keysight Zeeman HeNe lasers use both a QWP and HWP while similar lasers from other manufacturers only require a QWP?

      Assuming this is not simply a historical artifact, my conclusion is that the combination of QWP and HWP can sometimes result in more optimal orthogonality of the H and V F1/F2 components than a single waveplate, even if they have limited retardatation (tilt) adjustment.

    2. F1 and F2 should be perpendicular - precisely 90 degrees to each other. Anecdotal evidence suggests this is not always the case, both in tests with a polarizer on a rotation mount and because there may be a small beat signal present even without a polarizer in front of the detector. It may be that if the ellipticity of the Zeeman modes isn't the same, the result from the QWP and HWP may not be quite perpendicular.

      Is this true and if so, is it something that can be optimized or is optimized at the factory?

    3. Why does the F1/F2 orthogonality and/or mode purity vary slowly over a limited range on some lasers despite being locked?

      This may be related to similar periodic variations in REF frequency, possibly caused by slight reflections from various optical surfaces that change due to thermal expansion affecting the net cavity gain and thus the ellipticity of the Zeeman modes. This does not appear to be related to position on the split gain curve as adding an adjustment for that had no significant effect on anything except the F1/F2 power balance. Even the REF frequency changed by less than 1 percent when the power balance went from <1:1.2 to >1.2:1.

    4. Why do the settings of the QWP and HWP required for optimal orthogonality and/or orientation of the F1/F2 modes change slightly from one power cycle to the next, or even if lock is interrupted and re-acquired?

      Probably related to (3), above, but as a result of locking to a different "order" of mode sweep (and thus different temperature).

    I have yet to find any detailed information relating these aspects of HP/Agilent metrology laser technology. If anyone has knowledge or references related to the waveplate issue or anything else of relevance, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    HP/Agilent Laser Wavelength/Optical Frequency

    While most types of lasers, and in particular, HeNe lasers only operate over a relatively narrow wavelength (or equivalently, optical frequency) range, the HP/Agilent specifications include the nominal values to 8 or 9 significant digits. Thus, not only is the optical frequency known to a very precise value, it does not change much over the life of the laser. (The specifications for most HeNe lasers - even other stabilized HeNe lasers - only have "632.8 nm" or even simply "633 nm"!) The chart below lists the spec'd nominal wavelength in vacuum, in air (at STP assuming n = 1.00027593), and optical frequency, from HP/Agilent specifications:

      Laser   |<------- Wavelength ------>|      Optical
      Type       Vacuum            Air          Frequency
      5500A   632.99???? nm   632.81???? nm   473.6122?? THz
      5500B   632.99???? nm   632.81???? nm   473.6122?? THz
      5500C   632.99???? nm   632.81???? nm   473.6122?? THz
      5501A   632.99???? nm   632.81???? nm   473.6122?? THz
      5501B   632.991372 nm   632.816759 nm   473.612234 THz
      5517A   632.991372 nm   632.816759 nm   473.612234 THz
      5517B   632.991372 nm   632.816759 nm   473.612234 THz
      5517BL  632.991372 nm   632.816759 nm   473.612234 THz
      5517C   632.991354 nm   632.816741 nm   473.612248 THz
      5517D   632.991354 nm   632.816741 nm   473.612248 THz
      5517DL  632.991354 nm   632.816741 nm   473.612248 THz
      5517EL  632.99139  nm   632.816776 nm   473.612221 THz
      5517FL  632.99139  nm   632.816776 nm   473.612221 THz
      5517GL  632.99139  nm   632.816776 nm   473.612221 THz
      5518A   632.991372 nm   632.816759 nm   473.612234 THz
      5519A   632.991354 nm   632.816741 nm   473.612248 THz
      5519B   632.991354 nm   632.816741 nm   473.612248 THz

    It's not clear what accounts for the two different wavelengths (and thus optical frequencies) for most of lasers. There are no obvious physical differences to account for it. The tubes, beam samplers, and relevant portions of the control electronics are all identical. So, it's possible there was a change in isotopic gas-fill or pressure or something else between 5517A/5517B/5518A/5501B lasers and those that came after them. The difference of approximately 14 MHz is still way lower than the commercial-grade error spec of ±0.1 ppm (roughly ±47 MHz), so it really doesn't matter. For the Military-grade lasers, the exact optical frequency is measured and included in the calibration report. But a report for one laser I saw had the optical frequency over 10 MHz away from the spec'd value anyhow. My contact at NIST doesn't even know whether it's an actual change in wavelength/optical frequency or simply an upgrade to the calibration in the measurement electronics! Or a change in the speed of light. :) And a 2009 operation and service manual for 5517 lasers lists the wavelength of the 5517B/BL to be the same as for the others below it. This is the only Agilent document I've seen so far with the value 632.991354 nm for the 5517B/BL, so it may only apply to lasers made after 2009. That manual does have a number of bloopers in it so who knows? But this lends further support to the conclusions that the differences are not real. In addition, a 2015 Keysight (formerly Agilent!) datasheet shows the 5517A/B/BL/C/CL/D/DL lasers all having an optical frequency of 632.99137 nm yet the Laser and Optics Manual still has the more precisely inaccurate values above. :) The wavelength listed for the 5517EL/FL/GL is slightly longer which may be due to a different gas fill in the Short tubes used in these lasers. (It seems to go the wrong way for the higher lock temperature used with the "Short" tubes. Other documents still list 632.991354 nm.)

    I've compared the optical frequency of multiple 5517 and 5501B lasers and have found no evidence of any real difference, let alone one averaging 14 MHz as shown in the chart above. And while even the datasheet available from the Agilent Web site for these lasers still lists the two wavelengths, I highly doubt there are different tube fill procedures depending on model! My guess is that when the 5517C and 5517D were introduced around 1992, HP may have changed the gas-fill (as a cost reduction) along with some other parameters of the tube including mirror spacing rod length (for performance reasons). They then made the same changes to the other lasers going forward, but never updated the datasheet. Other than laser nutcases like me, who would ever know (including anyone currently working for Keysight)? Since environmental factors - particularly temperature - have orders of magnitude more impact on dimensional accuracy than the small discrepancy in specifications, and calibration against a physical standard should always be done when installing a new laser, it ends up being only of academic interest.

    Here are some data. These lasers are listed in more or less the order in which they were tested:

                  Locked   REF/     Balanced
     Laser Laser  Output   Split    Frequency
      ID   Type   Power    Freq.   Difference  Notes/Comments
      1    5517B  660 µW  2.3 MHz   -2.30 MHz  Faulty beam sampler was replaced
      2-0  5517B  480 µW  2.4 MHz   -1.44 MHz  Laser 2 with its (new) beam sampler
      2-1   " "    "  "    "   "    -1.35 MHz  Laser 2 with beam sampler 1
      2-2   " "    "  "    "   "    -6.75 MHz  Laser 2 with beam sampler 2
      3    5501B  450 µW  1.9 MHz    0.00 Mhz  Other lasers referenced to Laser 3
      4    5517E  120 µW  6.3 MHz   -2.10 MHz  Only unit with Type III Control PCB 
      5    5517D  120 µW  3.6 MHz   -8.25 MHz
      6    5517C  260 µW  2.7 MHz  -15.60 MHz  Tube run at 4.0 mA (not 3.5 mA)
      7    5517C  240 µW  2.9 MHz   -9.00 MHz       "                   "
      8    5517C  210 µW  2.7 MHz   -8.58 MHz       "                   "
      9-0  5517D   80 µW  3.7 MHz   -9.10 MHz  Laser 9 with its beam sampler
      9-1   " "    "  "    "   "    -7.10 MHz  Laser 9 with beam sampler 1
      9-2   " "    "  "    "   "   -11.10 MHz  Laser 9 with beam sampler 2
     10-0  5517A  550 µW  1.7 MHz   -7.63 MHz  Laser 10 with its beam sampler
     10-1   " "    "  "    "   "    -2.13 MHz  Laser 10 with beam sampler 1
     10-2   " "    "  "    "   "    -9.33 MHz  Laser 10 with beam sampler 2
     10-3   " "    "  "    "   "    -3.44 MHz  Laser 10 with beam sampler 3
     11    5501B  220 µW  2.1 MHz  -10.70 Mhz
     12    5501B  150 µW  1.8 MHz   -8.65 Mhz  Tube run at 4.0 mA (not 3.5 mA)
     13    5501A  100 µW  2.0 MHz  -23.48 MHz  Really high mileage!
     14    5501A   50 µW  2.1 MHz  -25.42 MHz     "           "
     15    5501A   35 µW  2.0 MHz  -34.47 MHz     "           "
     16    5517A  410 µW  1.6 MHz   -0.34 MHz  Tube run at 4.0 mA (not 3.5 mA)
     17    5517D  405 µW  3.7 MHz  +75    MHz  Rebuilt with non-HP/Agilent tube

    Diagram of Test Setup for HP/Agilent Laser Optical Frequency Comparison shows the way these measurements were made. The Polarizing Beam-Splitter (PBS) is used to combine the beams from the two lasers. It, and one of the lasers are on adjustable mounts since the beams need to be precisely aligned to get a beat. Half WavePlates (HWPs) in front of each laser allow the polarization of the beam to be rotated 90 degrees to select F1 or F2 through the PBS. The HWP enables the appropriate frequency components to be compared Depending on whether the laser is a 5517 or 5501A/B. A turning mirror a meter or more away can be inserted in front of or in place of the detector to project the beam onto a card for precise alignment.

    Photo of Test Setup for HP/Agilent Laser Optical Frequency Comparison shows how ugly it really is, but the scope on the right is displaying the actual beat signal between a pair of 5517Bs, around 1.5 MHz. This photo shows some additional components that aren't needed for the optical frequency comparison, but are part of my standard HP/Agilent/Keysight test setup. The output of the PBS could have been sent to a 10780C optical receiver instead of the biased photodiode which has a polarizer at 45 degrees to combine the two orthogonal F1/F2 components. The scope display is of the output of a Thorlabs DET110 biased photodiode. One disadvantage of using the 10780C or other optical receiver is that it has a lower frequency limit of about 100 kHz and and an upper frequency limit somewhere between 5 and 10 MHz. An RF/microwave spectrum analyzer could also be used for this, which with some minor modifications to the optics, could show all four frequency components (F1 and F2 from both lasers) at the same time. But that's for the advanced course!

    To test these lasers that may have been rebuilt by someone other than HP/Agilent/Keysight, a much higher bandwidth could potentially be required if they aren't using the same gas-fill or lock technique.

    For later measurements - and to be able to display that nice clean scope trace but make the setup really ugly - two external HeNe laser power supplies (the white boxes and the Variac) were added to eliminate the FM introduced by the switching noise/ripple of the internal HeNe laser power supplies. But for the tests of the difference frequency, they weren't used and wouldn't have affected the results since the high frequency FM would be averaged out. Older HP lasers were really terrible in this regard; newer HP/Agilent lasers with the VMI 217 or VMI 373 power supplies are much better since they include built-in ripple reduction. But my linear supplies have a beefed up filter capacitor bank along with the normal active current regulation so they are almost ripple/noise-free.

    The 5517 lasers were enclosed in standard cases (non-vented for all except the 5517E) and allowed to reach equilibrium (2 hours minimum). (Removing the cover may significantly change the optical frequency once equilibrium is reached.) The lasers in the photo don't have any clothes on, but, well, that's another matter! :) The 5501As were run naked so that the "Photodiode Offset" adjustment could be performed. (More below.) It might be best to do this via a hole in the cover as they do drift significantly with the cover in place. But I wasn't *that* enthusiastic!

    Laser 3, the first and healthiest 5501B to be tested, was chosen arbitrarily to be the reference for optical frequency. The Balanced Frequency Difference is the frequency of the mid-point between F1 and F2 for the subject (ID) laser minus the frequency of the mid-point between F1 and F2 for Laser 3. There's still a ±1 MHz or more uncertainty due to variations in the specific lock point of the two lasers being compared during any given run.

    The three 5501As are very well used and weak, but it was easy to obtain a beat between them and the 5517A, Laser 10 (which happened to be the last 5517 laser tested and thus conveniently left in place!). 5501As have a "Photodiode Offset" adjustment, which moves the lasing line on the split gain curve. It's the square pot (R4) on the Lock Reference PCB, clockwise rotation decreases optical frequency. I could have set them all to have a 0 MHz difference frequency, but this would have resulted in grossly unbalanced mode amplitudes for these high mileage lasers. So, they were adjusted according to the HP procedure - maximizing the F1-F2 REF frequency, which centers the lasing line between the split neon gain curve. Before doing this, it wasn't even possible to see the difference frequency with Laser 13 likely because it was too high for my instrumentation. This was probably because parts of Laser 13 had been swapped, including the tube, without making any adjustments. The Photodiode Offset adjustments on the other 2 5501As were quite close to optimal. However, this is a single turn pot which adjusts the mode ratio from 1:2 to 3:2, and thus the optical frequency varies significantly with very small changes in its position - possibly 50 MHz or more end-to-end. Going only by the REF frequency - which isn't perfectly stable - it's quite likely that there will be an uncertainty of 5 or 10 MHz. So, best would be to adjust this pot (make it a 10 turn pot!) through a hole in the cover after the laser has reached thermal equilibrium. And if what you want is a precise optical frequency and don't mind some possible mode imbalance, adjust it with respect to a reference laser like an iodine stabilized HeNe laser instead of for maximum REF frequency! :) And, although the optical frequency changes with the cover installed, the Photodiode Offset adjustment could still be optimal if the change is due to the tube temperature, and thus the gas pressure increasing. That would still maintain the same mode balance.

    Laser #17 was rebuilt by another company other than HP/Agilent. (I'm also not at liberty to reveal the company name.) It had its original laser tube removed from the magnet/optics assembly and replaced with a laser tube that is not from Agilent. This explains the large offset in optical frequency, which could result from any number of factors but is quite consistent with a tube uning non-isotopically pure neon. (See below.) The offset is probably of little practical consequence as long as it remains relatively constant.

    Aside from laser #17, there is really no consistent difference in average optical frequency based on laser type and if anything, it goes the wrong way! And note the change resulting from the swap of the beam sampler. Beam Sampler 1 was originally on Laser 2 and was resulting in the optical frequency dancing around, then swapped with Beam Sampler 2 which resulted in a large frequency offset, then with a third Beam Sampler which was finally well behaved and now remains in that laser. I have no reason to suspect anything is wrong with either Beam Sampler 1 or 2 and did test them for basic functionality with a voltage source and polarizer. All beam sampler assemblies I've checked regardless of what laser they came from have exactly the same part number though it's possible that the optics inside differ in some subtle way depending on laser type. There are at least two versions of the housing - one with a small aperture for 6 mm optics and another with a large aperture for 9 mm optics, but beyond that I don't know of any differences. Lasers 3 through 7 definitely have their original beam samplers. Though I don't have minimum specs for the 5517E, Laser 4 is probably relatively high mileage. And lasers 5 and 9 are high mileage as evidenced by their low (below spec) output power. Even though the output power of Lasers 6, 7, and 8 is well within spec, they are also definitely high mileage lasers being extremely slow start and unable to run on the normal 3.5 mA discharge current. This in itself shouldn't have a large effect on optical frequency unless the tube actually runs hotter (in which case the optical frequency should increase, more below). But the lock point temperature adjustment has not been changed on these lasers, so the equilibrium bore temperature should be similar to that of the others though the equilibrium laser tube envelope and laser temperature will be slightly higher.

    So, the actual optical frequency may be dominated by the amount of use (number of hours on the tube) which also tends to correlate with a decline in output power. Note that all HP/Agilent laser tubes are hard-sealed and thus the calender age (with very rare exceptions) is essentially irrelevant. For example, a laser run 2 hours a week for 10 years (with 2 weeks off for vacation each year) is really only a 1,250 hour (1.7 month) laser. The usage may overwhelm any real or fictitious optical frequency offset found in the specifications. While I don't know what the original output power was for most of these lasers, those with 400 µW or more start very quickly or instantly and are likely relatively young (usage-wise). Laser 1 is known to have been taken out of service due to a bad LCD in the beam sampler, so it could have seen relatively little use.

    There has been research showing that the neon gain center frequency tends to decline with use due to a drop in tube pressure and other factors. Helium has an effect on lasing center frequency of about +22 MHz/Torr, so a loss of He due to gas entrapment on the tube walls or cathode, leading to a drop in its partial pressure, can easily account for these large frequency differences. (Loss of He due to diffusion through the tube walls would also result in a decline in its partial pressure, but this loss mechanism should be minimal.) Major factors include:

         Cause              Sensitivity       Comments    
     Helium Pressure       +22 MHz/Torr       Pressure of He decreases with use
     Neon Pressure         -25 MHz/Torr       Pressure of Ne decreases with use
     Neon Isotopic Ratio   +10 MHz/% of 22Ne  Ratio of 22Ne:20Ne Decreases with use
     Temperature          +280 kHz/°C         Affected by specific lock point
     Magnetic field        ??? kHz/gauss      Affected by mode / gain symmetry

    Both He and Ne partial pressures descrease over the life of the tube but because the fill ratio is between 5:1 to 9:1 of He:Ne, the decrease in He pressure dominates and a frequency drift downward of several MHz/year is quite reasonable. If not filled with a pure Ne isotope, the Ne isotope ratio also will change slightly as the 22Ne will be trapped at a slightly a higher rate than 20Ne. Note the strong dependence on the Ne isotope ratio, a 1 GHz range! So, just over a 1 percent change in the ratio at the time of manufacture could account for the 12 MHz difference in nominal frequency specifications. Natural neon contains approximately 9.25% 22Ne. And, for any given measurement, there is uncertainty in the actual lock point as the laser warms up but that's probably only a maximum of ±1 MHz or so. The 280 kHz/°C is for a tube about 8-1/2 inches long - similar in length to most of the HP/Agilent tubes. However, note that for the HP/Agilent lasers, the temperature of the tube envelope is not controlled, only the mirror spacing rod for the 5501B and 5517s, and not at all for the 5501A. So, the temperature of the mirror spacing rod may have a significant impact on the optical frequency. This differs from most other stabilized HeNe lasers where the exterior of the tube is wrapped in a heater.

    Except for the effect of magnetic field (more below), this has been distilled from the paper: "Frequency stability measurements on polarization-stabilized He-Ne lasers", T. M. Niebauer, James E. Faller, H. M. Godwin, John L. Hall, and R. L. Barger, Applied Optics, vol. 27, no. 7, 1 April 1988, pp. 1285-1289. However, a later paper states the contribution from the Ne isotope ratio as being 8.75 MHz rather than 10 MHz per percent of 22Ne.

    So, it's quite possible that any differences in the optical frequency of these lasers when they were new is totally swamped by changes due to use. For example, if a laser has been run 24/7 for 3 years (middle age for these lasers!), its optical frequency could have gone down by 10 to 15 MHz due to the decline in gas pressure and isotope ratio changes. But the ultimate conclusions may be that (1) it's not worthwhile to assume anything about the nominal optical frequency on used HP/Agilent lasers, but if the optical frequency can be measured (or compared to that of a new laser), (2) the frequency shift may be a means of estimating how many hours or years they've been on! :-)

    Another effect that I have observed in testing with a variable magnetic field is that the intensity of the magnetic field may also shift both optical frequency components (in addition to changing the split frequency). Part of the effect is instantaneous, but the mode balance also changes and the feedback readjusts it. Tests so far have been done with (1) a new 5519B Keysight tube installed in a 5517A laser body with HP analog control PCB and (2) a new non-HP/Agilent/Keysight tube installed in a small 5517 laser body with analog HP control PCB, and my variable field electro-magnet. In both cases, there was an instantaneous change in optical frequency of a few MHz when the field was changed by 40 or 50 gauss but this then relaxed back to a stable equilibrium after a few seconds as the feedback loop readjusted the heater current. For the 5519B, the final equilibrium point didn't change significantly. For the other tube, the equilibrium point changed by about +150 kHz/gauss. 150 kHz is about 1/20th of the gain curve splitting sensitivity of 2.8 MHz/gauss, but many times the sensitivity of the split frequency versus field strength which may be order of 4 to 8 kHz/gauss. Using the value of 150 kHz/gauss and the same tube with a 250 gauss field for a 5517C or a 375 gauss field for a 5517D, optical frequency could differ by 18.7 MHz. I have never seen such a large difference. It is not known yet if the Keysight tube is immune, or just a coincidence. Since more than one variable was different (specific tube, beam sampler with its LCD polarization rotator, and control PCB), no conclusions can be drawn at this time.

    As a side note, Murphy must have taken a day off because the absolute optical frequencies of the two tubes differed by less than 10 MHz without specific control of any variable except that Ne isotope ratio for the gas-fill was selected to be what was though to be similar to what was used in the genuine tubes. By adjusting the field strength, it was trivial to achieve an average difference of 0 Hz (except for a 0.5 to 2 MHz low frequency variation resulting from the mediocre feedback scheme used on the control PCBs of both lasers). This unlike a previous test of a non-HP/Agilent/Keysight tube where the absolute difference optical frequency was around 75 MHz.

    There are at least two possible explanations for this behavior:

    1. The neon gain curve is asymmetric: As the split gain curves move apart, their effect on mode pulling would differ, resulting in a net change in split mode position. With pure 20Ne or 22Ne, the gain curve should be symmetric. But if it's a mixture, then this could apply.

    2. The lock point is offset from the split gain center: If the lock point isn't at the exact center of where the split gain curves intersect and/or the relative amplitudes of the two feedback signals is not identical, the effect of spreading the gain curves would also cause a net shift in optical frequency. The lock point in all HP/Agilent/Keysight lasers depends on a beam sampler that has been designed to equalize the amplitude of the H and V polarized components and on signal paths for each with gain determined by resistors with a 1% tolerance. There are no adjustments. Just considering the signal paths, if gain was 1% high on one side and 1% low on the other, some movement would be possible. But it could be worse. And with the high sensitivity to imbalance, it may not require imbalance.

    Despite all these potential sources of variability, for an application requiring an accurate stable optical frequency reference like calibration of a wavemeter, a reasonably healthy 5517 laser (any version) is probably a better choice than a laboratory stabilized HeNe laser like a Spectra-Physics 117A. The reason is that the design of the 5517 inherently locks to a balanced mode state (or equivalently, the center of the neon gain curve if no magnetic field were present), with no adjustments and little in the electronics to drift with age to change this. Lasers like the SP-117A have separate photodiodes and pre-amps for the two mode signals as well internal adjustments that can affect the lock point. Furthermore, the optical frequency specifications of all HP/Agilent lasers are known (even if there is an unexplained discrepancy of 14 MHz going from the 5517B to 5517C). This is not the case for many other stabilized HeNe lasers. And, if needed, the laser can easily be packed up and sent to NIST or elsewhere to have its optical frequency measured precisely without fear of it changing either from a few bumps during shipment, or over time if turned on periodically rather being run 24/7. I wouldn't recommend other HP lasers like the 5501B simply because healthy ones are becoming harder and harder to find. And the 5501A uses a different locking design which is similar to that of the other (non-HP/Agilent) lasers. However, a healthy 5518A or 5519A/B would also be suitable, using the same design as the 5517.

    Having said all that, how much does it really matter? The worse case uncertainty in optical frequency when locked based on reasonable assumptions about the likely range of isotope ratio and other factors is less than 0.5 GHz. When converted to wavelength, this represents a a 1 ppm error in dimensional measurement, similar to a 1 °C uncompensated error in temperature or 2.5 mm of Hg uncompensated error in pressure. Thus if the actual optical frequency is spec'd or measured, adding an offset to the environmental compensation would take care of it.

    Also see the section: Comparing the Optical Frequencies.

    Obtaining an Interferometry Laser for Experimentation

    While most people on this planet would consider an HP/Agilent laser most useful as a door-stop, there are reasons to want one other than for use in your private semiconductor foundry or custom optics diamond turning machine. :) In addition to their intended metrology applications, these lasers are also ideal as laboratory stabilized HeNe lasers and optical reference lasers. They could certainly substitute for lasers like the Coherent 200 and Spectra-Physics 117/A if one of the frequency components is blocked with a polarizer and the generally lower output power is adequate. And without selecting one of the frequency components, they are close enough together that the only effect is to reduce the coherence length from many km to perhaps *only* 100 meters! Since the optical wavelength/frequency of the HP/Agilent 5517 laser (all versions) is known and very stable over time, they can be used for wavemeter, spectrometer, and optical spectrum analyzer calibration. In this respect, they are far superior to common laboratory stabilized HeNe lasers, whose optical frequency is rarely even specified, nor controlled as it is in the HP/Agilent lasers. If ultimate accuracy is needed, the laser can be heterodyned with an iodine-stabilized HeNe laser to determine its precise optical frequency - and that can be written on the cover. (In case you happen to have one sitting around.) Since the HP/Agilent lasers do not drift much even over years if not run 24/7, once calibrated, such a laser will stand in for the even more expensive iodine-stabilized laser without being tied to one. The split frequency is also handy for testing ultra-high finesse interferometers (another thing everyone needs!) And, it would also be possible to use one of these lasers as the basis for an super-sensitive seismometer or the ultimate laser microphone or a LIGO simulation. :)

    So, if you're salivating for an HP/Agilent laser and can't live without one, a new 5517 costs somewhere between $8,000 and $12,000 depending on options! They are orderable but may have a several week lead time. (If you want some version of the flagship Z4203, the cost is probably at least double, but those are not easily adaptable for general purpose applications.) Used (or "previously owned" - which would be classier!) HP/Agilent lasers can be had much cheaper but caveat emptor. The only ones most people can afford for personal use would be found on eBay. But most of these lasers that end up being resold are taken out of service because they have an end-of-life tube. Interferometry lasers used for metrology are often run 24/7 from the day they are installed until they die. Even though the lifetime of the special HeNe laser tubes used in these lasers may be 50,000 hours, that's still only about 6-1/4 years. And guess where they then end up? :) If the seller hasn't powered the laser head (or doesn't admit to it) and lists the laser "as-is" with no returns, chances are excellent that it will serve as a nice doorstop but not much else, at least not without some effort. Unfortunately, except for the 5519A which plugs into a standard wall socket, these lasers require ±15 VDC for power with a Military-style connector that you won't find at Radio Shack (remember them?). It's easy to "hot wire" power from inside, but unless the seller is familiar with this sort of thing or has the mating power supply and cable, it may be better to just get a DOA warranty in writing and accept that you may have to pay shipping both ways if the laser is only good as a doorstop.

    These lasers also show up at surplus dealers but they tend to ask higher prices than would be considered acceptable for basic tinkering and many seem content to simply have the laser gathering dust than to let it go for a realistic price if untested. But I've also heard of at least one instance where such a laser was found at a garage sale. That price was almost certainly right!

    Also, note that when looking for lasers like this on eBay or elsewhere, the clothes these lasers wear are of little importance. Newer Agilent OEM 5517 lasers (which are mostly what show up now as late model surplus) tend to have a thin cheaply made gold-ish (alodined) or bare aluminum shroud rather than the beige or gray two-piece case of most older HP lasers. It has a rubber-ish gasket to somewhat seal it but this really but this only makes reassembly a royal pain. The gasket meterial is also conductive and serves to ground the cover, though I've never seen a situation where this either mattered. Perhaps it was added to satsify some RFI Standards requirement and might make a difference in a $100,000,000.00 Fab. (The gasket is easily removed if desired and the laser will work just fine without it.)

    What's inside is pretty much the same, though lasers built after around 2003 will likely have the Type II Control PCB with mostly SMT components rather than the older Type I Control PCB with all through-hole components, which may cause issues with locking when using the Short Tube. And in the trivial triviality department, Agilent's only concession to style seems to be in the color of the front and back plates: Beige for 5517Bs, silver for 5517Cs, and gold for 5517Ds! :) But this is only true of some samples and there doesn't appear to be any way to predict which ones.

    And for the most part, calender age doesn't matter. For lasers with the Type I Control PCB - even ones 30 years old, the only parts that go bad are a few electrolytic capacitors, easily replaced for a buck or two.

    Regardless of who is selling the laser, if they are able to power it, the three most important things to ask of them would be:

    1. Is there a good stable output beam that doesn't sputter, flicker, or cut out and restart, or simply cut out at any time over the next few minutes? And if so, how long does the beam take to appear?

      For all except the 5501B, "Yes" and "a few seconds or less" means the HeNe laser tube and power supply are probably good and happy working together. A laser that takes awhile to start may still be fully functional, but it can be annoying to wait 10 minutes for a beam, and associated equipment may expect the laser to be ready within a fixed amount of time. However, even a minute or more could still be acceptable. The Agilent specification is 45 seconds so even they accept that even new lasers don't always start instantly. And sometimes, there are remedies to reduce this.

      The 5501B is the one exception where the laser tube isn't turned on until near the end of the locking process. Thus a beam that doesn't appear immediately on the 5501B is a feature, not a bug. :) This was probably done in the design to minimize the DC current consumption during warmup to be backward compatible with the 5501A. My preference, where this isn't an issue, is to bypass the transistor that switches DC power to the HeNe laser power supply brick. From experience with many 5501Bs, when power is finally applied to the tube, the logic state machine on the Control PCB may reset, especially if the tube doesn't start immediately. In the worst case it could end up in an infinite loop. The transistor is next to the fuse near the top-center of the connector PCB. For testing just jumper between its metal tab and the top pin behind the PCB. Solder a wire between the top two pins to make it a permament "improvement". The tube will then be turned on immediately as with all the other lasers, though as noted above, could still take some time to light. But then a 5501B warms up just like a 5517. I've yet to have any complaints about this modification, even when the 5501B is used in commercial applications.

      Asking the seller to do this for a 5501B may be a bit unrealistic, so you will probably just have to take your chances if the beam doesn't appear after a few minutes and stay on.

    2. Does the READY LED come on and stay on (solid, not flashing) in about 10 seconds for the 5501A; around 4 minutes for most of the common 5517s, 5518A, and 5519A/B; and 6 to 9 minutes for the 5501B? (For all except the 5501A, READY will start flashing about half way through the waramup period. This is normal and indicates that the set-point temperature has been reached.)

      A "yes" answer to the READY question alone is usually sufficient to confirm proper operation with usable power for many purposes. However, cold start to READY on solid may be over 10 minutes - even up to 20 minutes for a few lasers like some versions of the 5517E, 5517FL, or 5517GL, and other 5517s using the Type III PCB, but these lasers are almost non-existent surplus. Aside from the really old 5500A/B/C and 5501A which lock in around 10 seconds, nearly all the thermally tuned lasers found on eBay even in 2014 should come ready in the typical 4 minutes (or 6 to 9 minutes for the 5501B). A slightly longer time is of no consequence, but if a standard laser takes several minutes longer, it's probably very low power and requires the extra time for the power to increase enough as the tube warms up for the laser to be convinced there is enough power available or the its internal REF signal is clean enough. However, such a laser could still be useful. It's also possible that someone twiddled an internal adjustment adjustment without knowing what they were doing! "Here's a screw, let's turn it.". ;-)

      However, some types of data processing systems like the HP-5508A Measurement Display will produce a hard error if the laser takes more than 10 minutes to become ready. (Power cycling the 5508A once the laser is ready will generally get around this even if it powers the laser, as it's likely to become ready much quicker once warmed up.)

    3. What is the output power once the READY indicator comes on solid? The HP/Agilent specification for minimum power is 180 µW for most of these lasers except the 5501A/B (and really old 5500A/B/C) and 5517BL, which are 120 µW, and less common lasers like the 5517E/FL, which is lower. Note that the locked output power may be at or near the minimum of the power as it varies due to mode sweep during warmup, especially for higher-REF lasers like the 5517D. So, simply measuring output power before READY is on solid may result in an overly optimistic reading and may even be more than double the locked output power for a high mileage laser.

      For many applications, much less than spec'd minimum power is quite sufficient. Even if the seller is unable to measure the output power, as long as READY comes on solid, it is probably at least 80 µW for all the lasers except the 5501A, which will lock at much lower power - down to 40 µW or less. Even this may be sufficient for a single axis system.

    Where the laser passes these tests, it will probably be more than adequate for an experimental, demo, test, educational, or research system which doesn't have many measurement axes and isn't run continuously for years. However, before considering such a laser for installation in a semiconductor wafer stepper producing next generation multi-core processors, many additional tests would need to be performed to determine its present health and life expectancy. In some installations, the laser is swapped out after a fixed number of hours, like fluorescent lamps! :) While in others, they are replaced at the point where their output power or REF frequency have changed by a certain percentage or is close to failing to meet HP/Agilent specs. Either of these approaches makes sense where the cost of down time is extremely high. So, for example, even though they may start instantly, run reliably, and have decent output power much greater than the HP/Agilent minimum, if their REF frequency is found to be at or above the range for that model laser, they may be flagged for replacement during preventive maintenance. (REF frequency tends to increase with use and is related to the decline in output power.) An example would be a 5517B outputting 500 µW with a REF frequency of 2.5 MHz. (The spec'd REF frequency range of a 5517B is 1.9 to 2.4 MHz.). However, most tools can tolerate a REF frequency way above the spec'd maximum, and certainly for the vast majority of exprimenters, this is totally irrelevant. A laser outputting 500 µW is generally darn healthy. :) And the REF frequency can be reduced relatively easily enabling an otherwise useless laser to function properly, at least for a modest amount of time. This (along with other rescue techniques HP/Agilent/Keysight won't tell you about) are described elsewhere in this chapter.

    (If anyone has an HP/Agilent laser with a record of the output power and REF frequency when new either from measurements, the label, or original paperwork, and what they are now, and if possible, an estimate of how much it has been run, please contact me via the Sci.Electronics.Repair FAQ Email Links Page. This will aid in my attempt to more accurately estimate previous use and life expectancy for these lasers.)

    Even if the laser plays dead, it could just be a bad HeNe laser power supply brick, or something else that's easily and (relatively) inexpensively repaired. Or, it could be *really* slow start. And on really old lasers, there's a service switch inside - someone may have left it in the wrong position! :)

    For details, see the section: Common Problems with HP/Agilent 5517 Lasers.

    And even if the HP/Agilent laser tube is certifiably dead, it is possible to install an inexpensive barcode scanner tube in its place that results in a usable system, at least for experimentation or demos. This isn't for the casual user, but if you're up to a modest challenge and have some basic mechanical, electronic, and optical skills, see the section: Installing a Common HeNe Laser Tube in an HP-5517 or 5501B. You can then say you built a $10,000 laser for $3.87. :)

    For more information on alternatives to purchasing new HP/Agilent lasers and critical issues in their selection and testing, see the companion document: Considerations in Evaluating Used or Rebuilt Hewlett Packard/Agilent Metrology Lasers.

    If you're intending to play with the laser for awhile and then try to sell it, get a 5517A/B/C. The market for these is pretty good. Some companies seem willing to practically buy them off the back of a truck if the operating condition is known to be good. :) But attempting to sell 5517Ds (and above) is more difficult - at least it takes longer to find a buyer - since they tend to be used in demanding applications and the risk of installing a laser of questionable functionality and future life is higher. For a hobbyist/experimenter, the model doesn't much matter. Go by whether it locks with decent power and cost. If a working 5517D turns up at an attractive price, that's just fine. I wouldn't recommend the 5517E/FL/GL and their variations though simply due to the generally much lower output power and higher REF frequency, both of which are more difficult to deal with and reselling them at a decent price is next to impossible.

    And a note on shipping: While the lasers with the Long Tube are not really particularly fragile as lasers go, a sharp physical shock can misalign the laser tube internally resulting in the power declining dramatically, possibly to zero. (Most common lasers would be destroyed by this treatment.) Careful tapping with a wood block may get at least some of the power back, but this is for the advanced course. :)

    While lasers with the Short Tube are still more robust than the typical common HeNe, since the mirror spacing rod is mounted only at one end, it can literally break off at that point. So, impress upon the seller the importance of careful packing with at least 2 inches of non-collapsible padding on all sides in a sturdy oversize box such that nothing (including a corner) can contact a hard surface even if dropped from 6 feet.

    And, if you do come across one of these lasers at a garage sale, just splurge, take the risk, and pay the $2 asking price. :) Or if it makes you feel better, haggle it down to $1. :-)

    Maximizing the Life of HP/Agilent Metrology Lasers

    Although the spec'd lifetime of the special tubes used in these lasers may be around 50,000 hours or slightly over 6 years, with a sticker price of something like $10,000, it still pays to do everything reasonable to maximize their usable life. (And, there is some evidence that newer lasers typically run only 3 to 4 years before going outside Agilent specifications.) Many if not most semiconductor fabs and other similar production facilities keep these lasers powered continuously regardless of whether they are actually being used. This may be based on the belief that (1) they take a long time to reach the required wavelength accuracy, (2) stuff tends to fail when powered cycled, and (3) HeNe lasers like to be run. None of these is particularly valid for these lasers. But even so, keeping them on 24/7 still may be most cost effective where minimizing any unscheduled downtime is more critical than replacement costs. But this is only valid if the laser is replaced during as part of normal preventive maintenance (based on operating hours or power decline) prior to failure. Unscheduled down-time of several hours to a day or more can be costly. However, for educational institutions and general research and development applications, there is no reason keep the laser on 24/7 if not being used. Power up the laser an hour or so before the system is needed and power it off when not needed for several hours or more. Here some comments on each issue:

    So, in short, the laser itself won't function any better if run continuously compared to being turned on at most 90 minutes before needed as long as it doesn't affect the environment in such a way as to change the calibration. (90 minutes is HP/Agilent's spec for warmup to full accuracy on an unvented laser, only 45 minutes on one with forced air cooling). And for less critical applications, simply waiting until READY comes on solid may be adequate. It should be possible to test for the overall effect by making a measurements of a known length in each axis when the laser comes READY, after 90 (or 45) minutes, and after 24 hours. If any differences found are acceptable, there is nothing to be gained by continuous operation.

    Where the laser might be used for a few hours a week, as in a diamond turning machine at a custom optics house, this should effectively extend the life of the laser to infinity.

    A switch can even be added for ONLY the 15 VDC to the HeNe laser power supply brick, leaving the control electronics running continuously. Return to service would then only be a couple minutes and the thermal environment would not change as much. And to be totally anal, a heater could be switched on in place of the tube itself. :)

    Installing Newer HP/Agilent Metrology Lasers in Older Systems

    As time passes, locating functional 5501A and to a somewhat lesser extent, 5501B lasers to repair an older system is becoming increasing difficult. While it's sometimes possible to find a laser from 1981 that has a like-new tube, the vast majority of them are good for little more than doorstops or high tech sculpture. Agilent will not support any of these older lasers.

    (The following deals with retrofitting systems using 5501A or 5501B lasers. For really old systems using 5500A/B/C lasers, a few more issues are present since the 5505A Measurement Display is more tightly coupled to the laser and somewhat more is involved to keep it happy.)

    Replacing 5501B laser with 5517 laser:

    The preferred approach is to install a more modern 5517 laser in place of a 5501A or 5501B. 5517 lasers are still in production and used working units are also readily available at very reasonable cost. This may require no modifications to the 5517 laser so if a replacement is required at a later time, it can be a drop-in.

    Only a few relatively minor differences need to be accommodated to substitute a 5517B for a 5501A or 5501B. With a bit of resourcefulness, the total cost (excluding the laser and labor) for this type of conversion will be under $100 in most cases:

    Installing 5517 tube in 5501B body:

    If the electronics in the 5501B are in good condition or can be repaired, it's also possible to install a new or used healthy 5517B tube assembly. This is simpler from the point of view of the user and preserves the external appearance of the Tool so Field Service doesn't get upset about unauthorized modifications. :) (A 5517C or 5517D could also be used if the higher REF frequency isn't an issue.) Physically and electrically, 5517B/C/D tubes are drop-in replacements for the 5501B as long as the beam diameter is the same. (Else, the beam expander will need to be replaced.) So that leaves F1/F2 orientation, REF frequency, and the temperature set-point adjustment:

    Note that even if everything is done perfectly a couple of factors may conspire to slightly degrade the REF/MEAS signal quality compared to that of a newer 5517:

    1. HeNe laser power supply ripple: The HeNe laser power supplies used in older HP lasers have enough current ripple to noticeably affect the optical signal. Current ripple gets translated into optical amplitude ripple, which will be picked up by a sensitive optical receiver, especially when the output power is relatively low. This will appear at 1 or 2 times the switching frequency of the chopper inside the power supply brick. There may also be magnetic coupling to the tube through the air from the power supply's high frequency transformer sitting inside the brick mounted underneath the tube. This coupling is minimized by an aluminum shielding plate screwed above the power supply brick on all newer lasers. Transferring the HeNe power supply brick and shield from a newer (sometime after the year 2000) laser will help.

    2. Heater current ripple: The 5501B is the only HP laser to use Pulse Width Modulation (PWM) to control the heater current, and the PWM switching frequency can also couple into the optical output. (It's not know whether the mechanism is thermal, mechanical, magnetic, or all three!) Later HP lasers use linear control so maybe HP learned their lesson. ;-) This issue is thus unique to the 5501B and may make the signal quality lower with a 5517 tube installed in a 5501B body than in a 5517 body. For the obsessive-compulsive, a PWM to linear converter can be added.

    However, neither of these is likely affect system performance in any detectable way, only the appearance of the signals on an oscilloscope. So it's probably not worth losing too much sleep over them.

    Modifying 5517 laser:

    Where the 5501B electronics are faulty, a usable 5501B is not available, or it is desired to upgrade to a 5517 laser but the F1/F2 orientation must match that of the 5501B, there is a hybrid approach that will also work. Two additional things need to be done beyond what's required to use a 5517 laser without this change:

    And doing any or all of this will void the warranty. :) But the result will be functionally indistinguishable from an original 5501B laser except that the peak current requirements on the +15 VDC power supply are higher - 2.5 A maximum.

    Installing Older HP/Agilent Metrology Lasers in Newer Systems

    It's much less likely that this would ever be required unless testing 5501A/B lasers using the electronics for 5517 or similar lasers. Or, you found a stash of new/NOS 5501Bs and it would be a darn shame not to use them. :) The same considerations as above apply - in reverse, which makes the power a non-issue and there is still the F1/F2 rotation and the lower REF frequency of 5501A/B lasers could be a limitation. But, apparently, there is an HP adapter cable for this purpose: ET-41915 "Laser Head Cable Adapter, 18-pin to 4-Pin Power and 4-Pin Signal". The cable part numbers are:

    To emphasize: This is NOT what is needed to use 5517 lasers in place of 5501A/Bs - it goes the other way! See the previous section.

    Repair of HP/Agilent Metrology Lasers

    The custom tubes at the heart of these lasers are what will dominate the cost of repair in most cases. Electronics problems are relatively uncommon, and generally minor when they do occur. Where the tube is found to be unusable, a replacement tube will be needed. New tubes are only available from Agilent and I wouldn't even want to think about their cost but figure 75 percent or more of that of a new laser. And they probably wouldn't even sell a tube without insisting on doing the installation as well (at an additional charge). One company that refurbishes tubes for precision HeNe lasers is Evergreen Laser Corporation. But I do not know if they will deal with HP/Agilent tubes, and rather doubt it due to the additional issues of gaining access to the glass tube itself buried inside the magnet assembly. And even if they will, the cost of a refurbished tube would probably exceed the cost of a complete guaranteed full spec laser on eBay! Another company I contacted who does repair these lasers said candidly that their success in refurbing tubes was spotty at best - and I think that was being overly generous. One company that does seem to claim a complete overhaul capability is Hong Ren Technology Co., Ltd. but I have no idea of their reliability. But keep in mind that for any of these, you're talking thousands of dollars, not garage sale or even eBay prices! From photos on their Web site showing a non-standard tube mount, it appears as though they may actually build their own tubes or provide tubes that are not from Agilent. Non-genuine tubes could have subtle differences in beam profile, wavelength, stability, lifetime, and may have rogue modes that will affect measurement performance.

    Examples of commercial rebuilds I've come across over the years include Rebuilt HP/Agilent/Keysight Laser Tube Assembly 1, Rebuilt HP/Agilent/Keysight Laser Tube Assembly 2, Rebuilt HP/Agilent/Keysight Tube Assembly 3, and Rebuilt HP/Agilent/Keysight Laser Tube Assembly 4. These all use a conventional HeNe laser tube and only differ in the finer details, mainly in the physical mounting and output optics. For more on all this, see the companion document: Considerations in Evaluating Used or Rebuilt Hewlett Packard/Agilent Metrology Lasers.

    However, even a tube deemed to be dead by Agilent due to low power or an inability to stay lit, may often be made usable for many applications (especially where only 1 or 2 measurement axes are required), for a test or educational system, or as an emergency spare, with at most some relatively minor low cost modifications to the laser, or possibly even simply an adjustment. But if the output power is so low that the beam actually disappears periodically while warming up, there won't even be a beat signal and such a tube is only good as a high tech paperweight with built-in magnetic paper clip holder. :)

    Assuming the tube is usable, except for Agilent 5517 lasers based on Type II or Type III Control PCBs, all of these lasers are very serviceable as far as the electronics are concerned. Pre-2004 lasers - many of what's found surplus even in 2013 - will almost certainly have the older Type I Control PCBs. (This is definitely true for pre-2000 lasers.) For these, even most of the HP house-numbered ICs have standard equivalents available from major electronics distributors, and none of the other electronic parts are special. Operation and service manuals are available which include detailed adjustment and troubleshooting information and complete schematics. And parts units can be obtained on eBay at low cost. Except for a blown fuse of my own doing, dried up electrolytic capacitors on really old lasers, a blown line driver chip, and bad REF photodiode, I've yet to see an Type I Control PCB with any serious problems including defective proprietary HP ICs. However, on a 5501B, the heater driver transistors and main fuses were blown as a result of dried up electronic capacitors on the Connector PCB. So, for 5501B lasers, it's probably good preventive maintenence to replace all 4 electrolytic capacitors on the Connector PCB and the 2 electrolytic capacitors on the Control PCB on a laser more than 10 or 15 years old as a precaution. (The third leg on the 2 silver and 2 blue electrolytic capacitors on the Connector PCB is for mechanical support only. They can be replaced with common caps of at least equal µF and voltage ratings. I prefer to use 105 °C types but that's probably not essential.) This is the only situation I know of in HP/Agilent where a high ESR/low uF capacitor will result in actual damage to other components. For more on the 5517 laser in general and the Type II PCBs in particular, see the section: HP/Agilent 5517 Laser Construction.

    Now, if you're independently wealthy and would like to have Agilent repair your laser, I've heard that an evaluation is about $500. Essentially, they confirm that it's an HP or Agilent laser and then tell you how much it will cost to repair, if they are willing to repair it at all. For a single failure, the cost is a flat rate between $1,500 and $2,000, but the evaluation fee will be applied toward that, thank goodness. :) A "single failure" probably includes a blown fuse, broken resistor, dried up capacitor, or bad IC. I don't know whether something like a degraded LCD in the beam sampler or blown HeNe laser power supply would qualify as a single failure, or if two dried up capacitors would be charged (no pun....) for separately. And, it's almost certain that if you read the fine print, the flat rate would exclude a weak or dead HeNe laser tube that required replacing the tube assembly even though it is technically a "single failure". In that case Agilent would simply return the laser after collecting their evaluation fee.

    For amusement, go to Find-A-Part: Keysight's Test and Measurement Parts Catalog and enter a laser model like "5517D". If you're not independently wealthy, you better be sitting down when viewing the prices. For example, (in 2009) the cover is $344, the Type II Control PCB is $1075 (not known which version), the HeNe laser power supply is $496, and a small screw is a bargain at $1.24 each. However, prices for the operation and service manuals are not totally ridiculous - $28.44 for the 5517A and $42.67 for the 5517B/C. But the exact parts available for each model laser seem to be somewhat random and forget about even being able to order a new tube assembly (or parts). They are listed as: "Not orderable, contact Agilent for repair service". Right. :-)

    And before doing something silly, getting inside HP/Agilent lasers is trivial. On the large lasers (5517A, 5518A, 5518A/B) it's just a matter of removing the 4 tiny screws on top and gently levering up the cover using a knife blade. On the small lasers (5501A/B, 5517B/C/D), rotate the front turret so the large hole is at the bottom. That will expose a slotted head screw - a 1/4 turn fastener. Push in and rotate 1/4 turn counter-clockwise and the front plate will pop off. The covers or shroud can then be removed. The only reason I've gone to this level of detail is that I had an academic type ask me if that screw was for tuning the laser frequency! :)

    Also see the section: Common Problems with HP/Agilent 5517 Lasers (which applies to other lasers like the 5501B as well). For operation and service manuals, see the section: Additional HP/Agilent Resources.

    Agilent Laser and Optics User's Manual

    This manual provides information on the lasers and optics (no surprise, huh?) aspects of Agilent (now Keysight, formerly Hewlett Packard) metrology systems based on two-frequency interferomters. Unfortunately in their obsession to justify their existance, the Keysight Web designers have removed these links, resulting in landing on a general Keysight "Knowledge Center Page". But don't despair, I have backup copies of both volumes at Sam's Backup of Agilent Laser and Optics User's Manual Volume 1 and Sam's Backup of Agilent Laser and Optics User's Manual Volume 2.

    So the following are what's in the manuals:

    Hewlett Packard/Agilent Part Number Reference

    Here is a partial list of HP/Agilent optics, electronics, and accessories including both current and obsolete products listed by part or model number:

     Part#        Description
     1000-0598    0.5 OD (31.6%) Neutral Density Filter (5518A/5519A/B test)
     05500-60025  5500A/B/C to 5505A cable
     05505-60048  Rack Mount Kit for 5505A
     05508-60021  Remote Control Unit (5528A)
     05517-60033  Differential to Single Ended REF Signal Breakout Box
     05518-60308  Replacement/upgrade turret assembly for 5518A
      5500A       Laser Transducer (w/interferometer, optical receiver, 0.4 m/s)
      5500B       Laser Transducer (w/interferometer, optical receiver, 0.4 m/s)
      5500C       Laser Transducer (w/optical receiver, 0.4 m/s)
      5060-0049   Extender Board, 15 pin
      5060-0630   Extender Board, 22 pin
      5501A       Laser Transducer (0.4 m/s)
      5501B       Laser Transducer (0.4 m/s)
      5505A       Measurement Display (5526A)
       K01-5505A   Extender Board (XA-14), 52 pin
      5507A       Electronics
      5508A       Measurement Display (5528A)
       5508-60020A Laser Interferometer Cable to Remote Control PCB?
      5510A       Automatic Compensator (5525A/5526A)
       H01-5510A   High Accuracy  Automatic Compensator (5525A/5526A)
       K15-5510A   Multiplexer for 5510A
      5517A       Laser Transducer (0.4 m/s)
      5517B/BL    Laser Transducer (0.5 m/s)
      5517C       Laser Transducer (0.7 m/s)
      5517D       Laser Transducer (1.0 m/s)
      5517DL      Laser Transducer (1.1 m/s)
      5517E       Laser Transducer (1.6 m/s)
      5517EL      Laser Transducer (1.77 m/s)
      5517F       Laser Transducer (1.7 m/s)
      5517FL      Laser Transducer (2.15 m/s)
      5517G/GL    Laser Transducer (2.2 m/s)
      5518A       Laser Transducer (w/optical receiver, 0.4 m/s), <SN2532A02139)
      5518A       Laser Transducer (w/optical receiver, 0.453 m/s, >=SN2532A02139)
      5519A       Laser Transducer (w/optical receiver, 0.7 m/s)
      5519B       Laser Transducer (w/optical receiver, 1.0 m/s) 
      5525A       Laser Measurement System
       K02-5525A   Resolution Extender
      5526A       Laser Measurement System
      5527A/B     Laser Position Transducer System
      5528A       Laser Measurement System
      5529A       Dynamic Calibrator
      5530        Dynamic Calibrator - Base System
      9211-1586   Transit Case for 5500A/B/C
      9211-1587   Transit Case for 5505A
      9211-1738   Transit Case for 5510A
     10550A       Retroreflector
     10550B       Retroreflector (includes retroreflector mount)
     10551A       Plane Mirror Convertor
     10552A       Resolution Extender
     10555A       Remote Interferometer
     10556A       Retroreflector (4 screw square mount)
     10557A       Turning Mirror
     10558A       Beam Bender
     10559A       Reflector Mount (Dual 4 screw square mount)
     10560A       Barometer
     10562A       Single Beam Interferometer
     10563A       Material Temperature Sensor
      H01-10563A   High Accuracy Material Temperature Sensor
     10564A       Air Temperature Sensor
     10565A       Remote Interferometer
     K03-10565A   Single Beam Inteerferometer
     K08-10565A   10" Non-Contacting Converter???
     10565B       Remote Interferometer with Retroreflector
     10567A       Dual Beam Splitter (50 precent)
     10579A       Straightness Adapter (Resolution Extender And Optics)
      10579-60001  Straightness Adapter Optics (One beam to two beam)
      10579-60004  Resolution Extender (Electronics)
     10580A       Laser Tripod (5500C)
     10581A       Plane Mirror Converter (5526A, 4 screw square mount)
     10585A       Metrology Program Package (5526A)
     10690A       Short Range Straightness Interferometer and Reflector
      10690-60001  Short Range Straightness Interferometer 
      10690-60002  Short Range Straightness Reflector
     10691A       Long Range Straightness Interferometer and Reflector
      10691-60001  Long Range Straightness Interferometer 
      10691-60002  Long Range Straightness Reflector
     10692A       Penta-Prism
     10692B       Optical Square
     10693A       Vertical Straightness Adapter
     10700A       33% Beam splitter
     10700B       4% Beam splitter
     10700C       15% Beam splitter
     10701A       50% Beam splitter
     10702-60001  Magnetic Alignment Target (two holes, 1/2" separation).
     10702A       Linear Interferometer
      K97-59995A   10702A with 2 low profile QWPs. May be custom.
     10703A       Linear Retroreflector
     10704A       Single Beam Retroreflector
     10705A       Single Beam Interferometer
      10705A-080   Fiber Optic Receiver Adapter
      C01-10705A   Plane mirror/specular reflective surface option
     10706-60001  Alignment aid (magnetic aperture) for 10716A
     10706-60202  Alignment aid (QWP reflector) for 10716A
     10706A       Plane Mirror Interferometer (PMI)
      10706A-080   Fiber Optic Receiver Adapter
     10706B       High Stability Plane Mirror Interferometer (PMI)
     10707A       Beam Bender
     10708A       Power Supply (May Not Apply)
     10710A/B     Adjustable Base (Small, Beam Bender, etc.)
     10711A/B     Adjustable Base (Large, Linear Interferometer, etc.)
     10713B       1 Inch Cube Corner (11.4 g, for use with 10702A or 10705A)
     10713C       1/2 Inch Cube Corner (1.4 g, for use with 10705A)
     10713D       1/4 Inch Cube Corner (0.2 g, for use with 10705A)
     10714A       Display Interface
     10715A       Differential Interferometer (DI)
      10715A-001   DI (turned configuration)
     10715C       Differential Interferometer (DI, improved non-linearity)
     10716A       High Resolution Plane Mirror Interferometer (PMI)
      10716A-001   High Resolution PMI (turned configuration)
     10717A       Wavelength Tracker
     10717C       Wavelength Tracker (0.5 nm non-linearity)
     10719A       One-Axis Differential Interferometer (DI)
      10719A-C02   One-Axis DI (low thermal drift)
     10721A       Two Axis Differential Interferometer
      10721A-C02   Two Axis DI (low thermal drift)
     10722A       Plane Mirror Converter (5501A)
     10723A       High Stability Adapter
     10724A       Plane Mirror Reflector
     10725A       50% Bare Beam Splitter
     10725B       4% Bare Beam Splitter
     10725C       15% Bare Beam Splitter
     10726A       Bare Beam Bender
     10728A       Plane Mirror
     10735A       Three-Axis Interferometer
     10736A       Three-Axis Interferometer
      10736A-001   Three-Axis Interferometer/Beam Bender
     10737L       Compact Three-Axis Interferometer (left configuration)
     10737R       Compact Three-Axis Interferometer (right configuration)
     10740A       Coupler (5501A)
     10741A       Laser Transducer Interface (10740A card)
     10742A       Laser Transducer Counter (10740A card)
     10743A       Extender Board (10740A)
     10744A       Fixturing Kit
     10745A       HP-IB Interface (10740A card)
     10746A       Binary Interface (10740A card)
     10747B       32 Bit Software for 55292A
     10747F       Metrology Applications Software
     10751        Air Sensor (7 pin LEMO)
     10751A/B/C/D Air Sensor (5528A)
     10751-60209  Air Sensor Adapter Cable (10 pin F 5508A to 7 pin LEMO M)
     10753A       Laser Tripod (5518A)
     10753B       Laser Tripod (5519A/B) with Kinematic Mounting Plate
     10755A       Compensation Interface
     10756A       Manual Compensator
     10757A/B/C   Material Temperature Sensor (5528A)
     10757D/E/F   Material Temperature Sensor (5528A)
     10757-60306  Material Sensor Adapter Cable (6 pin F 5508A to 5 pin LEMO M)
     10759A       Foot Spacing Kit
     10760A       Counter (10740A card)
     10761A       Multiplier (10740A card)
     10762A       Comparator (10740A card)
     10763A       English/Metric Output (10740A card)
     10764A/B     Fast Pulse Converter (10740A card)
      10764-60005  Laser Interferometer Cable Assembly
      10764-91009  Laser Interferometer Cable Assembly
     10764C-H05    Laser Interferometer Cable Assembly
     10766A       Linear Interferometer
     10767A       Linear Retroreflector
     10767B       Lightweight Retroreflector
     10768A       Diagonal Measurement Kit
      10768-20214  Base (Large)
     10769A       Turning Mirror
     10769B       Turning Mirror on Universal Mount
     10770A       Angular Interferometer
     10771A       Angular Reflector
      10772-67001  Turning Mirror Mount Assembly
     10772A       Turning Mirror
     10773A       Flatness Mirror
     10774A       Short Range Straightness Optics (interferometer and reflector)
     10775A       Long Range Straightness Optics (interferometer and reflector)
     10776A       Straightness Accessory Kit
      10776-20001  Adapter Plate
      10776-20008  Post
      10776-67001  Striaghtness Retroreflector
      10776-67002  Reflector Mount
      10776-67003  Base Assembly
     10777A       Optical Square
     10778A/B/C   Laser Power Cable (5501A/B, PN 10778-60001)
     10779A/B/C   Reference Cable (5501A/B, PN 10779-60001)
     10780A/B/C   Optical Receiver (free space)
     10780F/U     Optical Receiver (fiber-coupled)
      1251-3452    Mating 4 pin BNC Connector for 10780x
     10781A       Pulse Converter
      10781-60003  Cable assembly
     10782A       Service Kit without Laser Assembly (5501A)
      10782AOP001  Laser Assembly (5501A) only
     10783A       Numeric Display
     10784A       Interferometer Base
     10785A       Height Adjuster and Post
      10785-20005  Post
     10786A       Linear Measurement Transit Case
     10787A       Straightness And Squareness Transit Case
     10888A       Remote Control
     10790A/B/C   Receiver Cable (4 pin BNC plug both ends)
      10790-60001 Receiver Cable (4 pin BNC plug both ends) 4 meter
      10790-60003 Receiver Cable (4 pin BNC plug both ends) ? meter
      10790-60204 Receiver Cable (4 pin BNC plug both ends) 1 meter
     10790A-C10   Laser Interferometer Cable
     10791A/B/C   Laser Head Cable (5517, spade lugs for power, 4 pin BNC REF)
     10793A/B/C   Laser Head Cable (5517A to 5507A and 5518A to 5508A)
                   (One version may be PN 8120-3491 and 10793C may be
                    PN 10793-60203.)
     10793C-C04   10793C with right angle connectors
                   (Also a PN 10793C-60302 which may be length of 10793A.)
     10880A/B/C   Receiver Cable (4 pin BNC to 4 pin LEMO)
     10880A-C04   Receiver Cable (4 pin BNC to 4 pin LEMO, ~1.3 m)
      10880-91015 Receiver Cable (4 pin BNC to 4 pin LEMO, ~1.3 m)
      10880-91000 Receiver Cable (4 pin BNC to 7 pin LEMO ???)
     10881A/B/C   Laser Head Cable (5517, DIN for power, 4 pin LEMO for REF)
     10881D/E/F   Laser Head Cable (5517, spade lugs for power, 4 pin LEMO for REF)
     10882A/B/C   Laser Head Cable (5519A/B To 10887P)
     10883A/B/C   Laser Head Cable (5518A, DIN for power, 7 pin LEMO to 10887A) 
      10883-60202  Laser Head Cable (5518A, DIN for power, 7 pin LEMO to 10887A) 
     10884A       Power Supply (5517 lasers, universal switchmode, DIN)
     10884B       Power Supply (5517 lasers, universal switchmode, DIN)
     10885A       PC Axis Board
     10886A       PC Compensation Board
     10887A/B     PC Calibrator Board (5518A or 5519A/B)
     10887P       PC Programmable Calibrator Board (5519A/B)
      10887-60202  Laser Interferometer Cable (8 pin LEMO to wires)
     10888A       Remote Control (5529A)
     10889A/B     PC Servo Axis Board
     10895A       VME Laser Axis Board
     10896A/B     VME Laser Compensation Board
     10897B       VME Laser Axis Board
     10897A/B/C/D High Resolution VME Laser Axis Board
     10898A/D     VME Dual Laser Axis Board
     10934A       A-Quad-B Axis Control Board
     55280A       Linear Measurement Kit with Case
     55280B       Linear Measurement Kit
     55281A       Angular Optics Kit
     55281B       Linear/Angular Optics Kit
     55282A       Flatness Accessory Kit
     55283A-001   Straight Measurement Kit (Short Range)
     55283A-C01   Straight Measurement Kit (Long Range)
     55290A       Angular Position Measurement Kit
     55290A-744   Supplemental Fixturing Kit
     55290B       Rotary Axis Measurement Kit
     55291A       CNC Upload/Download Sofware
     55292A       USB Expansion Module for 10886A and 10887B boards
     C05-59995A   Reference Cable (5501A/B)
     C07-59995A   Power Cable (5501A/B)
     C08-59995A   Diagnostic Cable (5501A)
     C39-59995A   Laser Head Cable (5517A to 5507A and 5518A to 5508A, 1 meter)
     E1203C       Precision Beam Translator
     E1204C       Precision Horizontal Beam Bender
     E1705A       Fiber Optic Cable
     E1706A       Remote Sensor
     E1207C       Precision Vertical Beam Bender
     E1208C       33% Bare Beam Splitter
     E1208D       40% Bare Beam Splitter
     E1208E       50% Bare Beam Splitter
     E1208F       66% Bare Beam Splitter
     E1208G       60% Bare Beam Splitter
     E1250A/B     High Performance Receiver Cable
     E1251A/B     High Performance Laser Head Cable
     E1705A       Fiber Optic Cable (Normal, Vpin to Vpin)
     E1705B       Fiber Optic Cable (Normal, ST to Vpin)
     E1705C       Fiber Optic Cable (Normal, ST to ST)
     E1705E       Fiber Optic Cable Glass (Lower signal loss, ST to Vpin)
     E1705F       Fiber Optic Cable Glass (Lower signal loss, ST to ST)
     E1706A/C     Remote Sensor (Lend, polarizer, Vpin connector)     
     E1708A       Remote Dynamic Receiver
      E1708A-C05   Remote Dynamic Receiver (Option 1.2 m/s)
     E1709A       Remote High Performance Optical Receiver
     E1713A       Scale Servo Axis Board for E1720A.
     E1720A       Linear Encoder System
     E1734A       Transit Case (5519A, USB modules, sensors, cables, optics)
     E1734B       Transit Case (10753A tripod and accessories)
     E1735A       USB Axis Module
      E1735A-001   A-Quad-B Cable (3 meters)
     E1736A       USB Sensor Hub
     E1737A       Material Sensor with ISO 17035 Calibration
     E1738A       Air Temperature/Pressure/Humidity Sensor/w ISO 17035 Calibration
     E1739A       Sensor Cable (5 m)
     E1739B       Sensor Cable (10 m)
     E1739C       Sensor Cable (15 m)
     E1739D       Sensor Cable (25 m)
     E1826E/F/G   NGI Monolithic One-Axis Plane Mirror Interferometer
     E1827A       NGI Monolithic Two-Axis Vertical Beam Interferometer
     E1833C       15% Bare Beam Splitter
     E1833E       33% Bare Beam Splitter
     E1833G       50% Bare Beam Splitter
     E1833J       67% Bare Beam Splitter
     E1833M       100% Bare Beam Splitter (Beam Bender)
     E1837A       Two-Axis Vertical Beam Interferometer
     E1847A       Laser Head Power Cable (Spade lugs)
     E1848A       Laser Head Power Cable (Male DIN)
     E1848B       Laser Head Power Cable (Female DIN)
     ET-6880      5501 Laser Tester
     ET-41915     5501A/B Laser to 5517 Display/Power Supply Adapter Cable
     ET-319283    5519A/B to 5508A Adapter Cable (~6 inch)
                   (7 pin male LEMO to 18 pin male Laser Head Connector)
     ET-319283-2  5519A/B to 5508A Adapter Cable (~6 inch)
                   (LMS622, 7 pin male LEMO to 18 pin female Laser Head Connector)
     ET-31377-6001-A Laser Interferometer Cable 7 Pin LEMO F PHB.1B to 10 Pin M
                   (5508A Air Sensor to 5530A adapter cable.)
     ET-31378-6001-A Laser Interferometer Cable 5 pin LEMO F PHB.0B to 5 Pin M
                   (5530A Material Sensor to 5508A adapter cable.)
     N1203C       Precision Beam Translator
     N1204C       Precision Horizontal Beam Bender
     N1207C       Precision Vertical Beam Bender
     N1209A       Risley Prism Translator (RPT) Manipulator
     N1211A       Fiber AOM Laser Head (15 to 17 MHz split frequency)
     N1211A-001   RoC Cable, 6.7 m
     N1212A/B     Remote Optical Combiner, 6 mm/9 mm
     N1225A       Four Channel High Resolution Laser Axis Board for VME
     N1225A-200   Non-Linearity Compensation.
     N1231A       PCI Three-Axis Laser Board
     N1231B       PCI Three-Axis Laser Board with External Sampling
     N1250A/B/E/F High Performance Optical Receiver Cable (High performance)
     N1251B       Laser Head Power and REF Cable (High performance with 10884B,
                   female DIN)
     Z4201-60288  Optical Receiver Cable? (4 pin BNC to 4 pin male LEMO)
     Z4205T       ROC Fiber Launcher Adjusting Tool
     Z4379G-A08   Polarizing Beam Splitter (with fiber adapters?)
     Z4399A       NGI Monolithic Three-Axis Interferometer
     Z4420B       NGI Monolithic Five-Axis Interferometer

    Calibration of HP/Agilent/Keysight Metrology Lasers

    HP, then Agilent, and now Keysight provides a service which tests the laser using instruments traceable to the National Institute of Standards and Technology (NIST) to confirm that it meets specifications, and provides a "Cal Certificate" typically good for 1 year. These lasers do not require periodic calibration so typically, nothing is actually adjusted (and there is only one adjustment really possible and that's the temperature set-point). But what exactly do they test. This is the big mystery.

    The Cal Certificate only states that the laser meets specs. There is nothing on it in the way of measurements. Presumably at the very least, they test output power and REF frequency. But what about other parameters like F1/F2 mode balance, mode purity, time to lock, and in particular, absolute optical frequency?

    The paperwork provided with the Cal Certificate does provide a list of the test equipment used in the tests. A typical list for a 5517 laser is:

    One of the counters could be used to measure the REF frequency and the power meter for measuring output power. But what is the 5517B laser there for? At first I thought that perhaps they would beat the laser under test with the 5517B and measure the difference frequency using the other counter determine the optical frequency. But that would require a high speed photodiode (probably with a preamp), and some optical components, none of which are listed. The power meter sensor cannot be used for that purpose.

    As noted the Cal Certificate only states that the laser passed the tests, not what tests were done or what the measured values were. It lists a Web site to go to for more information: Keysight Infoline Service. Entering a laser model and serial number into "Infoline without login" results in nothing except that the laser passed (with a recommendation for the date of the next Cal) and "Modification recommended: Temperature adjustment process", which presumably is for the laser lock temperature set-point. But apparently, they can't be bothered to do that 5 minute procedure without charging extra. :) And, it doesn't appear that this statement means anything wrong was found, as the same recommendation may be found if the serial number of a brand new laser is entered.

    So is the laser's optical frequency measured or compared with that of the 5517B listed in the test equipment? To actually measure it would require an expensive iodine stabilized HeNe laser. Now Agilent may have had one in the past. I have an I2 stabilized laser head with an Agilent inventory sticker! ;-) But a healthy 5517B could be used as an optical frequency reference and do almost as good a job. But I doubt it is done either way for the basic Cal service due if nothing else to the extra time involved, though if they have a test fixture permanently set up, it wouldn't be difficult. However, as a practical matter, from my tests, the optical frequency of these lasers does not change enough to even begin to matter until they are nearly dead, so testing it simply isn't necessary.

    A slight possibility of getting more to the bottom of this is that more information might be forthcoming if one actually registered on the Keysight Web site as the owner of the laser but that is currently not known. This is being persued.

    Tests of Really Lively Agilent Tubes

    Here are some data on a couple of Agilent/Keysight tube assemblies acquired as junk. Both are probably new/NOS or nearly so even if their manufacturing dates are rather old.

    The first is an Agilent Z4203-60244 with a "Short" tube whose rear (HR) mirror alignment has been tuned up for maximum power. (Even before adjustment, it was probably over-spec.) The new spec for the Z4203-60224 is 880 µW (which is the highest power spec for any known HP/Agilent/Keysight tube), but this one is more than 40 percent greater.

    In fact, this is the highest power HP/Agilent/Keysight tube I've ever tested. It starts nearly instantly and the dropout current is 2.25 mA which doesn't change significantly as the tube warms up, indicating that it has seen little or no use. These tests were run after reaching thermal equilibrium at a typical operating temperature with 7.0 VDC on the heater:

       Tube     Tube     Peak
      Current  Voltage   Power
      2.5 mA   1.47 kV  1,250 µW
      3.0 mA   1.48 kV  1,384 µW
      3.5 mA   1.50 kV  1,461 µW
      4.0 mA   1.53 kV  1,533 µW
      4.5 mA   1.55 kV  1,571 µW
      5.0 mA   1.58 kV  1.590 µW
      5.5 mA   1.60 kV  1.596 µW

    Yes, that's almost 1.5 mW even at the spec current of 3.5 mA. (Note that these are peak power during mode sweep, so the locked power would be 10 to 15 percent lower.) The tube voltage includes the 100K ohm ballast. The current for maximum power is probably between 5.0 and 5.5 mA. But I don't dare run any extended tests at those higher currents for fear of damaging the ballast at the very least, since the tube would need to go through at least one complete mode sweep cycle and preferably several to determine maximum value. The actual specs for the Z4203-60224 are: output power with collimator greater than 880 µW and a split frequency range of 1 to 3 MHz. I don't know if that includes the waveplate assembly, which this one has and would reduce the power by 5 to 10 percent.

    Further, the split frequency at 3.5 mA and operating temperature is around 2.33 MHz, not something really low only usable as an N1211A (which would increase power somewhat). Therefore, it is also suitable for a 5517B, or with even higher output power if the magnetic field were reduced to put it in the 5517A range. When installed in an HP laser chassis as a 5517B, the locked output power from the laser is around 1 mW. It may also be usable as a 5517C if the magnetic field were increased, with almost the same power.

    That strange behavior with the output power being a maximum well above the spec'd default current of 3.5 mA is not unique to this tube. Virtually all HP/Agilent/Keysight tubes have their maximum power well above 3.5 mA. It is not known why this was done as most "normal" HeNes produce maximum power at their spec'd operating current. (At least they start out that way but it may creep up with use.) For this tube, the current for maximum power is more than 30 percent above the default current of 3.5 mA. That default hasn't changed since the introduction of the 5501B and 5517A in the early 1980s. Only a very few HeNe laser power supply bricks in HP lasers had adjustable current (and none in Agilent or Keysight lasers), but they were invariably set at the 3.5 mA default at the factory. (Though people like me have been known to adjust the current to get around an increase in dropout current.) Normally, optical noise will be minimized at the optimal current. However, the most likely explanation is that these tubes are filled at a higher than optimum pressure (for an operating current of 3.5 mA) to maximize life at the expense of output power. So the optimum current for maximum power may indeed be above 5 mA. But running at lower than optimum current often results in amplitude ripple of up to a few percent at a few MHz (unrelated to the Zeeman beat). That doesn't appear to be plasma oscillations though, which occurs at a lower frequency, typically around 700 kHz for these tubes. Neither seems to be an issue here. In fact, there was no detectable amplitude noise or ripple in a frequency range up to 10 MHz above 0.1% of the laser power (about the lowest I can measure) over a current range of 2.5 to 5.0 mA. Nor was there any obvious effect on the split frequency waveform itself. Issues would probably show up at higher current (just before the ballast exploded and the tube melted) but certainly not anywhere within the normal useful range of operating current (which I define as less than 4.1 mA). That's still kind of a mystery since non-HP/Agilent/Keysight tubes with a similar power output do often exhibit easily detectable amplitude ripple and noise between 3 and 4 mA.

    In general, dropout current does not seem to be a major issue for Short tubes. Even those where the power has dropped substantially from use (and could not be tuned up by HR mirror alignment) tend to stay lit reliably at 3.5 mA. For Long tubes, the dropout current is almost always higher that 3.5 mA if the power has dropped by more than 30 or 40 percent.

    A second tube assembly is from 2007 and produces nearly as much or perhaps even more power. The glass tube was seriously misaligned when originally potted inside the magnet, which is perhaps why it appears to have seen little or no use. Further, the tube is mounted in an N1211A frame but the part number printed on the label is 05517-68201, which in itself is strange because that's a PN for an OEM tube assembly with specs similar to that of a 5517B using a Long tube. The Short tube was not used in 5517Bs until around 2012. That PN was scratched out and "60224" was printed with a Sharpie™ on the magnet. And indeed, the output power and split frequency are similar to that of the 60224, above. So perhaps only the magnet is original and a 60224 tube was installed in it. The previous tenant was probably violently evicted as it's unlikely even trained monkeys were willing to undertake a non-destructive tube-ectomy regardless of how many bananas they were promised. ;-) And from 2007, it was probably dead as crud anyhow. But the potting was done without paying enough attention to alignment, which was then too far off to be used since N1211A feet are pegged and there may be no way to adjust alignment of the overall tube assembly when installed in the chassis. So the poor thing was set aside to be neglected and forgotten until the Bean Counters decided that unused inventory needed to be gotten off the books.

    This tube can be used in either a 5517B or 5517C by adjusting the magnetic field. Shims and slight trimming of the beam expander would correct the alignment for use in a normal case. The power output from the laser would be greater than 1.1 mW for either a 5517B or 5517C. I did actually convert it to a 5517B, which required the following:

    It would probably have been fine as a 5517B, but I decided a couple years later that restoring it to the original 1 mm beam was best (but keeping the shims), since it would be unlikely anyone would really be willing to pay a premium for a >1 mW 5517B. ;-) So it became a hyper-power 5517 lab rat Zeeman laser. ;-) With soft iron bars to reduce the split frequency to ~1.6 MHz, it currently produces more than 1.2 mW after warmup. (Home-built Zeeman lasers using new/NOS barcode scanner tubes can have similar power, though their maximum split frequency will be lower, limited by rogue modes.)

    Additional HP/Agilent Resources

    Complete operation and service manuals for HP products like the 5501A, 5501B, 5526A, and 5528A that are no longer supported - as well as other related information are available for download. Most of these have been provided by Jack Hudler scanned and enhanced from original manuals. Others are copies from the Agilent Web site. I have most of them acrhived at Sam's Archive of HP/Agilent Manuals. Their content is NOT accessible to minimize BOT traffic. Please contact me via the Sci.Electronics.Repair FAQ Email Links Page to download specific manuals.

    Notes on the HP-5500 Two Frequency HeNe Laser


    The 5500A (or a substantially identical laser that had no model designation) was the original Hewlett Packard metrology laser. The 5500B appears to have been nearly identical. The 5500A/B lasers were unique compared to all that followed in having the interferometer optics built-in and the label on the laser is actually: "5500A Laser Interferometer". Thus, the only required external optic required for a single-axis system was a retro-reflector (cube-corner). In fact, the 5500A might have been considered as an option to the 5505A Measurement Display as the only actual part number on it is: C01-5505A, though the decorative strips do say 5500A.

    To get inside the 5500A/B requires removing 4 screws - 1 on each side front and back.

    The HeNe laser tube in the 5500A/B is generally similar to the one in the 5500C and 5501A, but isn't quite identical and thus is not interchangeable, at least not without some work. A diagram is shown in Internal Structure of Hewlett Packard 5500A/B Laser Tube Assembly. The original patent for the 5500A/B laser tube is: U.S. Patent #3,771,066: Gas Laser. The most notable obvious differences between the 5500A/B tube and the one in 5500C and 5501A are in the PZT connector at the rear which is a ring (rather than a center terminal) that allows the waste beam from the HR mirror to escape, and the optics assembly at the front of the tube assembly which only has the beam expander - the waveplates are mounted externally (though strictly speaking these aren't part of the tube itself). And the glass tube is simply clamped to the mounting feet, which are not part of the tube assembly.

    Rather than using a portion of the main beam for feedback, there's a shielded can with a photodiode behind a Quarter WavePlate (QWP) and motor driven rotating polarizer that samples the waste beam from the back of the tube. The photodiode signal is used in a feedback loop to lock the laser so the modes are of equal amplitude. (See: U.S. Patent #3,701,042: D.C. Motor Circuit for Rotating a Polarizer and Providing a Detector Synchronizer Signal for a Laser Stabilizing System.) Ironically, this is actually closer in function to the LCD optical switch of the 5501B and later lasers, than the polarizing beam samplers of the 5500C and 5501A that followed the 5500A. Since the 5500C/5501A tube has no waste beam exiting the laser tube, duplicating this function would be a bit of a challenge.

    The 5500A/B is in the same size case as that of the 5500C. The main difference between them is what's at the front of the laser. The 5500A has interferometer optics and detectors for both REF and MEAS within the case. The shutter wheels can select normal or alignment apertures, and either nothing or a 45 degree polarizer in the return beam path. The 5500C has two channels of optical receivers (with the shutter wheel selecting between horizontal or vertical arrangement of the return beam) but no interferometer optics. However, it was possible to install linear interferometer optics inside the 5500C to give it 5500A/B functionality.

    The 5500A is also unique among HP lasers since it is the only one with a run-time (hour) meter! This seems to even have been dropped on the 5500B, at least on the sample I have.

    There are photos of a 5500A and 5500B in the Laser Equipment Gallery (Version 4.24 or higher) under "Hewlett Packard HeNe Lasers".

    For several original articles introducing HP's interferometer-based measurement system using the 5500A, see the Hewlett Packard Journal, August 1970.

    Also see Dave Meier's HP Laser Interferometer Evolution Page which includes a links to the early HP catalog pages.

    I have a 5500A laser (see gallery pages, above) which appears to be from around 1970 based on the date code found on a 74H10 TTL IC in the optical receiver. Except for the shape of the beam expander mount and color of the ballast resistor cover, my 5500A appears identical to the laser shown on the last page of the August 1970 HP Journal. An external HeNe laser power supply was used to perform initial tested before being connected to a 5505A Measurement Display. The laser tube starts and runs flawlessly with a raw output (after the beam expander but before the waveplates) of at least 370 µW and possibly as high as 450 µW. (The power varies with temperature as the tube warms up if not feedback stabilized and I didn't run it long enough by itself to determine the actual maximum power.) Even the low end of 370 µW would be considered excellent power for a much newer 5501A tube. The output power of the laser is between 106 and 150 µW (again depending on the temperature as it's not locked). If the locked output is anywhere near the higher end of this range, then it's basically like it was when it was last serviced. There is a note inside the laser saying: "120 µW August 1978". Perhaps the tube was also replaced at that time. The reason for the large difference between tube output power and laser output power is that the waveplates cut the power by 15 to 20 percent, and the internal interferometer optics suck up approximately half of the remainder since most of the F1 frequency component doesn't exit the laser.

    When first attached to a 5505A, the laser powered up and locked instantly, and within a couple minutes, I was able to make sub-micron measurements! But, then at some point while my back was turned, the original HeNe laser power supply inside the laser head failed. Hard to believe! Not like the thing has probably been turned on for the first time in 20+ years! :) I don't know if the failure was in the two transistor driver, or inside the potted HV module, which is beautifully made in clear semi-flexible plastic with no obvious damage, the remains of which (after salvaging the HV wire) are shown in HP-5500A HeNe Laser High Voltage Assembly. But there could be a shorted turn in the inverter transformer or a capacitor breaking down. The driver transistors passed ohmmeter tests and were getting equally warm, but the output was only going to around 1 kV and then dropping to 0 V, never lighting the tube. So, I replaced it with a small brick power supply from a barcode scanner, installed inside the original aluminum can to preserve authenticity. Unless one knew exactly where to look, there would be no way to tell that it wasn't totally original.

    One thing that's probably only of curiosity value is that both the HeNe laser HV power supply and the PZT HV power supply are driven from a common oscillator which must be running for the PZT tuning to work. Without tuning, the 5505A readout may still function, but the RESET button will keep flashing. Newer versions of the 5500C, as well as the 5501A use independent self-oscillating inverters in ugly bricks made of hard tan potting compound for these two power supplies. The earliest 5500Cs are probably similar to the 5500A/B.

    It's extremely easy to align the interferometer with my home-built authentic replica of the retroreflector mount shown in the 1970 HP Journal article. As long as it adjusted so the return beam enters the optical receiver aperture or even the tiny alignment holes in the laser head turret, the system is happy.

    And here is the genuine imitation authentic setup hot off my time machine:

    More information and photos from early HP manuals and brochures, and elsewhere can be found at Dave Meier's HP Laser Interferometer Evolution Page.

    The cable wiring is given in the next section since it is the same for the 5500A/B and 5500C.


    The 5500C is one of the earliest of the HP metrology lasers (or "Laser Transducer" in HP-speak), only preceeded by the 5500A/B (described above). Thus, the 5500C is somewhat similar to a 5518A or 5519A/B in that it has an internal optical receiver for the return beam. This is actually comprised of 4 photodiodes wired as two pairs with lenses surrounding the laser output aperture. The aperture wheel may select either the horizontal or vertical pair. Thus, two independent optical receiver channels are actually available. The "A" channel is fed from the top and left photodiode, and the "B" channel fed from the left and bottom photodiode. Thus, with suitable interferometer optics, a pair of measurement axes may be handled by a single 5500C by plugging a pair of 5505As into its rear connectors. The 5518A and 5519A/B only have a single optical receiver, but the assumption by the time of their introduction was that the smaller 5501A/B and 5517 lasers with external optical receivers would be used for multiple axis machines.

    There are photos of a 5500C in the Laser Equipment Gallery (Version 2.48 or higher) under "Hewlett Packard/Agilent HeNe Lasers".

    Also see Dave Meier's HP Laser Interferometer Evolution Page which includes a link to the early HP catalog pages.

    To get inside the 5500C requires removing 4 screws - 1 on each side front and back. There may be an interlock that turns off the laser tube when the cover is removed.

    However, simple adjustment of the laser tube current only requires removing the connector cover at the back of the laser head, 4 screws. There is a trim-pot (A1R4) which sets the current and a voltage test-point (A1TP) to monitor the current. The range is from approximately 2.4 to 8.1 V. The manual says it should be limited to less than 5 V. The calibration is 1 mA/V based on a 1K ohm series resistor in the tube cathode return. So 2.4 to 5.0 mA is acceptable. Counterclockwise increases current. Adjustment is normally only required when a new laser tube is installed. Right.... :)

    If doing this with an existing tube that has started flashing or sputtering, it is probably acceptable to simply increase the setting of A1R4 to about 0.3 V above where it stays on (more CCW). But avoid setting to anything above 5 mA or risk the wrath of the HP laser gods and other consequences. :( :)

    But the complete procedure in the manual is more involved:

    1. Set 5505A Display Unit power switch to off.
    2. Disconnect interconnecting cables from 5500C Laser Head.
    3. Remove 5500C rear cover plate.
    4. Reconnect interconnecting cable to 5500C J1.
    5. Remove 5500C top cover and receiver shield. Disable interlock switch A4S1.
    6. Connect voltmeter negative lead to frame ground and positive lead to A1TP1. Set voltmeter range to 10 VDC.
    7. Connect oscilloscope probe to emitter on master Doppler preamplifier board (A5A6). Connect probe ground lead to 5500C frame ground. Set oscilloscope controls to:

      • Vertical: 0.1 V/div, AC.
      • Horizontal: 0.5 µs/div, X5 magnify.
      • Trigger: Normal, internal, positive slope, DC.

    8. Set 5505A power switch to ON position.
    9. Position interferometer and reflector in front of 5500C so that return beam is directed to lower aperture.
    10. Rotate A1R4 (LI ADJ, laser current) CW until laser tube starts flashing. Note voltmeter reading just before tube begins to flash.
    11. Rotate A1R4 CCW while observing the oscilloscope display. Note voltmeter reading corresponding to best signal-to-noise ratio (least noise on observed waveform).
    12. Set A1R4 subject to the following conditions:

      1. Voltmeter reading must be less than +5 VDC.
      2. Voltmeter reading must be at least 0.3 VDC above the value noted in step j.
      3. A1R4 should be adjusted for best signal-to-noise ratio within the above limits. Typically this voltage will be about 3.5 VDC.

    13. Set 5505A power switch to OFF position, disconnect test equipment, enable interlock switch A4S1, and replace shield and covers.

    Whew. ;-)

    The signal quality can also be assessed without opening the laser head by using a normal interferometer configuration with a separate optical receiver like a 10780A and oscilloscope monitoring one of its outputs. There will be an optimal tube current for least noise, but it may be outside the acceptable current range. :(

    The 5500C uses a HeNe laser tube with PZT tuning that appears identical to the one in the 5501A, though the part number differs. A diagram is shown in Internal Structure of Hewlett Packard 5500C and 5501A Laser Tube Assemblies. See the section below on the 5501A for detailed descriptions and and photos. (The 5500A has a very similar, though not identical tube. See the description and patent reference in the previous section.) The beam sampler for the feedback stabilization is of the common modern polarizing beam-splitter variety with the control loop driving the PZT of the laser tube to adjust cavity length. But, unlike the 5501A which only requires DC power supplies, the 5500 requires the mating 5505A Measurement Display to even turn on and stabilize since its HeNe laser power supply and PZT power supply are controlled by the 5505A. Although the HeNe laser power supply could be run open loop with a variable DC voltage, this would not provide current regulation. However, the PZT power supply of later 5500Cs which appears to be a potted module inside more potting, can be used as a stand-alone PZT, PMT, or other variable HV low current power supply since its output is fairly linear with respect to input between close to 0 and 15 V, which is multiplied by approximately 100 to produce the output voltage. Tests show that it will run with an input well below 0.5 V. Although I have not seen it specifically stated, the PZT power supply appears to be capable of more than 2 kV based on the 5501A schematics. Both HV Control and PZT Control are really just the power input to a self oscillating inverter. (Very early versions of the 5500C and the 5500A have the inverter transformers and other high voltage components potted inside metal cans with the driver circuitry on separate PCBs fed from a common oscillator.)

    The pinout for the self contained PZT power supply module is:

    5500A/B/C and 5505A connector pinout

       Pin      Function
        A       Gnd
        B       DOPPLER (A)
        C       +5V
        D       LOCK (A)
        E       HV CON
        F       REF TRIP
        G       -15V
        H       BEAM AL
        J       PZT MON
        K       REF (A)
        L       GND
        M       REF (B)
        N       DOPPLER (B)
        P       NC
        R       LOCK (B)
        S       LASER I
        T       +15V
        U       PZT CON

    If constructing your own cable, the wires to pins B and N should be shielded twisted pair, shield to pin A, and the wires to pins K and M should be shielded twisted pair, shield to pin L. The shield probably isn't critial for relatively short cables, but use the twisted pair. Size the voltage (+5, +15, -15) and Gnd wires to handle a couple amps. HV Control will also need to supply some current.

    On most (probably later) versions, the HeNe laser tube can be powered with a variable DC power supply. The two connections are:

    If the cover is removed, there may be an interlock (microswitch) in series with the power to the tube. So that would need to be defeated. The useful range for the tube to turn on is from around 15 V to 30 V between these pins, with the tube operating at the optimal current at around 20 to 25 V. But it's best to start at 0 V and work up. ;-) As soon as the tube starts, reduce current to just above where it stays lit without flickering. To safely measure tube current, put a 1K resistor between the cathode terminal (on the side of the tube) and its connecting wire. Then measure voltage (V) across the resistor. The current is then I = V(mA). The optimal current (when new at least) is usually marked on the tube and is typically in the 3 to 3.5 mA range. If the optimal current isn't labeled, a rule of thumb is to set the current 0.5 mA above the point where the discharge drops out and starts flickering, or 3 mA, whichever is higher.

    However, it appears as though very early 5500Cs may have a transistor in the circuitry leading to the HeNe laser power supply. So, if testing it as described above results in no output beam and current being drawn from your DC power supply, it will be necessary to go inside and connect directly to the HeNe laser power supply brick on the PCB under the laser.

    HP-5500C Extreme Testing

    For a project where little information is available beyond photos which may have been classified ;-), a 5500C was modified with a linear interferometer (similar to a 5500A) along with a retro-reflector and mounted on an I-beam with a length of 62 inches. It was spun in a centrifuge at 376 rpm resulting in around 125 Gs at the end-points of the beam. See Modified HP-5500C with Linear Interferometer Mounted in Centrifuge. These photos show the enclosed centrifuge and instrumentation with 5505A Measurement Display, the centrifuge open for inspection or maintenance, the I-beam with modified 5500C and interferometer, additional instruments, and the slip ring array used to transfer power and signals to and from the rotating assembly.

    It's not clear what the objective of this stunt was, but apparently no one working on the project realized that various parts of the glass laser tube and interferometer optics are mounted using flexible RTV silicone adhesive so they would move under extreme G-forces, probably rendering any displacement measurements to be meaningless. ;( ;-)

    Notes on the HP-5501A and HP-5501B Two Frequency HeNe Lasers

    The 5501A is the successor to the 5500. However, the 5500C and 5500A had optical receivers for a return beam built-in. (Two in the case of the 5500C.) The 5501A requires an external optical receiver for each axis, but thus has no limitation on the number of axes other than there being adequate beam power.

    The 5501B is a functional replacement for the 5501A. Locking of the 5501B typically takes 5 to 9 minutes compared to 10 seconds or so for the 5501A, but this is of no consequence for machines that are run for hours or years. In terms of optical characteristics, and power requirements and reference signals (including connector pinouts), they are equivalent. However, the 5501B lacks the Diagnostic (J3) connector of the 5501A, so other system components may not be happy and some substitutes may need to be provided. Going the other way doesn't have this issue, but if a 5501A is installed in place of a 5501B, it may be necessary to press the Retune button from time-to-time whereas there is no such button or need on the 5501B! This may be anywhere from a few hours to never, but it would be a good idea to do this periodically at convenient times between measurement runs, at least until the system has reached thermal equilibrium. Performing a Retune cycle does not compromise the accuracy in any way. Once the Retune LED goes out, it's ready to go again. Even from a cold start, a laser may go 12 hours or more without requiring a Retune. After that, once a day may be more than sufficient. Laser/Interferomter Page.


    The 5501A, 5500C, and 5500A/B appear to be the only models that use a special HeNe laser tube which has its cavity length fine tuned by a PieZo Transducer (PZT). (The 5501B and all successors still use a special HeNe laser tube but it has an internal heater for fine cavity length control rather than a PZT.) The OC (front) mirror and PZT with the HR (rear) mirror inside the laser tube are held in place by a spring against a spacer tube (the actual bore of the HeNe laser) made of Zerodur, a special very low thermal expansion coeffient glass/ceramic developed by the Schott Glass. In fact, longitudinal mode cycling is virtually non-existent with these lasers.

    Compared to the 5500C, the 5501A is in a much smaller lighter case, similar to the later 5501B and 5517B/C/D lasers. It also has simplified optics and totally different electronics. See Interior of the HP-5501A Laser Head - Left Side and Interior of the HP-5501A Laser Head - Right Side. The HeNe laser tube dominates the interior space in both views. The high voltage piezo driver power supply brick is visible under the magnets at the center of the tube. The HeNe laser power supply brick is underneath the output-end of the tube. The piezo driver electronics circuit board at the far right end of the right side view. The optical sensor circuit board is at the far left of the left side view.

    HP-5501A Laser Tube Assembly shows a 5501A tube by itself. The naked tube is shown in HP-5501A Laser Tube Removed From Magnet and Output Optics Assembly. The normally enclosed part is really just a very thick-walled fine-ground bore inside an outer glass envelope. A spring (visible through the glass at the left) at the rear holds the PZT, HR mirror, bore, and OC mirror in place. No adjustment is possible. There are distinct multiple spots on the card because the output window is at a slight angle and not AR-coated.

    See Major Components of HP-5501A HeNe Laser Tube for an official autopsy photo montage of one that was end-of-life and had it's tip-off broken in shipping. Only minimal sacrifices to the gods of dead lasers were required since it was already deceased. :)

    The top photo includes an intact sample of an HP-5501A tube assembly with the waveplates and beam expander. Then below from left to right:

    The inset photo at the lower left shows the HR mirror, the two tiny spring contacts that pass through it to the PZT, the PZT disk, and the HR-end of the Zerodur bore.

    The inset photo at the lower right shows the OC-end of the Zerodur bore and concave OC mirror (which magnifies the printing on the Fragile sticker).

    The Zerodur bore is precision ground at both ends to form the laser resonator with no adjustments.

    Both the HeNe laser power supply and piezo power supply run off the -15 VDC power supply. An interlock switch (easily defeated) disables operation with the cover removed. In the 5500A and 5500C, these power supplies are regulated by the 5505A Measurement Display. In the 5501A, the potted power supply bricks have no inputs other than power. Rather, HeNe laser tube current and PZT voltage regulation are accomplished by controlling the input voltage. For the HeNe laser power supply brick in the 5501A, as well as later versions of the 5500C, while the passive HV components are buried in potting compound, the two 2N5192 driver transistors are mounted on the outside of the brick and are replaceable. However, from my experience, when the transistors blow, there is probably a fault in the potted section so replacing them doesn't help.

    The pinout for the self contained HeNe laser power supply module is:

    Note that in the laser, the input voltage will be driven to max until the tube starts and then ramp down over a fraction of a second to provide the regulated operating current. So, it could go much higher than the 5 kV or so actually required to start the typical tube.

    As an experiment, I've successfully replaced the 5501A HeNe laser power supply with a common barcode scanner brick, the Laser Drive model 103-23. This has an input range of 21 to 31 VDC at less than 0.5 A, and an output of 1.1 to 1.5 kV at 3.5 mA (fixed). The 3.5 mA is a bit higher than the labeled current on most 5501A tubes, but seems to be acceptable and actually beneficial for some high mileage tubes that like to run at a slightly higher current. But, adjustable versions of these supplies are readily available. The supply was connected between the HV Control (white/green wire) and -15 VDC (violet wire) with the pot set fully CCW (max current). This assures that the 5501A current regulator will not attempt to compete with the brick's internal regulator. However, with some HeNe laser power supplies, it may be possible to use the 5501A's regulator to *reduce* the current in a stable manner. This applies to other unregulated supplies like the Hughes 3595H as well. This is left as an exercise for the student as it may not work in general or bad things may happen. :( :)

    The PZT power supply module is fully potted. Its internal circuitry is similar to that of the HeNe laser power supply brick but designed for lower voltage and current. It may be used as a stand-alone PZT, PMT, or other variable voltage low current HV power supply. But assume that 15 V between the white/red and pinkish violet wires is the maximum safe input voltage.

    The pinout for the PZT power supply module is:

    The output of the laser tube is passed through a Quarter WavePlate (QWP) to convert the circular polarization to orthogonal linear polarization components, and then through a Half WavePlate (HWP) to rotate the linear polarization by an arbitrary, but fixed angle to line the two linearly polarized components up with subsequent optics. These waveplates are adjustable with respect to orientation around the optical axis of the laser as expected. But the angle of each waveplate along one of its principle axes with respect to the optical axis of the laser is also adjustable - presumably to optimize the QWP or HWP performance, but could also be required to adjust them so they are not quite perfect to compensate for imperfect polarization purity in the raw beam - or something. :) They are both very thin and may be zero order waveplates, possibly made of optical grade mica. The beam is then expanded and collimated and passed through an angled partially reflecting plate located just beyond the collimating lens on the laser tube assembly. This deflects about 20 percent of the beam to a polarizing beamsplitter which sends each component to its own photosensor to provide the frequency control feedback. The PBS is set at approximately a 30 degree angle in the 5501A so the separation is not pure. A control loop uses these signals to adjust the PZT, and thus resonator length, so that the two signals are of equal amplitude. The difference of the two signals is the frequency/phase reference (REF) generated by the laser.

    The laser stabilization control algorithm is actually dirt simple: The voltages from the photodiodes corresponding to the two polarization components are compared in an integrator which maintains the PZT voltage at a level so they are equal. (There is an adjustment to compensate for slight differences in amplitude resulting from beamsplitter ratio and photodiode sensitivity.) While crude and simple to implement, this approach is adequate to achieve the needed stability. The electronic reference frequency signal is derived from the residual difference frequency present in one of the polarization components as a result of the PBS orientation.

    While the spacer rod has a very low coefficient of thermal expansion, it isn't exactly zero, so as the system heats up (over hours), the cavity length will still change slightly. Eventually, the PZT voltage may be unable to compensate. The PZT voltage is compared with fixed upper and lower limits which are well within the range over which locking is assured. When either limit is passed, the "Tune Fault" flag is set turning on the "Retune" LED and asserting the "Retune_Status" signal. The laser may be retuned via a pushbutton or external TTL signal). This clamps the PZT control voltage at its lowest value for a short time and then releases it to ramp up to the lock point. Requiring external intervention (whether manually or by computer) assures that a measurement will never be made when the laser isn't stable, nor will one in progress be interrupted due to the laser relocking unexpectedly.

    When testing, continuous monitoring of the amplitudes of the F1 and F2 modes is recommended, or at least periodic checking to assure that they are still approximately equal. All of these lasers show some drift in both total power (which tends to increase) and the relative mode amplitudes. The latter is likely due to etalons effects from several uncoated optical surfaces between the tube's output mirror and the F1/F2 photodiodes. Error checking in the laser is not very comprehensive, so it's possible for a failure in the locking circuitry to go undetected even though F1 and F2 differ by a large amount. For example, if the integrator is unable to reach the upper or lower detection thresholds, F1/F2 could become very unbalanced without flagging an error.

    The 5501A laser head requires +15 VDC and -15 VDC for power. (There is also a +5 VDC pin but it is an output according to the manual.) The two voltages (and common) are all that is needed to operate the laser head but an interlock switch (on the right side at the rear of the case) must be depressed to turn on the laser tube. I haven't yet looked at the output with a photodiode or scanning Fabry-Perot interferometer but after a few seconds, the "Retune" LED goes off, similar to if the "Retune" button is pressed. And then there is a stable reference signal. I have since acquired an operation and service manual for the HP-5501A laser head which confirms the information above.

    Issues Unique to the 5501A

    Due to the very different design of the 5501A laser tube, there are some things that can only occur with it.

    HP-5501A reference connector J1

    See HP 5501A and 5501B Reference and Power Rear Panel Connectors for pin location.

       Pin      Function                              Socket View
     ---------------------------------------------         A
        A       Accessory +15 VDC fused                    o
        B       +15 VDC return                         D o   o B
        C       Reference (difference) frequency           o
        D       Complement of J1-C                         C

    HP-5501A power connector J2

    Power requirements for the 5501A are +15 VDC at 0.6 A and -15 VDC at 0.5 A.

    See HP 5501A and 5501B Reference and Power Rear Panel Connectors for pin location.

       Pin      Function                              Socket View
        A       +15 VDC input                          D o  o A
        B       -15 VDC input
        C       +5 VDC output (test-point)             C o  o B
        D       Power ground

    HP-5501A diagnostic connector J3

    See HP 5501A Diagnostic Rear Panel Connector for pin location.

       Pin  Function      I/O   Comments
        A   +15 VDC TEST   O    Sample for diagnostics
        B   -15 VDC TEST   O    Sample for diagnostics
        C   +5 VDC TEST    O    Sample for diagnostics
        D   SYS COM        -    Ground/return
        E   Retune_CMD-    I    Active low to initiate PZT tune/check cycle.
        F   Retune_Failure O    Active high output indicates failure of PZT
                                 tune/check cycle.
        J   Retune_Status  O    Active high when tune/check cycle is in progress.
        K   Laser_Cur_Err  O    Active high indicates laser tube current is
                                 outside acceptable limits.
        L   Error          O    Logical OR of J3-J, J3-K, and PZT voltage outside
                                 of specifications.
        M   L I Mon Test   O    Laser current sample for diagnostics.
        N   PZT Mon Test   O    PZT voltage sample for diagnostics.
        P   Ref OK Status  O    Active low diagnostic signal indicates laser
                                 is properly tuned.


    The 5501B uses a very different HeNe laser tube compared its predecessors which is also physically similar to the tube in older 5517B/C/D lasers. It is of more conventional contruction but has an internal heating coil for cavity tuning. The 5501A and 5501B are in the same size case and look very similar externally. See Interior of the HP-5501B Laser Head - Left Side and Interior of the HP-5501B Laser Head - Right Side. A diagram of an original one is shown in Internal Structure of Hewlett Packard 5501B Laser Tube Assembly though later versions and those from Agilent will be physically similar to 5517 tubes as shown in Internal Structure of Hewlett Packard 5517B/C/D Laser Tube Assemblies. (The differences are mainly in the physical construction of the beam expander and output optics support structure, and the use of a segmented rather than 1 piece magnet. None of this affects functionality.) And see Tube Assembly Used in HP-5501B and HP-5517B/C/D Two-Frequency HeNe Lasers for the a photo of the beast. The actual glass laser tube is shown in Tube Used in HP-5517B Two-Frequency HeNe Laser.

    Like the 5501A, the 5501B also requires only ±15 VDC to power up. The current requirements are 0.8 A at +15 VDC with a momentary surge of up to 3.5 A for ~2 ms at startup, and 0.7 A at -15 VDC. The conventional HeNe laser power supply brick in the 5501B runs on -15 VDC so that the current requirements for the 5501B are similar to those of the 5501A and can be dropped in as a replacement. CAUTION: If replacing the black brick in a 5501B with a more modern metal-cased HeNe laser power supply like the VMI PS 253 to insulate its case from the chassis. (This is unlike all the 5517 lasers where the brick runs on the +15 VDC.)

    There is no case interlock on most of these, but some really old versions had one to disable the laser tube from being powered if the covers were removed, and a "Service" switch to override this. :-) Both of the switches have long since been eliminated. The PCB pads and wiring for them are still present, but bypassed. It's worth removing both switches on lasers that have them and adding the required jumper diagonally between the center pads that are closest together on both switches. (Do NOT just add the jumper - the switches must be removed!)

    On s healthy 5501B, the powerup sequence is as follows:

  • When power is first applied, only the ±15 VDC power LEDs come on. (The laser tube is not powered.) The heater inside the tube is being driven to bring the mirror spacing rod temperature up to the set-point.

  • After 3 to 6 minutes, the "READY" LED begins to flash at about a 1 second rate. The temperature feedback loop is fine tuning the mirror spacing rod temperature.

  • After another 1.5 minutes or so, the "Laser On" LED comes on and the beam should appear. The optical feedback loop is enabled to lock the F1/F2 modes to the center of the Zeeman-split neon gain curve.

  • Finally, a minute or so after that, the READY LED comes on solid and should remain that way indicating that the laser is locked and ready for use.

    Note that this method of turning on the laser only after the temperature set-point has been reached is unique to the 5501B, probably to maintain backward compatibility with respect to the maximum DC current on the ±15 VDC power supplies. But it creates problems of its own on high mileage lasers. (More on this below.) All the other HP/Agilent lasers turn on the laser with the application of DC power. (And where supply current is not an issue, an easy modification can be made to 5501Bs to disable the delay.)

    Specific times for one test beginning from a cold start at an ambient temperature of about 65 °F were: (min:sec) 3:15, 1:35, and 0:48. The first of the times is called "preheat" and is determined by how long it takes for what HP calls the "mirror spacing rod or simply "laser rod" to reach operating temperature. This is the large glass bore of the laser tube to which the mirrors are clamped at either end. It thus controls cavity length. The temperature is sensed by disabling the heater drive and measuring the resistance of the heater coil every 25.6 seconds. The warmup is much shorter if the laser is restarted after having been running: 1:00, 1:20, and 0:50. Only after the READY LED is on solid, do the reference (REF) signals appear. The 5501B adjusts the cavity length so that the two polarized components of the beam (the Zeeman split longitudinal mode, F1/F2) have equal power. Interestingly, there is only one photodiode sensor which is alternatively switched between beams using a liquid crystal polarization rotator. A sample-and-hold then outputs to the error amplifier of the optical mode control feedback loop. (This is the same scheme used in all later HP/Agilent lasers.)

    For REF, there are two outputs of about 5 to 6 V p-p (centered about 0 V), 180 degrees out of phase. For the 5501B, the reference frequency is between 1.5 and 2.0 MHz. There is no need for a "Retune" button as with the PZT based system of the 5501A. Also unlike the 5501A, there are no other signals to or from the 5501B (no large Diagnostic connector), only the +5 VDC output on the power connector, and a fused +15 VDC output on the reference connector.

    Although the control board inside the 5501B looks similar to that of the "small" 5517 lasers, it is NOT interchangeable with them as some functions like the heater drive are located on the "Connector PCB" at the back-end of the case, which is also unique to the 5501B.

    When swapping tubes in 5501B, the only adjustment that needs to be performed is for the temperature set-point, which is the same for the 5517. See the section: HP/Agilent 5517/8/9 and 5501B Temperature Set-Point Adjustment.

    For basic testing to see if the laser tube works at all, there are two ways to force it to come on immediately. Either of these should be done before applying DC power:

    1. The power transistor that switches the HeNe laser tube on is at the top near the center of the Connector PCB at the back of the laser. See Closeup of HP-5501B Laser Connector PCB. The transistor is next to the clear plastic 2A fuse. CAREFULLY, use a small clip lead to jumper its collector to emitter. These are the top pin (emitter) and tab (collector).

    2. Change the position of the following jumpers (along the top of the Control PCB):

      • JMP2 (~REF ON): LO.
      • JMP4 (HTR OK): HI.
      • JMP7 (HEATER): OFF.
      • JMP8 (CLK FREQ): HI.

      Yeah, right, why would one do all this when a clip lead will suffice? ;-) (Well, if you are all thumbs, moving 4 jumpers may be less risky than creating a short circuit accidentally!)

    The laser should turn on immediately with DC power, though high mileage tubes may take awhile to start. On the 5501B, these tend to be somewhat problematic even if they have decent output power as the power transient when the tube finally starts may reset the state machine on the Control PCB. I've seen this on many 5501Bs, so it's not something unique to a single sample. Even if the tube starts instantly, the state machine may get reset. The result is that the laser may require a few extra minutes to finally lock as it repeats portions of the warmup sequence - or it may never lock. Where DC power supply capacity is adequate, I recommend leaving option (1) in place permanently by soldering the two pins of the power transistor closest to the opt of the Connector PCB together. Warmup will then be similar to that of 5517 lasers, though probably a minute or two longer. (I still suspect a bad capacitor to be causing this behavior, but it's not one of those that is normally replaced. The laser in the photo above has had its Connector Board capacitors replaced. The others that fail with high ESR are the two aluminum electrolytics near the center of the Control PCB.)

    When the test is completed, remove DC power, wait a few seconds for the DC voltages to decay to 0 V (all LEDs dark), then remove the jumper on the power transistor or move the jumpers to their NORM (far left) position.

    There were two problems with the first 5501B I acquired that I had to deal with. The first was that the tube wouldn't stay on stably at the 3.5 mA setting (fixed) of the power supply but works fine at 4 mA. Such a condition is usually due to the tube having been run for a long time, which wouldn't be surprising with a surplus 5501B laser head. Since the existing power supply has no current adjustment, I needed to find a similar size HeNe laser power supply brick (1"x1.5"x4" or smaller) that will run on 15 VDC to replace it that can be set for 4 to 4.5 mA. The tube seemed healthy enough otherwise. I installed one that runs the tube at 4 mA but draws more DC input current than the original, and possibly for that reason, the controller aborts and resets after about 1 second when it turns the laser on. For now, to get around this, I have connected the HeNe laser power supply directly to the raw -15 VDC and added a transistor to drive its enable input when the original laser power turns on. That appeared to work fine. But after replacing the cover, the laser tube wouldn't come on. :( I discovered that it needed the room light to start! I had thought this to be a relatively rare malady for HeNe laser tubes, but more common for neon lamps and glow-tube fluorescent lamp starters. However, it turns out that a decent percentage of HP/Agilent HeNe lasers start more quickly when illuminated. So, there is now a decorative red LED shining on the back of the tube which is lit when the laser is powered. An HeNe laser power supply with a higher starting voltage would probably make this kludge, oops, feature, unnecessary. But no one will ever know about it. :) While many of these higher mileage HP/Agilent lasers can benefit from this addition, since the 5501B turns the laser on and expects it to come on quickly, it is more critical than with the other lasers like the 5517s that really don't care whether the laser is outputting a beam or not, until they actually try to lock. However, in either case, if the laser takes too long to lock, associated equipment like the 5508A Measurement Display may flag it as a failure.

    Most other 5501B problems are similar to those of 5517 lasers. See the section: Common Problems with HP/Agilent 5517 Lasers. However, one failure mode is unique to the 5501B with its PWM drive scheme. First some background: To maintain DC power requirement compatibility with the 5501A, The PWM driver can provide negative pulses or positive pulses, but not at the same time. There are separate signals from the Control PCB to enable the positive and negative current pulses. If for some reason, they are both on at the same time, bad things happen. Normally, this is should be impossible, but becomes more and more common on high mileage 5501Bs that have not had their preventive maintenance rigorously performed. Specifically, it is believed that if some combination of the filter capacitors on the Connector PCB develop high Effective Series Resistance (ESR), it may be likely. It is therefore essential to test the ESR of these capacitors (and the two on the Control PCB) periodically, and replace them if it increases significantly. The normal ESR for the two large caps is typically around 0.4 ohms. Anything above 0.5 or 0.6 ohms is suspect, though actual problems probably require it to be much higher. For the smaller caps, a typical value is under 0.2 ohms and above 0.5 ohms would indicate replacement.

    With luck "the bad things" are only that one or both of the overly expensive clear plastic 2 A fuses blow. (They can be replaced with common 2 A fast-blow Picofuses™.) But it's possible that either or both switch transistors and even their associated buffer transistors will also blow (possibly to protect the fuse), though that is futile since the transistors typically fail shorted causing unlimited current to flow which blows the fuse anyhow. ;( :) Poor fuses. ;-)

            HP     Industry  Suggested
     ID   Part #    Part #   Substitute  Description   Function    Location
     Q6  1853-0363  X45H281    D45H11    80V 10A PNP   +Switch   Near trimpot
     Q7  1854-0635  D44H5      D44H11    80V 10A NPN   -Switch   Near Pwr Conn
     Q8  1853-0058  S32248     2N3906?   Gen Purp PNP  +Buffer   Near Q7
     Q9  1854-0215  2N3904     2N3904    Gen Purp NPN  -Buffer   Near Pwr Conn

    Normal transistor tests should identify defective parts, but just replacing all 4 may be the most expedient approach. The exact D4x parts may be difficult to locate now but acceptable variations do exist from places like DigiKey. And there is nothing critical about them - many substitutes exist.

    While you're at it, if there are interlock switches, remove them and install a jumper wire between their closest center pads. The purpose of the interlock switches was probably to avoid shocking experiences if the laser tube was disconnected from the power supply with power on. This "feature" went away a few decades ago, replaced by the jumper. ;-)

    HP-5501B reference connector J1

    See HP 5501 Reference and Power Rear Panel Connectors for pin location. This connector is labeled as REFERENCE on the rear panel but shown as J1 in the installation instructions. It is J6 on the Connector PCB schematic.

       Pin      Function                              Socket View
     ---------------------------------------------         A
        A       Accessory +15 VDC fused                    o
        B       +15 VDC return                         D o   o B
        C       Reference (difference) frequency           o
        D       Complement of J1-C                         C

    HP-5501B power connector J5

    See HP 5501 Reference and Power Rear Panel Connectors for pin location. This connector is only labeled as POWER on the rear panel but shown as J2 in the installation instructions. It is J5 on the Connector PCB schematic.

       Pin      Function                              Socket View
        A       +15 VDC input                          D o  o A
        B       -15 VDC input
        C       +5 VDC output (test-point)             C o  o B
        D       Power ground

    Notes on the Hewlett Packard 5505A Measurement Display

    The HP-5505A is a part of the HP-5525A/B and HP-5526A Laser Measurement Systems, which are designed for very precise single-axis measurements in a machine tool environment.

    The HP-5525A was used in the original HP interferometer introduced around 1970 and includes the HP-5505A Measurement Display and the HP-5500A two-frequency HeNe laser head. The 5500A laser has the interferometer optics built-in and thus only requires an external retroreflector (cube-corner) on the moving part to be measured. The HP-5525B upgraded to the 5500C laser head which requires external interferometer optics but allows for two axis measurements (with a pair of 5505As!). The 5526A seems to have added a variety of options and but it's not clear how it really differs from the 5525B.

    The 5525A/B and 5526A can be set up in the field with relative ease with a minimum number of individual components and no need for a control computer as its basic functions are built-in to the HP-5505A. It provides for the stand-alone precise measurement of position and velocity. But straightness and angle are not directly supported.

    The 5505A implementation of the display function is all done in MSI TTL logic with a pair of 36 bit counter/registers for REF and DOPPLER (same as MEAS for other HP lasers), with a decimal adder/subtractor to generate the result. This is all on multiple PCBs and while there is one labeled "Program", there is no actual microprocessor controlling the system.

    The 5525A, 5525B, and 5526 all require the 5505A display but differ in the laser and options. (There may be some minor changes required to convert an older 5505A to be used in a 5526A system.) The following is from the N4MW HP 5526A Documentation Page which also has links to the actual HP catalog pages for each system.

    5500Cs have also been showing up with internal linear interferometers like the 5500A. I haven't seen any reference to this as a standard product though. I wonder if they were retrofits for customers who found their original 5525A configuration adequate or whined when their 5500As went bad and wanted an exact replacement.

    The HP-5525A/B and HP-5526A are very obsolete, but many are still in use. 5505As show up on eBay, often for next to nothing. To non-interferometer geeks, the set of Nixie tubes is probably more valuable than a working unit! However, being so old, they often have problems, and at least some of the ICs like the Nixie tube drivers are proprietary parts and no longer available.

    For info (or lack thereof):

    The laser connector on the back of the 5505A is the same type and has the same pinout as that on the 5500A and 5500C heads. The 5508A supplies ±15 VDC power for the laser head. It also controls both HeNe laser power supply current regulation and PZT laser tuning.

    To use the 5505A with a 5500A, all that's required is a 05500-60025 cable and a retroreflector (cube-corner) as shown in Original HP-5525A with HP-5500A, HP-5505A, and Retroreflector - View 1. (Additonal photos can be found in the section above on the 5500A laser.) It's straightforward to make a cable. The connectors are standard and everything is wired 1:1 at both ends. To use the 5505A with the HP-5500C laser also requires external interferometers optics. All of the standard configurations that have separate outgoing and return beams should work.

    To use the 5508A with other HP laser heads will require a custom cable and possibly a separate optical receiver which can be any version of the 10780 (A, B, C, F, U). However, some circuitry may need to be added to the 5505A to keep it happy by making it think it still has control of PZT tuning.

    FWIW in the "well that's interesting department", here is the board set from another 5505A. This is a rather vintage sample, S/N: 2016A01966, which puts its manufacturing date around 1970:

     Slot       Name           Part #        Additional Markings
      A1   Analog Board      05505-60001     Series 1920        03L
      A2   Clock Board       05505-60002  B3 Series 952-2       03F
      A3   Accumulator Board 05505-60034     Series 1920     01403F
      A4   Accumulator Board 05505-60034     Series 1920     01403F
      A5   Adder Board       05505-60005     Series 952         03F
      A6   Algorithm Board   05505-60006     Series 952      00203F
      A7   Program Board     05505-60007     Series 2240     23103F
      A8   Function Board    05505-60058     Series 1920     23903F
      A9   Multiplier Board  05505-60049     Series 1948     23103F
      A10  D-Register        05505-60010                        03F
      A11  Display Board     05505-60011     Series 1324        03F
      A12  Power Suppy Board 05505-60012     Series 1940     01503F

    Notes on the HP/Agilent 5517 Two Frequency HeNe Laser

    There are several versions of the HP (now Agilent) 5517 series. All are very similar to the 5501B in terms of the physical characteristics of the glass HeNe laser tube and the control electronics. Except for the 5517A, all versions are also in the same small rectangular case as the 5501B. However, the 5501B has a pair of connectors, one for power and the other for the reference signal; all 5517s have a single connector for power and signals. (Pinout below.) For the 5517s (and 5518A and 5519A/B), F1, the lower frequency output is horizontally polarized and F2, the higher frequency output is vertically polarized. It is not known why the F1/F2 orientation changed when going from the 5501A/B to the 5517s.

    When warming up, the difference frequency only appears for 5 to 20 percent of the time during mode sweep - only when the Zeeman modes are near equal amplitude on the split neon gain curves. And this percentage tends to be lower for higher REF-frequency lasers. The difference frequency is maximum and the output power is minimum at the center of this region, which is also where it will eventually lock. This is normal behavior for these lasers based on what is shown in Axial Zeeman Split HeNe Laser Mode Behavior. Note that while there may be another longitudinal mode present for part of mode sweep, there will be no beat except from the main pair, and then only when relatively close to being positioned symetrically on the Zeeman-split neon gain curves and only the main F1/F2 mode is present when locked. While other "rogue" modes would not produce any beat signal, they could result in problems in the interferometer and possible transient errors.

    The only functional difference among 5517 models (and the laser part of the 5518A and 5519A/B) is in the spec'd range for the REF frequency. With suitable processing electronics, any 5517 that's physically compatible (e.g., same case style and beam diameter) can stand in for any other 5517 subject to the maximum velocity limitation for its REF frequency. The measured displacement, velocity, etc., will be the same. In fact, since the REF frequency tends to increase as the laser is used, it's not unusual for a mid-life 5517B (REF range of 1.9 to 2.4 MHz) to actually meet all 5517C specs (REF range of 2.4 to 3.0 MHz)!

    The heart of all these lasers is the HeNe laser tube assembly (henceforth often referred to as simply "the tube"). This consists of the actual glass HeNe laser tube (much more below) mounted inside the Zeeman magnet and a cast or machined structure which also includes the output optics. A typical tube assembly from a 5517B is shown in Tube Assembly Used in Agilent 5517B/C/D Two-Frequency HeNe Lasers. Except for nutcases like me, these tube assemblies are considered to be non-repairable as disassembly is virtually impossible. Much more below.

    The output optics consists of a beam expander/collimator (the black object just to the right of the central aluminum cylinder) and an additional optical assembly to the right of this whose front and rear halves contain what appear to be AR-coated optical quality mica pelicles oriented at slight, but different angles. The front and rear sections can be rotated independently and they were sealed with blue paint once the perfect orientations were found. The two mica (or whatever) pieces of the optics assembly (just after the beam expander) are adjustable waveplates. The first one is a Quarter-Wave Plate (QWP) to convert the circular polarization of the Zeeman split output of the HeNe laser tube to linear polarization and the second one is a Half WavePlate (HWP) to rotate the resulting linearly polarized components to be aligned along the horizontal and vertical axes. These can then be separated out with a polarizing beamsplitter at the detectors.

    The locked F1/F2 amplitudes from these lasers are usually not quite equal. This is due in part to the beam sampler not being perfectly non-polarizing, so the horizontal polarization experiences less loss than the vertical polarization. But in addition, although electronics-induced imbalance should be very small, the LCD switch device may not be ideal. And the locked beat frequency varies a bit after locking and does not stay at its maximum value as would be expected if the stabilization was optimal. This is not a quirk of one particular laser I've use for these experiments as I've tested dozens with similar behavior - some worse than others. The cause may be various back-reflections from the multiple optical surfaces outside the laser cavity. On a common cherry-flavored HeNe laser these would not produce any detectable effects, but when dealing with small differences in very large numbers like the optical frequency, they become very evident. So, perhaps these lasers aren't as perfect as we might hope! :)

    The HP/Agilent lasers do not employ any sophisticated method of stabilization such as locking the Zeeman beat frequency (which changes slightly depending on where the modes are on the neon gain curve) to a crystal reference. They simply use the amplitudes of two orthogonally polarized signals in an analog feedback circuit as is common with most other stabilized HeNe lasers. However, here, the two polarizations are of the two Zeeman split components of the single oscillating mode rather than two separate longitudinal modes. The error signal is the difference between their amplitudes, which is forced to zero by temperature tuning of the cavity. And, in fact, there is no real need to have the frequency be precisely known or even constant over the long term, as long as it is stable over the short term. More below.

    The warmup/locking algorithm is straightforward, though just a bit different than used in many other stabilized lasers. When the laser is first turned on, it is in "Warmup Mode" and the heater, which is wrapped around the internal bore of the laser tube, is driven to reach a fixed temperature (set by the only pot on the electronics PCB). The temperature is sensed by periodically measuring the heater's resistance. This is done by disabling the heater driver, passing a small fixed current through heater wire (for 2.56 seconds out of each 25.6 second period), and storing the resulting voltage in a sample-and-hold. Since the heater wire changes resistance with temperature, this eliminates the need for a separate temperature sensor inside the tube. Once the temperature set-point is reached (the voltage from the pot approaches the voltage on the sample-and-hold), the feedback switches to Optimal Mode and alternately samples the two polarized Zeeman split sub-mode signals with their voltage difference being the error signal in the feedback loop, which is driven to zero by adjusting the temperature, and thus cavity length. In fact, from the relative shapes of the red and blue mode cycles, it can be seen that from about the last half dozen mode cycles till just before locking, the tube is actully steadily cooling rather than heating. With the heater located inside the laser tube, the time from power on to a locked condition is typically only about 4 minutes and should also be less susceptible to ambient conditions. In fact, for this example, from the relative shapes of the red and blue mode cycles, it can be seen that during most of the time from power on (a cold start) to lock, the laser tube is heating (about 75 cycles), but it switches to steady cooling (about 6 cycles) just before locking. The behvior may change slightly from one power cycle to the next, and from one laser to the next. ;-)

    Later versions of the 5517 lasers have a totally redesigned electronics board using surface mount technology with a single Xilinx FPLD containing most of the digital circuitry. I don't know exactly when this changeover took place to this Type II Control PCB, but it appears to be sometime in late 2003. The original Type I Control PCB was becoming rather dated as to parts availability so perhaps Agilent was not simply reinventing the wheel. :) The Type II Control PCB is functionally equivalent to the Type I Control PCB. But there are at least two versions optimized for the Long tube and Short tube. There is also a much more complex Type III Control PCB, which appears in a few 5517 (usually 5517D/E/F/G lasers), reason unknown. I had original thought it was the successor to the Type II Control PCB but these have been present on Agilent lasers with a manufacturing date of 2001, well before the Type II Control PCB appeared. More on the Control PCBs and locking schemes below.

    Power requirements for most of the common 5517 lasers are +15 VDC at 3 A and -15 VDC at 300 mA. Note that the +15 VDC current is much higher than for the 5501, so when replacing a 5501A or 5501B with a 5517B, the DC power supply may need to be upgraded.

    HP/Agilent 5517A

    The 5517A was the first of the 5517s and is in the larger case with a trapezoidal shape but does not include an optical receiver with an external input like the physically similar 5518A and 5519A/B. The three mounting holes are tapped M8x1.25. (You were no doubt unable to sleep not knowing this vital information!)

    Photos of virtually all 5517 laser models may be found in the Laser Equipment Gallery (Version 2.42 or higher) under "Hewlett Packard/Agilent HeNe Lasers.

    Converting a 5517A into a 5518A is simply a matter of installing the internal optical receiver PCB and replacing or removing the shutter assembly on the front of the laser. (Replacement with a shutter assembly from a 5518A laser is necessary if there is a desire to use the modified laser for straightness measurements since it has a separate setting for these.) Of all the 5517s, the 5517A (as well as the 5518A and 5519A/B) are the only ones to have a tube assembly that might appear to be of lower manufacturing quality as shown in Tube Assembly Used in HP-5517A, 5518A, and 5519A/B Two-Frequency HeNe Lasers, and is larger than the the tube assembly in the others (and the one in the 5501B). But the real reason may be that it is cast with precise locating pegs so that a tube can be swapped without requiring even minimal alignment. The actual glass laser tube is physically similar for all models except the 5517E/F/G, and some later versions of the 5517B/C/D, which have shorter tubes. Compare Internal Structure of Hewlett Packard 5517A, 5518A, and 5519A/B Laser Tube Assemblies and Internal Structure of Hewlett Packard 5517B/C/D Laser Tube Assemblies. And the 5517A tube assembly is physically interchangeable with the 5518A. For installation in a 5519A/B, there is a small piece of metal that needs to be cut away from older 5517A tube assemblies to provide clearance for the 5519's internal DC power supply. More below. Interior of the 5517A Laser Head shows the major laser/optics components of the Hewlett-Packard 5517A laser. The actual glass HeNe laser tube is inside the gray cylindrical housing which also has the cylindrical magnet for Zeeman splitting the HeNe laser lines to create the difference reference frequency in the interferometer application. See the previous sections for more information on these two-frequency lasers.

    There's no reason that a version of the 5517A couldn't be made available in the smaller style case but there was probably never any demand.

    HP/Agilent 5517B/C/D

    The 5517B and all subsequent 5517 lasers (as well as the 5501A and 5501B) are in a more compact rectangular case with identical dimensions and mounting holes configurations. The covers and front plate with shutter wheel are also interchangeable, though decorator colors vary. :) The tube assemblies consist of cast mounting brackets/feet along with a machined mount for the output optics. When swapping tubes, some horizontal alignment is usually needed before tightening down the tube mounting screws, but there should be no need for vertical alignment.

    Photos of virtually all 5517 laser models may be found in the Laser Equipment Gallery (Version 2.42 or higher) under "Hewlett Packard/Agilent HeNe Lasers.

    With minor exceptions, any 5517 tube assembly, HeNe laser power supply brick, and Control PCB may be installed in any 5517 case, requiring only a single adjustment of the lock temperature set-point to be done if the tube/Control PCB combination was changed.

    A later model (2014) Keysight 5517D has seen little change except for some cost reduction. The only significant difference compared to a standard HP or early Agilent 5517D is the use of the Short tube. And this one appears to have been cost reduced since (1) the protective plastic cover on the rear of the tube is thinner, (2) the useless trim-pot on the Connector PCB is no longer present, and (3) the aluminum parts are either bare or clear coated and no longer Alodined (goldish chromate coating). Can you believe that the bean counters at Keysight might have saved $1 on a $10,000 laser? And, the reference to the Patlex patent on the tube label is gone. ;-) Photos of this laser may be found in the Laser Equipment Gallery (Version 4.75 or higher) under "Hewlett Packard/Agilent/Keysight HeNe Lasers.

    Agilent 5517E/F/G

    Up until mid-2012, the primary evidence of the existence of a 5517E is that I found one on eBay in 2009. :) Even now, there is almost no information relating to these lasers with E, F, or G suffixes on the Agilent Web site beyond a brief summary including the REF frequency ranges and options list, and a button to order one. :) And the only thing Google returns when searching for "Agilent 5517EL" is that "Quick Fact Sheet", a Far East service company, and Sam's Laser FAQ. :) (The "L" suffix is for the vented case and these are the only versions now available. One reason for this may be that by keeping the laser tube cooler, the REF frequency can be spec'd to be marginally higher, as REF decreases with tube temperature.) Interestingly, even after paying some exorbitant price for a 5517E/F/G, the manual costs extra.

    The 5517E/F/G are the only major variation on the 5517 theme to have been introduced by Agilent. They may have been an attempt to push the basic Zeeman-split two-frequency laser concept to its limits and compete with the Zygo 7701/2 and other lasers, with their 20 MHz REF/split frequency. Agilent has also developed the N1211A Fiber AOM Laser (described below) providing even higher REF which has even more significant change, though the tube assembly is generally similar to those in other 5517s. Based on how often these have either appeared on eBay :) or from requests for repair, it's fairly obvious that they never caught on. This is likely for two reasons: Due to the limitations of Zeeman-split HeNe physics, the output power spec is significantly lower for the 5517E/F/G (believed to be 65 µW) compared to the other 5517 lasers (and even lower when compared to the Zygo lasers). This significantly limits the number of axes that can be controlled from a single laser, as well as reducing the lifetime of the laser since the power doesn't need to decline very far to be unusable. And at least as significant, even the 5517G doesn't provide a REF/split frequency that comes anywhere near that of the Zygo 7701/2 lasers - the maximum being 7.2 MHz for the 5517G. The only Agilent laser that comes close is the N1211A "Fiber AOM Laser". While the N1211A starts with a laser tube assembly similar to that of the 5517s, its REF frequency is more or less irrelevant as a pair of AOMs shifts the optical frequencies apart by an arbitrary amount. See the section: Notes on the Agilent N1211A Fiber AOM Laser Head.

    My 5517E is in the gold case, so perhaps it is only available as an OEM or "military calibrated" product? The tube also has no label, so the one I have may have been an early prototype. It runs at 6.3 MHz, which is slightly above the spec'd minimum of 5.8 MHz. The associated spec for the 5517E - 1.6 m/s maximum velocity is much higher than that of the standard 5517D's resulting in over a 50 percent greater velocity measurement capability. The textbook party line had been that axial Zeeman HeNe lasers above about 4 MHz were simply not viable. The requirement for a higher REF/split frequency likely means that the magnetic field is stronger and thus the total extent of the Zeeman-split neon gain curve is wider (necessitating a shorter cavity length with the larger FSR to suppress rogue modes, which as a side effect, also increases REF/split frequency at the expense of output power). An informal measurement of the magnetic field of a 5517E did show it to be 5 to 10 percent stronger than that of any other 5517 laser, though there was quite a bit of variability even for the same models (e.g., 5517B). The tube assembly looks almost identical to all the others except that it is about 0.75" shorter up front beyond the section with the magnet and the glass tube itself is only 6.3" compared to 8" for the long tubes. In fact, in order to work at all at these high REF frequencies require a careful selection of mirror reflectivity, magnetic field, cavity length, and no doubt many other parameters of the laser design. The 5517E/F/G lasers are all operating on the hairy edge of what's possible with Zeeman-split HeNe laser technology, balancing desired high REF/split frequency, acceptable output power, and avoidance of rogue modes in the output. These lasers have a very low spec'd minimum output power (65 µW for the 5517FL, and probably the other 5517E/F/Gs as well) compared to the 5517A/B/C/Ds (180 µW). A new "lively" 5517E may produce 120 µW compared to over 600 µW for many 5517Bs, and even more for some 5517As. A photo of a 5517E I saw recently had a backplate output power of 110 µW and REF frequency of 6.1 MHz. The physical design of the tube squeezes almost every last drop of performance out of it, and even then, the power is low. It's really a last gasp on Agilent's part to retain customers needing higher performance who might have otherwise switched to a Zygo laser with its 20 MHz REF frequency. (Zygo lasers use an AOM to split the frequency rather than the Zeeman effect and has no problems with output power.)

    Several views of a naked 5517E laser are shown in Agilent 5517E Laser Head With Cover Removed.

    Without actually dissecting a Short tube, the cavity length can be estimated by measuring the longitudinal mode spacing using a Scanning Fabry Perot Interferometer (SFPI). During warmup, two longitudinal modes are present over a portion of the mode sweep cycle, so their spacing compared to the FSR of the SFPI provides a good estimate of the FSR of the laser, and thus its cavity length. For the 5517E and 5517FL, the mode spacing/FSR is approximately 1.5 GHz implying a cavity length of around 10 cm (~4 inches). This is about 20 percent shorter than the cavity length of the 5517A/B/C/D lasers. (Though newer 5517s may also use the shorter tube.)

    Unfortunately my X-ray vision is somewhat limited. Even X-ray Views of Typical Long-LV (5501B), Long-HV (5517C), and Short (5517D) HP/Agilent HeNe Laser Tubes doesn't reveal much. Yes, as of around 2012, *all* 5517 lasers use the Short tube. Of what is visible, the most obvious difference is that the HR-end of the tube has a metal cap on it instead of the glass with spring affair of all the other 5517 lasers. At the OC-end of the tube, reducing the space between the mirror and discharge escape hole would decrease the cavity length by a sufficient amount, but wouldn't require any major redesign. So, at first (before even seeing bare tube) I assumed the design would be similar to that of the 5517A/B/C/D tubes, only shorter. But when I finally was able to remove the beam expander on one for inspection (before acquiring a bare tube), that was found to not be the case at all, with the OC mirror attached to the end of the mirror spacing rod, but the entire affair is unsupported at the front end. And that can have consequences. Strange.

    Finally in 2015, I was able to obtain a few certifiably dead Short tubes with the unsupported bore broken off and good for nothing but dissection in the interest of science. :) A close examination of the remains confirms what was inferred previously. The HR is attached to a post with glass frit and the OC is attached to a cage affair on the front of the mirror spacing rod, also with glass frit. A metal (the actual back of the cage) has a small hole drilled through its center. Whether it simply lines up with the bore or is smaller or larger cannot be determined without more drastic destructive measures.

    Here are some photos and diagrams:

    This particular 5517E has the most incredibly complicated Control PCB of any HP/Agilent laser I had see before finding it, even compared to the Type II Control PCB (see below). I've since found similar Control PCBs on a few other Agilent (post-2000) 5517 lasers, but they are quite rare and one was on a high REF 5517D, so they are not unique to 5517E/F/G lasers. It includes a SHARC DSP, two Lattice FPLDs, and a lot of other digital circuitry, purpose unknown. They also seem to have gone back to PWM for the heater drive since there is no power transistor on a heatsink, as with the original Type I and the updated Type II Control PCBs. However, that collection of inductors visible in the lower left of the photo may be there to clean up the drive to the heater and remove the high frequency switching noise. Since the locking should be basically the same as for the other lasers, this level of complexity is perplexing unless this particular unit was designed to have much better stability - perhaps the "military calibrated" version. Unfortunately, the Type III Control PCB lacks all the familiar jumpers and the temperature set-point pot, and adds a couple of micro DIP-switches and connectors, purpose also unknown. It's possible that adjustments can be made in the firmware via RS232, or the laser may automagically determine the optimum operating point during the extended warmup period. This particular 5517E may have been some sort of prototype or test unit as there is no label on the tube, only a magnetic field strength of 363 G, though that, too, is not known. Aside from the unknowns, everything else is obvious. :)

    Here are several closeup photos:

    Functionally, this 5517E with the Type III Control PCB behaves more or less like the other 5517 lasers. The user LEDs are the same but there are 4 LEDs on the control board that I'm sure provide a wealth of information if one knows how to interpret them. My sample takes over 5 minutes for READY to start flashing. READY also stops flashing once or twice for a couple minutes, before it resumes flashing, and then locks after about 9 minutes. Whether these long times and peculiar flashing behavior are normal or indicate some problem, is also unknown. However, with a similar Control PCB and heslthy 5517B tube, the behavior is similarly strange. More on this in the sections: HP/Agilent 5517 Laser Control PCBs and Locking Sequence and Agilent 5517 Laser RS232 Communications. Other 5517E/F/G lasers I've seen use a version of the Type II Control PCB and lock in the normal time.

    The tube is somewhat low power compared to what's normal for other 5517s - about 120 µW locked. But I have no specs on 5517E minimum output power, so with the shorter tube and likely stronger magnetic field, that might be acceptable. In fact, the minimum power spec for the 5517FL is only 65 µW, so the 5517E may be similar. And given that other sample with 110 µW as the value when new, 120 µW on a used laser may in fact be absolutely wonderful. :-) Once locked, it's quite stable with minimal drift in REF frequency. Given the huge amount of computation power available, it may count mode sweep cycles instead of using a temperature set-point (or in addition), and might also adapt automagically to a replacement tube - or require a factory upload of tube parameters via one of those unlabeled connectors!

    The high REF frequency of 6.3 MHz works fine with my home-built SG-MD1 measurement display, but comes up as "LASr FAIL" on a 5508A. This isn't all that surprising since 6.3 MHz is almost twice the maximum REF frequency of the 5518A for which the 5508A was intended. However, the same type of Control PCB with a 5517B tube locks fine but also fails keep the the 5508A happy, so it is more likely due to wimpy line drivers or something like that. :)

    This montage of Agilent 5517FL Laser and Components shows views of a 5517FL in various stages of disassembly. (Sorry about the photo quality - I do not have this laser.) It had a listed output power of 160 µW and REF/split frequency of 7.12 MHz. (The minimum specs are 65 µW and 7.0 MHz, respectively.) The overall construction is similar to that of the 5517E including the overhead-mounted ballast resistor, though the HeNe laser power supply brick is in a fully shielded metal box. The portion of the tube assembly housing the beam expander is longer than the one in the 5517E and the same as that of the 5517B/C/Ds, but of no significance since it doesn't affect anything beyond looks. The design and size of the tube is also similar except that it's a bit more polished with a real Agilent label! But, from the photos, it appears as though the heater resistance adapter found in my 5517E is not used, so the tube heater resistance must be higher. However, this unit had the normal (for recent Agilent lasers) Type II Control PCB rather than the fancier one found in the 5517E. A pair of production (based on the tube label) 5517Ds and a 5517FL had the Type III Control PCB, and indeed, no heater resistance adapter.

    Additional photos of the 5517E and 5517FL (and other 5517s) may be found in the Laser Equipment Gallery (Version 3.18 or higher) under "Hewlett Packard/Agilent HeNe Lasers".

    Testing of HP/Agilent 5517 Lasers

    The following represents the minimum that should be performed to assure that a 5517 (as well as 5518A, 5519A/B) laser will serve as more than a bookend may be of use at least as an emergency spare, or for a hobbyist or experimenter. And failure doesn't necessarily mean that the laser is useless. More extensive testing would be required in either case to determine its condition.

    The appropriate DC power supplies and laser head cable will be required for all except the 5519A/B, which plugs into the AC line. (Testing of a 5501B laser is similar except that by design, the laser beam doesn't appear for several minutes into warmup. See the sections on the 5501B for more details.) Aim the laser at a white card or wall to view the beam.

    Assuming a beam does appear, it should stay on without any flickering or sputtering. Continue watching it for the next several minutes. Power down if it does not stay lit - damage to the tube and/or HeNe laser power supply may occur if it continues to drop out and restart. Note that on high mileage lasers, there may be a significant periodic variation in beam intensity during warmup due to normal mode sweep. This should not be confused with flickering or sputtering. The smooth variation probably means the output power is relatively low but the laser may still be usable. However, a slight variation may be present even on a new laser.

    A very few versions of the 5517 may take 10 minutes or longer for these two steps to occur. But it's highly unlikely you'll ever run across one of those. However, occasionally, a laser with marginal output power will take somewhat longer than the normal 4 to 5 minutes to lock as the laser power gradually increases after full warmup.

    Once READY is on solid, the laser is locked and usable. But to have any confidence in its true condition, additional tests need to be performed. The most important are to measure the laser output power and REF frequency to compare with either values on the laser's backplate (if present) or specifications for the specific model. However, it is now known that the laser is most likely good for more than a doorstop. :)

    If the laser beam appears and remains on but the laser doesn't lock, either the laser output power is too low (typically less than 80 to 120 µW) or there is a problem elsewhere in the laser. If the beam does not come on or does not remain on, the tube or HeNe laser power supply may be bad. See the sections below for more information.

    And for much more than you probably want (or need) to know, see the companion document: Considerations in Evaluating Used or Rebuilt Hewlett Packard/Agilent Metrology Lasers.

    HP/Agilent 5517 Mode Behavior

    Like any laser, these metrology lasers exhibit longitudinal mode sweep or mode cycling as they warm up and the cavity expands. However, the Zeeman-splitting and locking makes things more interesting:

    Plot of Hewlett Packard Model 5517C Stabilized Laser During Warmup shows how a typical 5517 laser behaves. Note that the entire warmup period from laser on to locked is only around 3.5 minutes because of the internal location of the heater for the active mode as noted above. A laser with the more common external heater could take 20 minutes or more to lock. The control algorithm is a bit more sophisticated than used on some other stabilized lasers, checking periodically for the status, and switching from "Warmup Mode" to "Optical Mode" about half way through the warmup period, at which point the READY LED starts flashing. A short while after it locks is when the READY LED comes on solid.

    Plot of Hewlett Packard Model 5517C Stabilized Laser Near End of Warmup shows the 5 mode cycles just before locking and the final transition to the locked state. The peculiar shape of these Zeeman-split modes is clearly evident in this expanded view. Part of this is due to the locking algorithm switching between heating and cooling, but mostly it's a result of the effects of the magnetic field. More below.

    The beat frequency is shown for the last 5 cycles and after locking in both these plots. This is the actual measured frequency captured along with the vertically and horizontally polarized modes and total output power. (Showing the frequency plot earlier would be a mess.) The beat only appears for a small percentage of the mode cycles with some variation during the time it is present, peaking when the F1 and F2 amplitudes are equal, and only when F1 is rising with increasing temperature. There is no beat when F1 and F2 are equal but F2 is rising with increasing temperature. The reason for this becomes evident from the simplified diagram in Axial Zeeman Split HeNe Laser Mode Behavior, or more accurately in HP-5517 Zeeman Split HeNe Laser Mode Behavior. The second diagram has been specifically crafted based on the mode plots, above, as well as SFPI Display of Lasing Mode Power Envelope of Horizontal Polarized Output of Healthy HP/Agilent 5517B Laser, of mode sweep on a storage scope as the laser warmed up. (The envelope of the vertical polarized output would be a mirror image of this one but I don't have a color digital scope to view them at the same time. The single peak visible within the envelope is really a pair of Zeeman-split modes but the resolution of the SFPI is over an order of magnitude to small to resolve them.) Thus, the plots in the second diagram more accurately represent the actual behavior of the 5517. And the clutter of the gain curves and associated junk has been removed. :) Both diagrams show snapshots of most of a mode sweep cycle starting with the cavity being 1/4 wavelength too short and ending with it being 1/8 wavlength too long. (The case of 1/4 wavelength too long would be the same as the first, 1/4 wavelength too short). Only when the longitudinal mode is near the center of the Zeeman-split neon gain curves will there be a beat. In addition, the mode amplitudes are changing rapidly as the cavity expands at those high slope locations on the gain curves. When the cavity length changes (longer or shorter) by 1/4 of the lasing wavelength of approximately 633 nm, the amplitudes are again equal, but the two separate longitudinal modes are oscillating far apart and there is no beat. Note that the red and blue plots include the F1 and F2 amplitudes, but also may have contributions from another longitudinal mode derived from the same split gain curve which will thus have the same original circular polarization. But when centered and locked, only the desired Zeeman-split modes are oscillating.

    Note that as the tube ages with use, the gain declines and the width of each gain curve that is above the lasing threshold decreases. Eventually, with a really high mileage tube, there may be no overlap at all and the beam will probably disappear for a part of the mode sweep cycle. But it is exactly at that point where the Zeeman beat would be generated, so it will also disappear entirely. Lasers are generally taken out of service long before this happens, but I recently found one whose output power was so low that this behavior was present - or absent depending on your point of view!

    The second diagram above, HP-5517 Zeeman Split HeNe Laser Mode Behavior would likely apply most accurately to a nearly new 5517 since that's what it was more or less based on. As expected, when the split mode is centered, there are no other modes oscillating. But if slightly off-center, there is a strong mode at a distance of 1 longitudinal mode spacing from it. Normal and Zeeman-Split HeNe Laser Mode Power Curves. Compares a "normal" (common cherry flavored HeNe laser), and two 5517s. One has seen a fair amount of use while the other is close to new. Note that the similarity in the general shape of the "hat" - the top portion of the lasing output power curves, but the new laser has the added "skirt" below, which has a similar amplitude. The skirt is present in the region where there are two longitudinal modes lasing with one of them being a Zeeman-split mode. Thus, when and if a skirt will be present, and its height relative to the hat region, will be affected by the cavity mode spacing and magnetic field strength. As far as the mode sweep is concerned, the skirt mainly adds an offset to the total output power. Two other 5517Bs in various stages of life show similar skirts. A relatively low mileage unit (but not quite as new as the one in the diagram) looked much the same but with a slightly higher ratio of hat:skirt height. :) And one that had been really high mileage whose magnetic field was reduced to bring down REF had a 3:1 ratio of hat:skirt. With the reduced field, the central region is wider, but the hat is otherwise similar. So there are 3 lasing regions in these Zeeman-split mode plots as shown in Mode Competition in HP/Agilent 5517 Zeeman-Split HeNe Laser:

    Here is how the article "An Instant-On Laser for Length Measurement" by Glenn M .Burgwald and William P. Krugein describes the operation of the laser tube in the Hewlett-Parckard Journal, Aug., 1970.

    "If an axial magnetic field is applied to a laser which is free from polarization anisotropy in either the mirrors or the plasma tube, the output splits into two frequencies of left and right circular polarization. First-order theory predicts that the frequency splitting is proportional to magnetic field strength and to the ratio of line Q to cavity Q. In the new laser, magnetic field strength is adjusted for a difference frequency of about 2.0 MHz. Line center is virtually midway between the displaced lines, so proper cavity tuning can be assured by adjusting for equal intensities of the lines."

    This was written with respect to the earliest HP metrology lasers but the principles are the same for the 5517s (as well as the 5501B). And they show a gain curve diagram even simpler than the one above. See that article for more details. The first order theory is consistent with my measurements and speculation where I use "cavity loss" instead of "the ratio of line Q to cavity Q" but they are equivalent.

    The peculiar shape of the real mode plots almost certainly due to mode competition between the pair of Zeeman modes, and at times between the Zeeman modes and a normal mode that may also be present, and between two normal modes if only they are present. But so far I have found no references anywhere. The split gain profiles need to be asymmetric to account for it, and this has been confirmed by testing several 5517 lasers on a Scanning Fabry-Perot Interferometer (SFPI). The simpified explanation of Zeeman splitting rarely takes into account what happens in the real World which distorts the gain profiles in these lasers as a result of mode competition for the same pool of excited atoms. This happens in short normal HeNe lasers as well, but it isn't as dramatic. (More on this below.) So, drawing a pair of nice bell-shaped gain curves really isn't accurate. The net effect is depicted in HP-5517 Zeeman Split HeNe Laser Mode Behavior. Here, the lasing mode power curves have been modified so that the results would be roughly similar to what was in the plots, above. And HP-5517 Zeeman Split HeNe Laser Mode Behavior Versus Mode Position on Gain Curve shows one complete mode cycle along with little split lasing mode power curves.

    Also note the second longitudinal mode (in addition to a Zeeman-split mode) present for a part of the mode sweep cycle. Extra "rogue" modes should never be present when HP/Agilent lasers are locked, though one may appear at times as in the diagram when warming up and the Zeeman-split modes are not centered on the split neon gain curves. If any are present when locked and they align with the X and Y axes, then the only effect will be to slightly decrease the MEAS or detected REF signal level with respect to laser output power since any difference frequency is way outside the passband of any electronics. However, if they are not aligned with the X and Y axes (e.g., at 30 degrees), they will cause level changes in the envelope of the signal from the optical receiver's photodiode due to self-interference in the interferometer. This is similar to what would happen if the primary Zeeman modes were misaligned, or not pure. The consequences could be transient position errors but only during motion. The end-points would be accurate since the optical receivers only respond to AC. There's a fine balance between the desire for a high split frequency (which extends the split gain curves) and the desire to suppress these "rogue" modes. So, for example, increasing the magnetic field to boost split frequency may produce rogue modes if the cavity length isn't also decreased.

    Normal and Zeeman-Split HeNe Laser Mode Power Curves compares behavior quite close to what's actually observed. I can't guarantee that these are to scale, but they do show the general shape with and without a magnetic field. The lasing mode placement in the diagrams is such that there are two equal amplitude normal modes without the magnetic field and two equal amplitude Zeeman modes with the magnetic field, with a cavity length selected to just suppress rogue modes in the latter case.

    I was curious (actually quite curious) to see what the mode behavior of a typical HP laser would be without a magnetic field. One must be quite curious - in fact quite quite quite curious - to do this as it requires removing the glass tube from the magnet intact - which is literally an all day (well all morning) affair using knives, dental tools, and other instruments of torture to dig out the rubbery potting compound securing the tube inside the magnet/optics assembly. With the strong axial magnetic field doing all sorts of wonderful things and based on the effects of Zeeman splitting, the plots of mode sweep with and without the magnet should be dramatically different. Although I had already removed several tubes from their magnets, they were all end-of-life with rather low power so there would always be questions as to whether whatever was found would also apply to a healthy tube. I had a 5517C that would lock with decent output power (265 µW) but required over 4 mA for the discharge to remain stable during warmup. Such a tube should behave reasonably normally, laser-wise, but since the future life of high dropout current tubes is unpredictable, it would not likely be in demand and could be sacrificed in the name of science (and curiosity). (In principle, the tube could be remounted and used but I doubt that will ever happen.)

    First, a plot of the mode sweep of this laser was made as a reference. Its appearance was similar to that of those shown above. Then the major surgery was performed to remove the glass tube. The initial results were quite strange. It appeared as though there were always 2 modes that were nearly identical except for a burst of randomness where the mode sweep would normally do its mode hop thing. And these were present at both polarizations! It was as though the output was totally non-polarized - rotating a polarizer had almost no effect! The removal process was rather violent at times (but I won't go into all the gory details!), so I put the tube back in its magnet to confirm that it had not been damaged. It hadn't. After ruminating on this totally peculiar mode sweep during my afternoon walk, I began to suspect something in the environment like a stray magnetic field resulting in a transverse Zeeman effect producing a split mode with a very small difference frequency. That sort of strange mode behavior is a characteristic of the mode sweep of a transverse Zeeman laser at some range of relatively low magnetic field strength. (See the section: Transverse Zeeman Stabilized HeNe Lasers.) And there was another HP laser sitting less than 1 foot away! Sure enough, removing that laser produced a mode sweep more along the lines of what would be expected with a HeNe laser tube having a cavity length of 127 mm (longitudinal mode spacing of 1.2 GHz). See: HP-5517C HeNe Laser Tube Mode Sweep Behavior. The top plots are of the normal 5517C laser with the lock point being where the red and blue (F1/F2) polarized modes (lined up with the horizontal and vertical axes) cross, located at minimum output power. The angular shape is due to the distortion of the split neon gain curves resulting from the magnetic field, tube geometry, and other factors. It corresponds fairly accurately to the diagrams shown above. The bottom plots are of the same tube without the magnet and waveplates so that the polarized modes are the (non-Zeeman-split) longitudinal modes. The time axes of the two sets of plots are similar and the plots are approximately aligned one above the other more or less where the mode is centered on the Zeeman split neon gain curve (top) and the normal neon gain curve (bottom). For this tube, the normal polarized modes also line up with the horizontal and vertical axes - probably not entirely a coincidence. (This is not required since the waveplates can correct for any mode angle but it would simplify the alignment process.) However, the appearance is still not quite typical, as it's somewhat polarized with bumps. The red mode in the bottom plot doesn't quite go to 0 as it would in a linearly polarized laser. And where the bumps are, the mode orientation should reverse, but it does not. And although the appearance would suggest a neon gain curve with a relatively flat top (just a small depression in the middle) and steep sides and a lasing width of only about 1.3 GHz, not the 1.5 or 1.6 GHz normally used, this could also be at least in part a result of mode competition. However, with the lower mirror reflectance and thus increased lasing threshold of the higher REF frequency 5517 lasers, there would be a narrower effective gain bandwidth. But some stray magnetic field must still be influencing its behavior to cause the polarized mode behavior. In fact, this tube is extremely sensitive to magnetic fields - much more so than the average run-of-the-mill HeNe laser. But I'm not sure that even if stray magnetic fields were totally eliminated, the tube would revert to totally normal behavior. I removed all sources of stray magnetic fields I could identify including a loudspeaker a couple feet away, and degaussed the tube and housing (though there don't appear to be any ferrous materials there), but there was no change. Perhaps a set of Helmholtz coils to eliminate the Earth's magnetic field would be able to create a sufficiently field-free region of space. This specific tube is not unique though - a 5517B tube produced a generally similar set of plots and has the same sensitivity to magnetic fields and a 5501A also behaved in a non-typical way with no external magnetic field. What this probably indicates is a polarization anisotropy in the laser tube very close to zero, required for Zeeman splitting to predictable and consistent or be present at all. Or the opposite. :) There is one thing that is very asymmetric: the anode-end discharge enters the bore from one side and the cathode-end discharge exits the bore from the opposite side. In most modern HeNe laser tube, these are generally fairly symmetric in both cases. And on the SFPI, there appears to be something very strange going on when looking at the output through a polarizer. It may be a large frequency modulation of the optical frequency as the appearance is of a full amplitude high frequency oscillation in the mode display, but only with a polarizer. The might be due to the HeNe laser power supply ripple, or low level plasma oscillations, or aliens attempting to communicate with Earth. :-) Without a polarizer, the appearance is normal.

    Also, note the depression in the blue mode (and total power). Although that may indicate the presence of a Lamb dip - and many aspects of the physical design of the HP/Agilent tubes are consistent with the requirements of a Lamb dip laser - it could simply be an artifact of the way the neon gain curve lines up with the longitudinal mode spacing.

    Since all HP/Agilent thermally-tuned lasers employ a tube with similar construction, I would expect their mode sweep behavior to be similar as well. And I doubt this behavior has anything to do with usage - it is simply a characteristic resulting from the design. But it would be nice to be able test a new 5517 tube sans magnet. However, the probability of this happening is somewhat below that of pigs flying.

    In fact, plotting the horizontal and vertical components of the polarized modes of a healthy Agilent 5517C tube with no magnetic field as it is rotated through approximately 120 degrees in 15 to 20 degree increments shows some even stranger mode behavior. See Mode Sweep of Agilent 5517C HeNe Laser Tube with No Magnetic Field. The plots of the other 5517C laser, above, are very well behaved in comparison. This tube came from a laser removed from service due to random dropouts, likely a result of a bad cathode connection. But it still had output power well above 300 uW and REF below 3.0 MHz. It's not even really possible to define a set of polarization axes as would be the case with a "normal" HeNe laser tube. The best that can be concluded is that the mode variation is largest in the first plot and smallest in the 5th plot. For a normal tube, there would be an orientation with zero amplitude. However, as expected, the total power plot is well behaved.

    To reiterate, I believe this sort of crazyness to be normal for HP/Agilent laser tubes including all 5517s, the 5501B, and probably the 5500A/B/C and 5501A as well. Without a magnetic field, they are all highly random polarized (and I mean RANDOM). Healthier tubes may actually actually be more random. In these plots, there is only a weak tendency toward a set of polarization axes and any disturbance will screw up the polarization. In fact, the slightly vibration will result in wild variations in relative mode power during part of mode sweep. Similar vibration will have no effect on "normal" tubes. I do not know if this behavior is due to the tube structure or the mirrors, or some magic sauce. ;-)

    However, applying a modest axial magnetic field wipes away fingerprints. :) Once the field strength reaches a value of around 100 G, the randomness disappears. This threshold tends to be slightly lower for HP/Agilent tubes compared to common internal tubes like those from Melles Griot or JDS Uniphase. It is likely related to the randomness with no field which may promote Zeeman splitting of the lasing mode. But at the normal operating field strength of 200 G or above, it's essentially impossible to tell the tubes apart from mode behavior.

    The older 5501A also behaves much the same as shown in Modes and Beat Frequency of HP-5501A HeNe Laser Tube 1 With Normal Axial Magnetic Field.

    The Agilent 5517E/F/G lasers use a tube with a slightly shorter cavity, but behavior is generally similar with the magnetic field present. During mode sweep, there is a modest variation in total power and this is a somewhat larger percentage of the total power than with the longer 5517A/B/C/D tubes even when new. However, what may not be obvious is that with no magnetic field, the output power from these short tubes may decline dramatically - or even disappear entirely - during a part of mode sweep. Their longitudinal mode spacing is around 1.5 GHz, and two modes just barely fit under the neon gain curve. At that point, the modes are on the tails of the gain curve, near or below lasing threshold. With one sample, the power variation with no magnetic field was from 0 to 450 micro;W. But with the magnet, it was only 300 to 360 µW. The normal lock point would be near the minimum of 300 µW, but that's a lot more than 0 µW! :)

    Performing the same tests with one of these short tubes produces even stranger results as shown in HP/Aglient 5517E HeNe Laser Tube Mode Sweep Behavior. The conditions are essentially the same as for the 5517C plots, above. Again, the behavior is smooth and predictable with the magnetic field, but the polarization is totally chaotic at times without it. I don't really know what's happening most of the time. Changes may be taking place on a time scale faster than the sampling rate of the data acquisition system, which is only 60 samples/second. But note that the total power (green) curve is still smooth as expected. It's only when observed through a polarizer the it turns to randomness. Even when the polarization isn't changing randomly, it isn't what one would see with most "normal" HeNe lasers. And here is a closeup of the randomness: HP/Agilent 5517E HeNe Laser Tube Mode Behavior with No Magnet - Expanded. Rather wild, no? :) Someone asked how this tube got through Agilent Quality Control. The answer is that I believe it is this way by design, or at least this behavior when not inside a magnetic field is a byproduct of the design which minimizes asymmetries in the mirrors, bore, and other aspects of the laser tube construction.

    Out of further curiosity, I did the same experiment with my custom SG-5517 laser which uses a Spectra-Physics 007 HeNe laser tube. These plots are shown in SP-007 HeNe Laser Tube Zeeman Split Mode Sweep Behavior. Again, the lock point is where the blue and red modes cross close to minimum total output power. While the general character of the plots is similar to those for the genuine HP-5517C, the details differ dramatically. And the SP-007 has none of the hyper-sensitivity to stray magnetic fields that is present with the 5517 tubes so the normal longitudinal modes (no magnet) look like those of a short well behaved random polarized tube. (However, in the interests of full disclosure, these plots were not of the same physical tube, only the same model tube since I didn't want to disassemble the SG-5517.)

    Out of further further curiosity, I tried the same experiment with the totally screwed up (as far as mode behavior is concerned) Far East tube. (See the section: A Far East HeNe Laser Tube.) It's about 6 inches long but unlike most typical short tubes, it has a very long radius hemispherical cavity, with a curved mirror of around 1 meter RoC - 3 to 4 times what is common, so the mode volume should only taper a small amount within the active discharge. The gain curves displayed on the SFPI did not show the dramatic asymmetry present with the HP-5517C or even the SP-007. While far from conclusive given the overall pecularity of this tube as well as other basic differences compared to the HP-5517C and SP-007, it is, well, interesting. No plots, sorry. :-)

    Excel 1001A/B/F metrology lasers, which are performance clones of those from HP, use tubes of conventional design. Their mode behavior is definitely not the normal shape and is somewhat asymmetric, but also not nearly as skewed as that of the HP lasers. This might make some sense if they have the 25 to 30 cm OC found in typical short HeNe laser tubes and thus a somewhat long radius hemispherical cavity. Then again, that may be totally bogus. :)

    So what about the asymmetric shape of the Zeeman-split gain curves? At his point, it is almost certainly a result of mode competition between the two Zeeman-split modes, and with and between any normal modes that may also be present. The magnetic field splits the gain curve and shifts the two copies apart but doesn't do much more.

    However, when I first became obcessed with the strange shape, many mechanisms were considered and ruled out. For example, that the magnetic field takes the original symmetric gain curves and smears them out non-uniformly, or affects the Doppler broadening non-uniformly. But doing so requires that the population for the right-circularly polarized mode be the mirror image of the population for the left-circularly polarized mode, not simply that the populations are smeared and shifted. This severely limits the explanation. Or, even more off the top, that Zeeman splitting produces more than one pair of shifted neon gain curves (i.e., "hyperfine structure") and the weighted sum then represents the net gain curves. Here are some other possibilities that have been pretty much eliminated:

    The closest to an explanation for the peculiar shape has come from Harold Metcalf, Distinguished Teaching Professor at Stony Brook University, Stony Brook, NY, and that was simple mode competition. In retrospect, mode competition is a natural fit for the observed shapes of the lasing mode power curves. This agrees with the behavior in the very straight slopes and nearly flat region of output power within the region where there is a split frequency present. Basically, although the gain curves are visualized as being separate, there is still only one population of excited atoms, so the two components of the split lasing mode are competing for a limited resource. My interpretation is that at either end of the split frequency region, only a single mode is oscillating while in the exact center, there are two modes with equal amplitudes. Being resource-limited, the sum of the two modes in the split mode must be similar to the single mode at either extreme, and the simplest equation that will then join them is a straight line, which also explains the nearly constant output power in this region. Something similar is happening in the region where there are two separate longitudinal modes oscillating since again, even though the neon gain curves are split, it's still a single population of excited atoms. At the very least, it's easy to argue that the total power as a result of the sum of the two separate longitudinal modes isn't going to be double or more of the power in the split mode when it alone is present (though it is slightly greater). However, it's even harder to visualize the behavior in those regions. It may be time for Matlab.....

    There's probably a research paper from the 1960s or 1970s that will make everything perfectly clear. But all those I've found so far have been less than entirely useful. Translation: I couldn't make heads or tails out of the hairy math and there were no pictures or cartoons. ;-)

    HP/Agilent 5517 Mode Sweep Demonstrations

    A short PowerPoint show has been created that compares the mode sweep behavior of the 5517 laser with that of a common (non-Zeeman-split) HeNe laser having a similar cavity length, as well as a 5517 laser tube without a magnetic field. The plots have been crafted to be close in character to how they would actually appear for a like-new 5517B using an ultra-high resolution Scanning Fabry-Perot Interferometer (SFPI) with a polarizing beam-splitter and separate photodiodes for the F1/horizontally polarized and F2/vertically polarized outputs of the laser. With this specific example, the gain is high enough that in conjunction with the cavity length and magnetic field, all three regimes are present - Zeeman modes only, Zeeman modes and normal mode, and normal modes only. For a high mileage 5517 with lower gain, the plot would be mostly "hat" with little or no "skirt" (see above). However, note that for the F1/F2 split modes to be resolvable as separate lasing lines - or at all - a real SFPI would need to have a finesse of several thousand (corresponding to sub-MHz resolution) since their actual spacing is more than 2 orders of magnitude smaller than depicted - around 0.3 percent of the width of the neon gain curves rather than the few percent shown in the animation. Also incorporated are rough estimates of output (beam) power and split frequency (when present) of a typical 5517C laser. The individual frames of the animation of the 5517 laser are more closely spaced in the critical region where the Zeeman-split modes are present during the final locking process. (And of course real lasers have a continuous mode sweep without discrete frames!) The normal lock point (READY solid) for a 5517 laser is when the two polarized outputs are exactly equal. A number in the upper right corner of each slide shows the position of the cavity modes relative to the gain center with respect to the FSR of the laser (1.2 GHz for the 5517A/B/C/D and 5501B, about 1.5 GHz for the 5517E/F/G). The animation goes through two complete mode sweep cycles and then simulates locking ending with the Zeeman-split modes centered and LOCKED and REF ON. Of course, the locking process for an actual 5517 is just a wee bit more complex though the end-point is the same. And the final approach to the lock point is close to being critically damped, not the underdamped ringing depicted, but that would be too boring. :-)

    The slide show may be started by going to HP/Agilent 5517 Zeeman-Split HeNe Laser Mode Sweep Animation and runs in a separate window. ESC to exit. This is known to be compatible with PowerPoint 2007. Once it's stable, if ever :), a version compatible with PowerPoint 1997-2003 may be available. (while the present one is supposed to work with these, the animations fail to load in PP2003. I don't know if there's aproblem with what PP 2007 created, or my version of PP 2003, which has probably never been updated with bug fixes.)

    In addition to a very few explanatory slides, there are animations for a normal (non-Zeeman-split) HeNe laser tube such as that from a barcode scanner, a like-new 5517B laser tube, and a somewhat high mileage 5517C laser tube with no magnet. These all have the same cavity length. Unfortunately, testing the same 5517B tube with no magnet isn't likely to happen. :-)

    (If you're wondering what happened to the neon gain curves, lasing threshold, and other junk that used to be in the plots and shows, I decided (1) they were hard to draw, (2) didn't add anything useful, and (3) I couldn't figure out what to do with the gain curves beyond the lasing region anyhow. So they are history!)

    While viewing the animation of the normal HeNe laser, observe the following:

    While viewing the animation of the 5517 mode sweep, observe the following:

    While viewing the animation of the naked 5517 laser tube (no magnet), observe the following:

    All the frames of the normal 5517 mode sweep are also available in a separate PP show as HP/Agilent 5517 Zeeman-Split HeNe Laser Mode Sweep Sequence. Note how the output power - which is the sum of all the red and blue modes present as denoted as Total Power changes by less than 2 percent within the region where there is the single split longitudinal mode (frame IDs 925 to 075). (If drawn to scale, the separation between the two lasing lines would not even be visible.) Where there are two normal modes (red and blue separated by the FSR of the tube cavity), the output power is somewhat greater, possibly because the lasing modes are far away from each-other most of the time and thus have different sets of excited atoms to stimulate. End-of-life lasers will have much lower power within the Zeeman-split region but a more dramatic increase in power away from it since the gain relative to the lasing threshold is smaller.

    All the frames of the mode sweep of the normal tube (no magnet) are also available as HeNe Laser Mode Sweep: 127 mm (~5 inch) Cavity Length Showing Effect of Mode Competition.

    Enjoy! ;-)

    HP/Agilent 5517 Laser Construction

    The following also applies with only minor changes to the 5518A and the 5519A/B with differences noted where relevant. But I rather doubt that they (or the 5517A) have equivalent Type II or Type III Control PCBs like the small 5517s.

    These lasers consist of the laser tube assembly, potted (brick) HeNe laser power supply, beam sampler, Connector PCB, and Control PCB. mounted on a an metal chassis. Any of the parts can be replaced in under 5 minutes using common tools, with only minimal or no adjustment or alignment.

    Laser tube assembly:

    All of these consist of the actual glass HeNe laser tube potted with a rubbery material inside its Zeeman magnet, beam expander, and adjustable waveplates. The heater/cathode is attached via a 2 pin plug while the anode has its own single pin high voltage connector. The HeNe laser tube ballast resistance of about 100K ohms is conformal molded into the silicone insulated HV cable. The bifilar-wound heater inside the laser tube has a typical resistance (cold) of 8 ohms on most tubes. (The only exceptions I know of were early 5517Es, some of which may have even been a prototype, where it was between 4 and 6 ohms.) When at operating temperature, the resistance is spec'd to be higher by a constant factor since the actual temperature can be determined based on the known thermal coefficient of resistance of the heater wire.

    Only 4 screws hold the tube assembly to the chassis for lasers in the small cases (all the 5517s except the 5517A, as well as the 5501B). One or two will usually be flat head screws which provide either a fixed axis for horizontal (pan) alignment, or self alignment (no adjustment permitted). All of these tube assemblies appear physically similar, except for an early 5517E and the N1211A that are slightly shorter. (They, of course, differ with respect to the REF/split frequency.) The larger tube assemblies found in the 5517A, 5518A, and 5519A/B mount with 3 bolts and have machined alignment pins so no adjustments are needed or possible. They, too, are physically identical except for one small area of the casting that needs to be cut away if installing a 5517A tube into a 5519A/B laser to clear the internal ±15 VDC power supply.

    HeNe laser power supply:

    Very old (perhaps roughly pre-1990) 5517s (and 5501Bs) used the Laser Drive model 111-ADJ-1, Power Technology model 0950-0470 (same as HP part number) HeHe laser power supplies which had adjustable current via a pot on the laser Connector PCB, or an EMCO High Voltage model LP1600A. It is not known which of these was the first, or if they coexisted for awhile. However, based on how often they show up in used lasers, the 111-ADJ-1 was the only one in wide use. The adjustable power supplies were *only* found in thermally tuned lasers (5501Bs, 5517s, etc.) even though the service manual for these lasers NEVER suggests setting the current to anything other than the default 3.5 mA. (The power supply used in the 5501A and those that preceeded it also had adjustable current, but they were not stand-alone bricks.) (At least most of the 111-ADJ-1s were adjustable. I did find one that had the same part number but no third wire.) All later lasers use VMI power supplies with a fixed current of 3.5 mA. However, the pot is still present on the Connector PCB even on lasers made in 2013 (and probably to the present day), but normally serves no purpose. (It is useful if installing a 111-ADJ-1 to enable a high mileage sputtering tube to run stably at 4 mW. But I find it hard to believe that Agilent left there to accommodate the needs of hackers!)

    There are several versions of the VMI power supplies used in these lasers. The two oldest ones (VMI PS 148 and VMI PS 217) have the same HP part number of 0950-0470. The older Power Technology also has the same model number but would not be interchangeable as it will only mount in a 5517A or 5518A (or 5519A/B but predates that by at least a decade)! Then sometime after the year 2000, the power supply changed to the VMI PS 373 with Agilent part number of 0950-4459, which is found in most lasers up until around 2009. I know that switching from the PS 148 to the PS 217 reduced the residual current ripple from over 3 percent to well below 1 percent because I measured it. I do not know what changes were made in the PS 373 (but it also has very low ripple), nor what other differences there may be between these models. VMI claims all of this is proprietary information. Can you believe that? :) However, I have de-potted a dead PS 373 as shown in Photos of VMI PS 373 HeNe Laser Power Supply and De-Potted Components and reverse engineered its schematic. See the section: VMI 373 HeNe Laser Power Supply. Another version is the VMI PS 253 (Agilent part number 0950-4073), found in several late model lasers including the 5517FL/GL and N1211A. It is potted inside a metal case with removable lid so the appearance is unmistakable. See Photos of VMI PS 253 HeNe Laser Power Supply and Partially De-Potted Components. Preliminary dissection indicates that it is similar to a PS 373 but with a modified PCB layout, an additional filter capacitor and inductor, and some minor changes to the ripple reducer and high voltage resistors. The PCB is actually slightly larger than that of a PS 373 with additional cut-away sections, presumably to prevent arc-over. It is screwed to the metal case in three places providing direct grounding, and thus eliminating the need for the aluminum shield plate often found bolted to the older bricks. I'm not planning on fully reverse engineering the PS 253 but may do some spot checking to identify any major differences. The most recent power supply I've seen is the VMI PS 504 Agilent part number 0950-5216), which is very similar to the PS 373 and may simply be its replacement. It is believed that all of these power supplies have similar (default) current and voltage specifications and are fully interchangeable among 5517, 5518A/B, 5519A, and 5501B lasers (except for mounting in some cases). Here is a summary:

        Manufacturer       Model       Part No.   Comments
          EMCO HV         LP1600A        None     Early, perhaps first for 5517A
      Power Technology   0950-0470     0950-0470  Mounting for 5517A, 5518A/B only
        Laser Drive    101-1500-3.5-HP   None     Successor to EMCO, fixed
        Laser Drive      111-ADJ-1       None     Successor to EMCO, adjustable
            VMI           PS 148       0950-0470  First VMI supply
            VMI           PS 217       0950-0470  Lower ripple than the PS 148
            VMI           PS 253       0950-4073  Shielded case 0.1% ripple RMS
            VMI           PS 373       0950-4459  0.1% ripple RMS or less
            VMI           PS 504       0950-5216  0.1% ripple RMS, Latest in 2014
            VMI           PS 503       0950-5221  0.01% ripple RMS,Latest in 2014

    As far as I am aware, all of these power supplies have basically the same compliance voltage specifications and are thus electrically compatible with all HP/Agilent 5517/5518/5519 and 5501B tubes. Except for the Power Technology 0950-0470 and Laser Drive 111-ADJ-1 (which are adjuatable from around 3 to 4 mA via a trim-pot on the Connector PCB), they have a fixed nominal current of 3.5 mA and a voltage compliance range that more than accommodates long and short tubes over their entire life (as the tube voltage typically increases). (Some Laser Drive 111-ADJ-1s have no third wire and are thus similar to the 101-1500-3.5-HP, go figure.) Exactly why 5517 lasers built in 2014 still have the current adjust trim-pot on the Connector PCB is a mystery - it does nothing and has done nothing for at least 20 years and adds to the cost! Where are the bean counters when most needed? Only the 5519A/B laser has done away with it but should a need arise to install an older power supply, then all that is needed would be a 3.1K resistor in series with a 10K ohm trim-pot (or suitable fixed resistor) between the top two pins of the connector. The PS 503 and PS 504 supplies are the standard ones used in 2014. The PS 504 is used in most lasers but the PS 503 is designated as a repair part for the 5517CL and 5517DL.

    CAUTION: While the voltage and current specifications for the PS 253, PS 503, and PS 504 are the same as the others, those (and any other metal-case power supplies that may appear), CANNOT be used to replace the brick in the 5501B unless totally insulated from the laser chassis because the power supply case is connected internally to its negative input. The 5501B uses -15 VDC and GND as the inputs to the brick to decrease the load on the positive supply for compatibility with the dinosaur 5501A. So, -15 VDC gets shorted to GND when one of these is plugged in. :(

    For a long time, the only defective power supplies I've ever found in HP lasers were nearly all Laser Drive 111-ADJ-1s. And one type of failure may result in the adjustment pot having no effect with the power supply pumping way excessive current (like 6 or 8 mA) through the tube. With luck, the ballast resistor catches fire and explodes before the tube is damaged. :( :) (However, these defective bricks are quite useful as experimental supplies for driving tubes from below 0.5 mW up to a 2 or 3 mW by controlling input voltage.) I've also see a few PS 148s that had excessive ripple, so something in its output filter had blown. But if I hadn't been checking ripple on a bunch of these power supplies, it probably would have gone unnoticed; As long as the tube stays lit, performance of a healthy laser probably wouldn't be affected in any significant way, though the MEAS signal might have a bit more fuzz on a scope display due to the current ripple producing amplitude ripple in the laser output power. However, the additional ripple would make the effective dropout current go up, so a marginal tube might start sputtering on that supply. But now having tested several dozen late model lasers (post 2004), several defective PS 373s have turned up. They either draw no current, have lost regulation, or draw excessive current. Some also seem to be very sensitive with respect to input - they will start and run fine in a laser but may refuse to start reliably or at all if powered from a bench supply. Or vice-versa. Go figure. And on a few, while the output current remains well regulated, it has dropped to around 3.41 or 3.42 mA - or was set that way at the factory due to bad quality control! Reduced current would normally not matter very much, but could result in a premature failure should the tube's dropout current increase, as it often does after long hours of runtime. Now Agilent wouldn't want their lasers to fail early, would they? :-) So perhaps the PS 504 is simply a more reliable replacement for the PS 373.

    It's straightforward to modify the PS 217, 253, and 373 power supplies to be able to adjust the operating current. Most are potted in a soft rubbery material which can easily be removed precisely at the location of the trim-pot that sets the current. This should also be possible for the VMI PS 503 and 504, whose construction is similar to that of the 373. However, the trim-pot locations may differ, though probably not by much, and they may use tougher potting material since they are newer, see below. The others are potted in a hard material which would be virtually impossible to remove without destruction of the power supply. (But I do know someone who can probably do it if the future of the Universe was at stake.)

    However, recent versions of the VMI PS 373 (SNs somewhere above 10,000) are filled with a tougher potting compound that is more difficult to remove without damage to the underlying components. It's still rubbery but might require more extensive and gentle surgery. (This probably also applies to newer supplies as well.) So, the following procedures may need to be modified. I just haven't been desperate ehough to try.

    Any type of tool that can be used to careful excavate out the potting material above the trim-pot location will be satisfactory. I use the remains of a tire pressure gauge. The thin-walled metal tube is the perfect size to press into the rubber and extract a "core". ;-) Take care not to damage anything on the PCB - Don't go all the way down and don't rock it back and forth near bottom. Push it about one half the way through and remove it. With luck, a partial core will come with it that can be used to somewhat seal the hole once done. Then carefully pick away at the rubbery material to expose the top of the trim-pot. A small flat blade screw driver may be used to adjust it. Take care not to touch any circuitry during the excavation or while doing the adjustment. And don't be too forceful as that can rip the trim-pot off its solder pads. However, even if that happens, all is not lost. It's possible to cut away the side, excavate the area, and replace the 25K ohm trim-pot entirely. Only two pads (nearest the edge in the PS 373) need to be connected. Since it's low voltage in that area, there is little risk.

    In most, a very small multi-turn trim-pot is used for current where the direction to increase current is counter-clockwise. However, some had a single turn trim-pot where clockwise increased current. And additional variations would not be surprising. So, it's essential to monitor current while performing the adjustment. The effect of the voltage limit trim-pot has been tested on only one VMI PS 253 and it increased clockwise.

    Note that the voltage limit trim-pot normally can be left alone on VMI PS 373s when increasing current. However for the VMI PS 253 and other metal-cased power supplies - where it is necessary to increase current significantly - the tube may flicker or sputter and never "catch", especially when restarting hot. This also depends on how quickly the input voltage ramps up. There are three known solutions that may be satisfactory:

    1. Increase the voltage limit: 1/8th turn (45 degrees) will usually suffice. In theory, setting the voltage limit too high could eliminate some protection for out of tolerance conditions, though it is not clear what these may be. ;( ;-) So rotating it to the stop is probably OK. But if tue Universe explodes because of this, I don't want to knw about it. ;-)

    2. Add 2 or 3 silicon diodes in series with the 15 VDC input: For some reason, this tends to make starting more reliable and the input voltage is still high enough for it to be reliable.

    3. Wait to power up: If the problem only occurs when restarting hot, just be patient and give it 5 or 10 minutes. ;-)

    The high voltage connectors are from AMP/TE Connectivity LGH/LKH series and may be:

    Beam sampler:

    These consist of a first non-polarizing angled plate to sample a portion of the output beam and a second non-polarizing angled plate to take this and split it between the LCD switch with a photodiode behind (above) it mounted on the small PCB, and the REF photodiode with a polarizer at 45 degrees in front of it. The LCD switch attaches to the Control PCB via a 4 pin connector - 2 pins for the LCD drive and 2 pins for the photodiode behind it. The REF photodiode is actually mounted on the Control PCB and simply pokes its head into the beam sampler assembly. Beam samplers for the 5501B and all small-case 5517s are identical and interchangeable. The beam sampler for the 5517A, 5518A, and 5519A/B is optically identical but the housing differs enough that cannot be swapped without modification. But the only part that is likely to go bad is the LCD switch, which is the same for all and can easily be replaced. For a laser with a healthy tube that doesn't lock, a defective LCD switch is the most likely cause. They can delaminate resulting in a portion of the switched area not responding. The LCD panel is available from the Keysight Parts Online Store for *only* $62 in 2024. It is part number: 5088-1474. But considering that the HeNe laser power supply brick is over $2,000 from them, I guess $62 isn't too bad. Registration is required to buy parts but it is free. And shipping is also free for these at least. ;-)

    Connector PCB:

    Aside from the Mil-style connector to the outside world and the 24 pin connector to the Control PCB, this has some filter capacitors; fuses for +15 VDC and -15 VDC; and the Power, Laser On (really same as Power), and READY LEDs. The one pot does nothing except for really old lasers with the Laser Drive 111-ADJ-1 HeNe laser power supply brick.

    Very old lasers had a case interlock switch to disable the laser tube from being powered if the covers were removed, and a "Service" switch to override this. :) Both of these switches have been eliminated, though the PCB pads and wiring for them are still present, but bypassed. It's worth removing both switches on lasers that have them and adding the required jumper diagonally between the center pads that are closest together on both switches. (Do NOT just add the jumper - the switches must be removed!)

    HP-5517A Connector PCB with Interlock and Service Switches shows the component side of a typical sample. This is from a 5517A laser, but those from all the other 5517s have a similar set of parts, though the shape and layout differ for the "small case" 5517B/C/D/E/F/G lasers. It has a "05518-60001" part number even though it's from a 5517A. Their Connector PCBs are identical. (The Connector PCB for the 5501B is quite different, as are those for the older HP lasers.)

    The third leg on the large silver electrolytic capacitor on the Connector PCB is for mechanical support only. It can be replaced with a common cap of at least equal µF and voltage rating. It will be smaller and lighter and the lead wires will be sufficient support. The use of 105 °C types may be desirable but probably not essential.

    Control PCB:

    After removing the cover on one of these lasers, the most obvious assembly aside from the tube is the PCB which controls the heater inside the laser tube based on inputs from a pair of photodiodes - Mode (X and Y selected by the LCD switch) and REF. The 5517 laser Control PCBs are known as the "A3 Control/Reference Board" in HP/Agilent manuals.

    Since the introduction of the 5517A laser through the early 2000s, all the these lasers used essentially the same Type I Control PCB (designated the "A3 Control/Reference Board" by HP). There are slight differences between those for the large-case lasers (5517A, 5518A, and 5519A/B) and small-case lasers (5517B/C/D), but these are mostly physical. The Type I Control PCB was based on SSI TTL logic for timing and an analog feedback loop. The A3 board in the 5517A is physically larger and not interchangeable with those in the small lasers, but is nearly identical electrically. The main functional difference is the addition of a second line driver and a connector so it can provide the MEAS signal used in 5518A and 5519A/B lasers. (Interestingly, the Type I Control PCB also has a second line driver, normally never used, though it can be jumpered to provide a second REF output.) There had been virtually no change in the design over 15 years or more, except that a hand-soldered modification to the internal REF receiver makes newer lasers require somewhat higher optical power to lock than older ones. This is probably to avoid false locking when there may be substantial ripple from the HeNe laser power supply or due to plasma oscillations in the laser beam. There are no useful indicators on the Type I Control PCB, only one wimpy LED that duplicates the function of READY on the backpanel. There are no LEDs on the Control PCB in the 5517A.

    But since sometime after Agilent was created, at least two new versions of the Control PCB have appeared. FPGAs/FPLDs have replaced the TTL in the Type II Control PCB and a reasonably high performance microprocessor or DSP is at the heart of the Type III control PCB. However, there are still some analog parts. If one wants to count transistors, I bet the Type III Control PCB has over 1,000 times the number of transistors as the Type I Control PCB! These are also almost entirely surface mount (SMT), with major parts on both sides of the PCB in the that version.

    The first of these, the Type II Control PCB (for both size lasers), appeared in lasers manufactured around 2004. The standard version is electrically and physically interchangeable with the older Type I Control PCB and is now the most common by far. It is based on a Xilinx XC2S50 Spartan-II FPGA which replaces the discrete TTL state machine and most of the logic, and everything else is in more modern (and available) SMT parts. While very reliable, a failure for any reason other than an obvious problem like a blown fuse or bad DC regulator with no underlying cause would likely render it non-repairable except by Agilent or an authorized service center since it's then just a black box with no real way to easily troubleshoot. A service manual may exist but I've never seen one. And even if it did, sophisticated test equipment and a surface mount rework station would be required to have any chance at repair. However, this version has all the same jumpers and temperature set pot, so normal testing and adjustment is similar to that of the Type I Control PCB. The easiest solution would then be to simply swap in a known good board (either version), of which there should be plenty available in lasers with bad tubes.

    It's not clear what, if anything, the Xilinx-based controller adds to the laser, other than to make it more proprietary and difficult to service. After all, features are not being constantly changed or added, nor will there be security issues due to computer viruses - it doesn't run Windows! :) So, periodic firmware upgrades and bug fixes really aren't required, which is a primary benefit of using Field Programmable Gate Array (FPGA) technology like that of the Xilinx parts. The Xilinx XC18V01 PROM can be reloaded but I kind of doubt this would ever be necessary! However, there is no doubt that the discrete through-hole TTL and analog parts dating from the 1970s are becoming more difficult to find and expensive. So the Type II Control PCB is now used for all the common Agilent 5517B/C/D lasers. Swapping in a Type I Control PCB results in no obvious differences in performance, and that would be my suggested method of repair unless there were special requirements. And 5517B lasers with these Control PCBs have been turning with 2004 and later manufacturing dates, probably removed from service in wafer fabs after degrading to the edge of Agilent specs for REF frequency, or a specific number or hours of service. An identical laser from the same source and a manufacturing date of 2003 had a Type I Control PCB. So, that may mark the transition to the newer technology. However, for lasers with Short tubes and/or a REF frequency above 4 MHz like the 5517E, a special version of the Type II Control PCB may be required (or at least specified, though possibly not essential).

    All the jumpers and their approximate locations are identical and the time spent in the major states and time to lock (READY on solid) are about the same as with the Type I Control PCB. The behavior while warming up and after locking is indistinguishable from that of the Type I Control PCB so it really isn't even possible to determine which one is inside the laser without removing the cover. However, there may be subtle differences in the effects of changing jumpers on function and test-points, so it's probably best to do this with the laser powered off. The only obviously similar electronic component common to the two is the large white film capacitor for the feedback integrator, and perhaps the heater driver transistor. So, it's likely that the objective of this redesign was simply to eiliminate all the older SSI/MSI TTL logic and other obsolete through-hole parts, but that it is functionally identical to the Type I Control PCB with essentially the same logic inside the Xilinx FPGAs and linear circuitry in SMT ICs.

    More recently, I discovered another type, called the "Type III Control PCB". It was first found in a 5517E which based on IC date codes is from around 2003. At the time, I had never heard of a 5517E, and it wasn't on the Agilent Web site - or anywhere else bsides Sam's Laser FAQ! I thought the Type III Control PCB might be unique to the -E version. But I have since also seen one in a 5517D-C29, manufactured in 2004 and a 5517D-C15 manufactured in 2001. Thus, it's not a later revision, but possibly a special request. :) What is common to all these lasers is relatively low actual or labeled output power (80 to 120 uW) and high REF (above 5 MHz). The Type III Control PCB seems to be a total redesign, with no effort made to be at all similar to the Type I Control PCB with its jumpers and test points, or even in the specific algorithm it uses during warmup and locking. In addition to no jumpers, test-points, or adjustments, it takes 2 to 3 times as long to lock, and has an RS232 port. :) More below.

    Here are the three types of Control PCBs used in the 5517B/C/D/E lasers.

    As noted, the one in the 5517A differs slightly in form factor and has a small amount of extra circuitry for use in the 5518A and 5519A/B. (The original HP and now Agilent part numbers actually begin with 05518.) The Agilent Web site lists a Type II version of this board, though I have never seen one in an actual laser.

    The Type I and Type II Control PCBs function in a virtually identical manner, requiring about 2 minutes for the READY LED to start flashing, and another 2 minutes to come on solid. And as noted, they also have more or less the same jumpers and test-points, as well as the temperature set-point pot. The Type II Control PCB may in fact simply be essentially an emulation of the Type I Control PCB using an FPGA and more modern surface mount parts. That same large integrator capacitor is present, though the smaller sample-and-hold caps are a different type.

    The Type III Control PCB is not at all similar to the others. It has none of the same jumpers and several different test-points, an unused connector and a large unpopulated header (functions unknown), and no pots at all. It does have an RS232 port no doubt for setup and testing and almost certainly for access to a digitally-maintained run-time meter. There are also a pair of micro-DIP switches - and a pushbutton, which I fianally dared to push, and as expected, seems to be master reset. :) It is based on a SHARC processor with many digital and analog SMT parts as well as large Lattice FPGAs on both the front and back of the PCB.

    Here is a summary of the HP/Agilent part numbers for the known three types of Control PCBs:

     05518-60003 5517A/5518A/5519A/B Type I Control PCB (HP)
     05517-60003 5517B/BL/C/CL/D/DL Type I Control PCB (HP)
     05519-60004 5517A/5519A/B Type II Control PCB (Agilent)
     05517-60031 5517B/BL/C/CL/D/DL Type II Control PCB (Agilent)
     05517-60131 5517B/BL/C/CL/D/DL Type II Control PCB (Replaces 05517-60031)
     05517-60132 5517FL Type II Control PCB (Modified for high-REF lasers?)
     05517-60025 5517DL/E/FL Type III Control PCB (Early one for high-REF lasers?)

    (The "60" of the part number will appear as "68" on the barcode sticker and "20" on the PCB copper.)

    For more, on the Control PCBs and their operation, see the sections: HP/Agilent 5517 Laser Control PCBs and Locking Sequence, HP-5517E/F, and Agilent 5517 Laser RS232 Communications.

    The entire purpose of redesigning the controller more than once is somewhat perplexing, though in retrospect, they probably weren't done sequentially since the Type III seems to actually predate the Type II. Perhaps the Type III was done for some special application or with the intent of it replacing the Type I, but then found to be too expensive and difficult to manufacture for general use. Given the likely relatively costly components including the SHARC processor and Lattice FPLDs, this might not be surprising, even given the high cost of the entire laser. The same very limited inputs (a pair of photodiodes sensing the modes through the relatively slow speed LCD switch and another photodiode behind a polarizer generating the REF signal) and outputs (tube heater current) are used in all 5517 lasers so it wouldn't seem to be possible to implement a significantly higher level of frequency accuracy or stabilization no matter what sort of control scheme is used. About the only thing that might be done is to actually compute the REF frequency from the REF photodiode signal and fine tune the lock position to maximize it once the basic stabilization using mode balance has been achieved. The peak of the REF frequency function may be a more accurate means of locating the Zeeman-split gain curve center. But except for NIST-level precision, the analog method is really just fine, so even if this scheme were implemented, it's not clear what customers would require it. And from my observations of the REF frequency while locking, it doesn't seem to make any effort to maximize it, but stabilizes at a point much lower than the peak, with the same sort of slow variation once locked as the Type I Control PCB. And the Type III Control PCB is not all that common. Of the more than 100 5517 lasers I've probably seen, only 4 used it.

    So, a combination of several explanations make the most sense:

    1. Components used on the Type I Control PCB were likely becoming obsolete so that some (though likely trivial) redesign would have been needed in any event. But that's taken care of by the Type II Control PCB.

    2. Basing the controller on a digital platform makes it easy to set up and test in the factory. Those extra connectors must have a purpose! (And we now know that one is for the RS232 port.)

    3. Basing the controller on a digital platform makes it virtually impossible for anyone other than Agilent to repair or adjust (such as when a laser tube is replaced). The RS232 port is no doubt part of this, though the unidentified connector next to it probably provides additional status information. Of course, with the amount of computing power on this thing, it should be able to determine the relevant parameters of any laser tube without any need for calibration (and predict the weather at the same time!).

    4. Digital anything is "in" and the complex appearance helps to justify the five figure price tag. :-)

    As of 2014, I've only found four (4) lasers with the Type III PCB, all with manufacturing dates well before 2005. Other things they had in common were a high REF frequency and low rated output power. So perhaps there was some reason why the Type I Control PCB was considered unsuitable when these lasers were first produced. All other lasers I've seen from 2004 or later use the Type II Control PCB.

    If anyone has more information on these Control PCBs, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    Also see the section: Common Problems with HP/Agilent 5517 Lasers.

    HP/Agilent 5517 (and 5501B) Glass HeNe Laser Tube

    The glass HeNe laser tube used in all the 5517s as well as the 5518A, 5519A/B, and 5501B is very well made but not particularly unusual (at least based on a casual examination) except for a heater coil wrapped around the bore inside the glass envelope, used to precisely control cavity length. However, the design is heavily optimized for this specific application. Here are some photos and detailed diagrams of the internal construction of the older "Long" thermally tuned HP/Agilent laser tubes. (The newest "Short" tubes are described later.)

    Long-LV and Long-HV tubes

    This general design dates from the first thermally-tuned laser tubes used in the original 5501B and 5517A in the early 1980s through sometime before 2012 depending on the specific model.

    In terms of cavity length, the 5501B and 5517 tubes are very similar to typical 1 mW barcode scanner HeNe laser tubes. Compared to those (which can exceed 1.5 mW when new), the HP/Agilent tubes are rather wimpy. Yet, the discharge length for even the Short HP/Agilent tube is longer than that of the typical barcode scanner tube. For higher REF frequency lasers, some of this discrepancy may be accounted for by the lower OC mirror reflectance and reduced gain curve overlap required to achieve the desired Zeeman beat frequency range for each model. This is consistent with the generally lower output power for higher REF frequency lasers like 5517Ds which almost always have much lower power than 5517As when new. This is even more dramatic with 5517E/F/G lasers where the output power when new may be under 100 µW! But the 5501A with an OC mirror reflectance similar to that of barcode scanner tubes and large gain curve overlap still never has an output power anywhere near 1 mW and it is rarely over 500 µW. The cause may be related to the near hemispherical cavity geometry and relatively wide bore of the Long tubes and all that preceeded them. (The cavity of the Short tube is similar to a barcode scanner tube so should have similar output power.)

    Unlike most other modern internal mirror HeNe laser tubes, the Long tubes have no mirror adjustments. The mirrors are held in place against the thick glass bore (or mirror spacing rod as HP calls it) by spring pressure alone. So, the ends of the bore and mirrors must be ground to a precision sufficient for alignment to be near perfect. The distance between the mirrors (that I've seen) is 126, 127. or 132.5 mm depending on the date of manufacture of 5517A/B/C and 5501B lasers. This corresponds to an FSR or longitudinal mode spacing of about 1.19, 1.18, and 1.13 GHz, respectively. The purpose of having multiple glass backing disks behind the OC mirror is not really known, but in conjunction with the springs, provides a means of setting the mirror spacing rod axial position fairly precisely. disks.

    And slight movement of the mirror spacing rod/bore within the outer glass sleeve is clearly evident from even relatively minor tapping on the ends of a bare glass laser tube. But how much this would actually affect performance is unclear as long as the path for the anode-end discharge through the capillary isn't cut off. Unless the bore got stuck and one of the mirrors was then actually loose, the cavity geometry would be unchanged and the only effect would be that the distance to the beam expander would differ by at most a few mm, which should not account for a dramatic reduction in output power. Compare the top two tubes in X-ray Views of Typical Long-LV (5501B), Long-HV (5517C), and Short (5517D) HP/Agilent HeNe Laser Tubes.

    I've heard from someone who used to work at Agilent that bare tubes are handled VERY CAREFULLY until they are installed in the magnet assembly. This is probably the reason why. Whether any in-situ alignment is ever actually performed during manufacture is not known. But perhaps that's how it's done - by whacking with a BIG hammer! :) Steinway and Sons has a "Pounding Room" where new pianos are broken in. Perhaps Agilent has a "Whacking Chamber" where new lasers are aligned. :)

    Originally, I was expecting the low REF frequency 5517A to have some obvious difference compared to the high REF frequency 5517D, but except for the small change in mirror spacing rod length going from 132.5 mm to 127 or 126 mm, this does not appear to be the case. Since then, it has become clear that the primary determining factors of REF frequency are mirror reflectivity and magnet strength, with a small contribution from cavity length. But the likely reason for the shorter cavity length is to allow for a higher magnet strength before rogue modes appear.

    I have also checked a working 5517D using a Scanning Fabry Perot Interferometer (SFPI) to measure the mode spacing during warmup (and thus cavity length based on c/2f) and confirmed what I had already concluded from physical examination - that a 5517D is the same as a 5517B with a similar date of manufacture (1.18 GHz, 127 mm). So, if they are all virtually the same, what is the purpose of the extra space in front with 5 backing disks and the extra length spring? The most likely explanation is still that the selectable number of backing disks and long spring provides a more precise means of fine tuning the mirror spacing rod position.

    Here is a summary of the number of OC backing disks (#OC BDs) I've found searching through my inventory of (mostly dead) HP/Agilent laser with Long-LV or Long-HV tube assemblies:

       Model   Year   #OC BDs   Comments
       5501B   1987      0      Segmented Magnet
       5501B   1989      0      Segmented Magnet
       5501B   1991      0      Segmented Magnet
       5501B   1993      0
       5501B   1994      1      Segmented Magnet
       5501B   1994      1
       5501B   1994      1
       5501B   1995      1
       5501B   1996      1
       5501B   1996      1
       5501B   1997      0
       5501B   1998      0
       5501B   2005      0
       5501B   2005      0
       5501B   2006      0
       5517A   1994      1
       5517A   1996      1
       5517B   1988      0      Segmented Magnet
       5517B   1988      0      Segmented Magnet
       5517B   1992      0      Segmented Magnet
       5517B   1995      1
       5517B   1995      1
       5517B   1995      1
       5517B   1996      1
       5517B   2004      0
       5517B   2005      0
       5517C   1991      1
       5517C   1994      1
       5517C   1994      1
       5517C   1995      0
       5517C   1996      1
       5517C   1997      0
       5517C   2000      0
       5517C   2003      0
       5517C   2003      0
       5517C   2004      0
       5517C   2004      0
       5517D   1994      0
       5517D   1997      0
       5517D   2001      0
       5517D   2004      0
       5519A   2000      0
       5519A   2001      0

    Of possible note is the absense of an HR backing disk on any laser later than 1996. Perhaps the spring behind the HR has been changed around that time to be longer or stiffer, or someone found that fewer OC backing disks would compensate. And there are none seen before 1994, which is perhaps when the tubes changed from Long-LV to Long-HV, and also when the segmented magnet disappeared. (More below.)

    The heater connections with red and purple wire stubs sticking out can be seen at the left of the tube. The purple one also attaches to the cathode via a piece of springy sheet metal - no welds. The anode connection goes through the outer glass envelope but there is no glass seal into the bore, simply a hole drilled in it to coincide with the anode location. Even after totally disassembling multiple tubes, it's not clear what prevents the discharge from bypassing the bore - no trace of any kind of sealant has ever been seen. The fit between the bore and surrounding glass cylinder is quite close but even this wouldn't normally guarantee that the discharge goes through the bore, especially on hard-start tubes. After all, a high voltage source like a Tesla coil is often used to locate micro-cracks in glass or ceramic vacuum systems. The reason this works is that the mean free path within a tight-fitting (but not sealed) joint like this is too short to allow a buildup of current. The Alnico magnet in newer lasers is a cylinder 4" (L) by 2" (OD) with wall thickness of 1/4". The inner diameter is just a few mm larger than the tube. The magnets in older lasers have the same dimensions but are made up of 4 equal length segments. The magnet extends at the left and right ends of the tube to approximately where the discharge begins and ends, at least with the longer tubes.

    Short tubes

    The Short tube represents the most significant change in design since thermally-stabilized lasers were introduced around 1983. Several modifications enable HeNe Zeeman technology to achieve greater REF frequencies while maintaining adequate output power and similar long life. See Agilent 5517 "Short" HeNe Laser Tube for a photo, X-ray, and diagram of a typical sample. :) The Short tubes have a mirror spacing rod that is unsupported at the front (output end) with the OC mirror attached to its face with a of cage device and adhesive. It is believed that the cage enables the OC to be adjusted in some type of optical alignment jig before the tube is sealed. At the back, the mirror spacing rod is rigidly attached to the glass envelope and a metal anode cylinder assembly. The HR mirror is attached what what appears to be glass frit to a post protruding from a thin disk inside the metal anode cylinder. This allows for fine alignment after assembly by deforming the disk using a special tool - essentally a narrow insulated rod. The discharge passes by the HR mirror so the narrow bore can start quite close in front of the mirror thus maximizing its length. The same bifilar-wound resistance heater is present, but extending only about half the length of the one in the Long tube and using thinner wire to maintain a similar resistance. The cathode and heater connections are similar to those of the Long tubes. The Short tube laser appear to use the same magnet as the longer ones even though the discharge ends far inside it, at least at one end. The shorter cavity in itself boosts the REF frequency and also permits a higher magnetic field to be used without producing rogue modes. By pulling out all the stops, so-to-speak, nearly every aspect of the Short tube's design has been optimized enabling the maximum REF frequency for Agilent 5517 lasers to nearly double compared to the original 5517D while maintaining acceptable output power and similar lifetime.

    Long and Short tube voltage:

    I have measured the operating voltage at 3.5 mA of several laser tubes in their original magnet, or if only a bare tube, corrected for a drop of 10 to 20 V when installed in a magnet. These are ordered within groups by increasing tube voltage:

                                Tube     |<-- Voltage -->|
      Laser/Tube Model          Type      Total      Tube     Comments
      New reject 5517D (2012)   Short   1.434 kV   1.084 kV   This group has the
      Like New 5517D (2013)      " "    1.490 kV   1.140 kV    latest tubes.
      New/NOS 5519B (2017)       " "    1.510 kV   1.160 kV
      Like New 5517C (2001)    Long-HV  1.544 kV   1.194 kV   This group has most
      Healthy N1211A (2006)     "   "   1.558 kV   1.208 kV    tubes from around
      Like New 5517D (2009)     "   "   1.560 kV   1.210 kV    1990 to 2009.  The
      Healthy 5517C (2006)      "   "   1.575 kV   1.225 kV    typical operating
      Weak 5501B (2006)         "   "   1.578 kV   1.228 kV    voltage is about
      Weak 5517B (2004)         "   "   1.587 kV   1.237 kV    50 V higher than
      Weak 5517C (1997)         "   "   1.596 kV   1.246 kV    for Short tubes.
      High Mileage 5517D        "   "   1.600 kV   1.250 kV
      Weak N1211A (2006)        "   "   1.616 kV   1.266 kV
      Weak 5517B (1995)         "   "   1.624 kV   1.274 kV
      Weak 5501B (1994)         "   "   1.626 kV   1.276 kV
      Very weak 5501B (1998)    "   "   1.627 kV   1.277 kV
      High Mileage 5517C        "   "   1.630 kV   1.280 kV
      Very weak 5517B (1995)    "   "   1.631 kV   1.281 kV
      High Mileage 5517A        "   "   1.638 kV   1.288 kV
      Declining 5517B (1988)   Long-LV  1.398 kV   1.048 kV   This group has very
      Very weak 5501B (1987)    "   "   1.448 kV   1.098 kV    old long tubes.
      Very weak 5501B (198?)    "   "   1.471 kV   1.121 kV    Tube voltage about
      End-of-Life 5501B (198?)  "   "   1.474 kV   1.124 kV    50 V lower than for
      End-of-Life 5501B         "   "   1.498 kV   1.148 kV    Short tubes.


    All of these have somewhat greater voltages than would be expected for a laser with a power output of less than 1 mW. Originally, I thought this might be due to the magnet. But installing a tube in a typical magnet actually reduces the total voltage by 10 to 20 V.

    The one labeled "Declining 5517B" is strange. It appears healthy in all respects - starting, running, discharge color, and performance. But the power was steadily going down hour by hour. It is quite old, possibly one of the earliest 5517Bs from 1988, and has the segmented magnet that until now I'd seen mostly on 5501Bs and 5517As of similar vintage. (The other 2 tubes in this group also have segmented magnets.) The laser was found to have a bad beam sampler, so I thought that it may have been taken out of service due to not locking. However, the locked output power started out at around 415 µW after warmup, and I figured this would be a nice laser. But the output power had decreased by almost 25 percent after running for a few days. Though the rate of decline was getting smaller, even at 325 µW, it was still falling at 1 to 2 µW per day. (Got all that?!) Thus, another possibility is that it may have been replaced during preventive maintenance due to the decrease in power, which for a tube with a typical new power of 600 µW, could be at a threshold of around 400 µW. And the beam sampler may have gone bad sitting on the shelf for many years. However, the REF frequency behavior is also strange. It has increased from 2.04 MHz to 2.25 MHz as the power has gone down. That alone is to be expected, but the initial value of 2.04 MHz is way lower than in any unmodified 5517B laser I've ever tested, or even the label for any 5517B laser I've seen. (Most are between 2.2 and 2.3 MHz.) If the output power were to ever level off, perhaps the REF value would end up more like that of a new laser! But a simpler explanation for the low REF is that the magnetic field has declined over the years. Accurate field measurements can't be made unless the tube is removed from the magnet, and even then a somewhat lower value for the magnetic field really would not be conclusive as there is always some variation. And here's the really strange part: If allowed to sit unused for awhile, the output power will increase a small but statistically significant amount. After 2 months, it peaked at over 353 µW! I'm now running it periodically just long enough for the power to peak. Here is the data so far on the recovery:

          Date        Power      REF
      10-Apr-2014    415 µW    2.04 MHz
      18-Apr-2014    324 µW    2.25 MHz
      27-Apr-2014    328 µW    2.23 MHz
      19-May-2014    337 µW    2.22 MHz
      30-May-2014    338 µW    2.21 MHz
      20-Jun-2014    353 µW    2.16 MHz
      19-Jul-2014    373 µW    2.12 MHz
      21-Aug-2014    383 µW    2.09 MHz
      20-Sep-2014    384 µW    2.09 MHz
      19-Oct-2014    386 µW    2.08 MHz
      16-May-2015    398 µW    2.05 MHz
      24-Sep-2015    412 µW    2.03 MHz
     *24-Sep-2014    419 µW    2.01 MHz

    After a bit over 15 months being off except for testing, power and REF are back to almost where they started. This would suggest that there is some type of gas-fill/contamination problem but it's difficult to come up with a scenario where the tube voltage isn't significantly affected.

    However, the latest tests were done just around the time I was investigating issues with fractured mirror spacing rods. Lasers with broken rods could still have satisfactory performance but the power and REF might be erratic. This laser doesn't quite meet those criteria, but there is currently no satisfactory explanation for its behavior either. And, indeed the basic test for a fractured rod - pressing on the glass protrusion at the back of this tube - did result in the locked power varying between 380 µW and 436 µW. While not as dramatic as in some instances, this is way more percentage change than is usually present in undamaged lasers. After the manipulation (wiggling and tapping), the locked power peaked at around 419 µW with a REF of 2.01 MHz. I do NOT believe a fractured rod is the cause of the declining power, but it is a complicating factor. And the slight increase from 412 µW to 419 µW may not even be due to damage, but just the rod having moved longitudinally slightly, which is possible with any Long tube laser.

    Originally when I acquired this laser, I thought the low tube voltage (around 150 V below that of a typical more recent new 5517B) was related to the declining output power and a symptom of gas contamination, despite the discharge color and brightness being perfect. But then I measured an end-of-life 5501B from 1987 and its tube voltage was consistent with having been a similar low value when healthy. A 5501B tube from 1997 had a tube voltage similar to that of newer (Long-HV) 5517s. So, really old tubes had a slightly different design. All of the tubes with a voltage below 1.15 kV also had a segmented magnet, indicating that they are quite old even if their exact manufacturing date is not known.

    In fact, it appeared as though not only was something different about the active discharge length or bore diameter, but the distance between mirrors was slightly different for the Long-LV and Long-HV tubes as well. Using a Scanning Fabry-Perot Interferometer, the longitudinal mode spacing for a Long-LV tube was found to be approximately 1.13 GHz (for a mirror spacing of 132.5 mm) compared to 1.18 GHz (and 127 mm) calculated and measured for a Long-HV tube. This difference is large enough to be real, not a measurement error. However, with respect to tube voltage, it went the wrong way! The Long-LV tubes with their lower operating voltage have a larger mirror spacing! So the mirror spacing, discharge length, and possibly other design parameters like the bore diameter, were changed. The primary rational for decreasing the mirror spacing may have to allow for the use of higher magnetic fields before rogue modes appear and also increasing REF frequency slightly (at the same magnetic field), both possibly necessary when HP came out with the 5517C and 5517D in the early 1990s.

    Thus, there would actually appear to be three types of tubes: Short, Long (High Voltage, HV), and Long (Low Voltage, LV), with the latter only appearing in very old lasers. The tube voltage of both Long-LV and Long-HV tubes increases by around 100 V over their lifetime. I do not have enough data for the Short tubes but would expect a similar trend.

    It turns out I had a box of 5501B and 5517A laser tube guts stashed away. Upon rechecking the dimensions of the mirror spacing rods, they were all from Long-LV tubes! I had never even measured them, assuming they would be identical to the original 5517B tube I dissected several years ago. That one has been loaned out so I was unable to compare it to the others. With this mystery, the urge to compare the two types became irresistible! I found a very weak hard start nearly end-of-life 5517C from 2006 and issued the appropriate chants and incantations to the gods of dead lasers before sacrificing it in the interest of research. Indeed, the lengths of the mirror spacing rods differ with the Long-LV being the predicted 132.5 mm. However, its mirror spacing rod has a length of only 126 mm, not the 127 mm I'm quite sure I had measured on the HP 5517B. Solve one mystery and create another! But that's a minor anomaly and I'll leave the value at 127 mm for the purposes of calculation. The discharge escape holes are in very nearly the same locations relative to the ends of both rods but the stepped bore of the newer Long-HV tube is substantially narrower, especially at the output end. So that would explain the higher voltage even with the shorter discharge length. And even though it is shorter, the narrower bore probably results in higher overall gain and thus output power. At the same time, I was able to measure the mirror reflectivity of the 5517C OC and found its value to be in agreement with predictions. More on this later.

    Here are the HP/Agilent/Keysight HeNe laser tube parameters as best I have been able to determine them so far, compared with a typical short barcode scanner HeNe laser tube. Although only the 5500A, 5501B, 5517C, and 5517E are shown, the tube voltage, total length, planar HR mirror, divergence (without beam expander), and beam diameter are similar or identical for other models using the same type tube. There is some uncertainty in the mirror spacing for other Long-HV tubes as a late model (2006) 5517C had a value of 126 mm instead of 127 mm. But while definitely a design change, this is comparatively minor. The parameters for the barcode scanner tube are for a range of typical models. The example is a Melles Griot 05-LHR-006 which is similar to barcode scanner tubes from other manufacturers including Siemens (now LASOS) and JDS Uniphase (now Lumentum). The divergence for a particular model barcode scanner tube is usually achieved by either the specific curvature of the outer surface of the OC mirror glass or with an external lens glued to it, but the cavity design including the OC RoC (radius of curvature) is usually similar. These are all random polarized tubes (and must be for the Zeeman splitting to work properly). Click on the (Example) model number for a diagram of the internal structure of each tube type.

        First Year -> 1974      1976      1983      1992      2003       1980s
           Example -> 5500A     5501A     5501B     5517C     5517E    05-LHR-006
       Designation -> PZT-1     PZT-2    Long-LV   Long-HV    Short     Barcode
     Output Power    0.3-1 mW  0.3-1 mW  0.3-1 mW  0.3-1 mW  0.1-1 mW  0.4-1.5 mW
     REF Frequency   1.5-2 MHz 1.5-2 MHz 1.5-2 MHz 1.5-4 MHz 2-7.2 MHz 0.8-1.7 MHz
     Total Length     170 mm?   170 mm    194 mm    194 mm    160 mm   110-155 mm
     Cavity Length    123 mm    130 mm   132.5 mm   127 mm   101.6 mm  100-150 mm
     Cavity FSR      1.22 GHz  1.153 GHz 1.13 GHz  1.18 GHz  1.48 GHz  1.0-1.5 GHz
     Cavity Geom.      N HS      N HS      N HS      N HS     LR HS      LR HS
     HR Mirror RoC    Planar    Planar    Planar    Planar    Planar     Planar
     HR Mirror Diam. 25.3 mm?  25.3 mm    9.32 mm  9.32 mm     5 mm     6-7.75 mm
     OC Mirror RoC     ????     132 mm    136 mm    136 mm   ~136 mm   200-300 mm
     OC Reflectance   98.74%?   98.74%     98.5%     98.0%    97.8%    99.0%-99.5%
     OC Mirror Diam. 25.1 mm?  25.1 mm    9.35 mm  9.35 mm    10 mm     6-7.75 mm
     Mirror Align.      NA        NA        NA        NA        HR      HR and OC
     Bore Diameter   1/1.6 mm? 1/1.6 mm  1/1.5 mm  0.8/1 mm  ~0.5 mm   0.4-0.6 mm
     Beam Diameter     ~1 mm    ~1 mm     ~1 mm     ~1 mm    ~0.5 mm   0.4-0.6 mm
     Divergence       ~10 mR?   ~10 mR    ~10 mR    ~10 mR   ~2.0 mR  1.7,2.7,8 mR
     Disch. Length     96 mm?   100 mm    105 mm    100 mm     82 mm    50-75 mm
     Oper. Current     3-5 mA   3-5 mA    3.5 mA    3.5 mA    3.5 mA     3-4 mA
     Oper. Voltage    1.3 kV?   1.4 kV   1.05 kV    1.2 kV   1.15 kV   0.7-1.1 kV
     Anode Ballast     136K?     136K      100K      100K      100K     75K-100K
     Htr Res. Cold      NA        NA      8 ohms    8 ohms    9 ohms       NA
     Htr Res. H:C       NA        NA      1.285     1.285      1.4         NA


    1. Values for the 5517C (except OC mirror reflectance) also apply to the 5517A/B/D from the mid-90s (after the Long-HV replaced the Long-LV tube) through the early 2000s. After the early 2000s, 5517A/B/C/Ds and N1211As with Long-HV tubes have a cavity length of only 126 mm for an FSR of 1.19 GHz. They also apply to the 5518A and 5519A/B over similar periods. As of around 2012, it appears as all new Agilent/Keysight lasers use versions of the Short tube except possibly the N1211A.

    2. I have not dissected a 5500A (which is identical to the 5500B). Thus, parameters with a "?" at the end of the value are educated guesses based on those of the 5501A, which is virtually identical. The 5501A uses the same tube as the 5500C.

    3. I have only dissected a few Short tubes, probably all from 5517ELs. The reflectance (R) was measured as 97.9%. As of sometime between 2012 and 2018, virtually all new tubes use the Short design. The basic optical design of the Short tube is more cloesly related to that of barcode scanner tubes than to previous HP/Agilent tubes with their wide tapered or stepped bores and large divergence.

    4. There may be at least two versions of Short tubes that (aside from mirror reflectivity/split frequency) have slightly different beam diameters (and divergence). Since it is believed they all have the same bore ("Mirror Spacing Rod"), this must be done via the RoC and/or curvature of the outer surface of the OC mirror. The evidence for this is that identical 6 mm beam expanders from/for Short tubes behave differently with the expanded beam not filling the output aperture from some Short tubes.

    5. "Cavity FSR" (longitudinal mode spacing) was estimated on the 5500A using a Scanning Fabry-Perot Interferometer (SFPI), and both measured on an SFPI and confirmed using physical measurements of cavity length for the others.

    6. "Mirror Align." denotes the mirror(s) whose alignment may be fine tuned on the intact laser tube.

    7. "Cavity Geometry" is either "Near Hemispherical" (N HS) or "Long Radius Hemispherical" (LR HS).

    8. "Bore Diameter" shows the two estimated values in tubes that use a stepped-bore. The Short tube appears to have a constant diameter bore. For the barcode scanner tube, it is the range of typical (constant diameter) bore diameters.

    9. The values for "Mirror Diameter" for HP/Agilent tubes are based on measurements of a single sample and it is known that there may be considerable variation.

    10. The "OC reflectance" applies to the example model where relevant. It is relatively low for HP/Agilent tubes which have a maximum output power below 1 mW, expecially the 97.8% of the 5517E. The OC mirror reflectance is one of the primary parameters that determines the REF frequency. All other factors being equal, reducing the reflectance (or equivalently increasing the mirror transmission) increases cavity loss which increases the REF frequency. But it also may decrease output power, possibly to 0 if too low. Tubes of the same cavity length with higher REF frequencies (like the 7.2 MHz of the 5517GL) may have even lower OC reflectivities, though magnet strength can also be used (up to a point) to adjust REF frequency. The shorter cavity of the Short tube results in a higher REF frequency (at the same magnetic field) and also allows a higher magnetic field to be used (further boosting REF) without creating rogue modes.

    11. The "First Year" dates of introduction are estimates. For the barcode scanner, it is intended to be when the more compact modern hard-sealed HeNe laser tubes were first used in commercial scanners.

    Differences between Long and Short HP/Agilent Tubes

    Up until well after HP became Agilent, there was only a one-size-fits-all glass tube design for the thermally-tuned lasers. The differences to accomodate the various laser models (5501B and 5517A/B/C/D) were achieved through selection of OC mirror reflectivity and magnet strength. (Though there is a possibility the reduction in cavity length from 132.5 to 127 mm going from Long-LV to Long-HV tubes was related to the introduction of 5517C and 5517D lasers. The cavity spacing did decrease to 126 mm (and other even shorter ones), to eliminate any chance of rogue modes in higher-REF lasers like 5517Ds and even early 5517Es.) Then, probably in order to be able to implement the 5517E/EL with decent output power (and then the 5517FL and 5517GL), a shorter tube appeared. Functionally, the primary difference is in the cavity length: The long tube has a mirror spacing most commonly between 126 and 132.5 mm depending on version, while the short tube has a mirror spacing of around 100 mm. For the same mirror reflectance, the shorter cavity results in a higher REF/split frequency (by roughly 5/4, all other factors being equal), It also allows for a higher magnetic field before rogue modes appear, in approximately the same ratio. Taken together, the maximum REF frequency nearly doubles. With some tweaks to the internal construction, the overall gain isn't reduced proportionally, or perhaps not at all. The most significant is the use of a long radius hemispherical cavity with narrow bore rather than the near-hemispherical cavity with wider stepped bore of all previous HP/Agilent lasers. A narrower bore has higher gain and the larger mode volume fill makes more efficient use of the discharge. This more than counters the slight reduction in length of the discharge necessitated by the shorter cavity. Thus, the Short tubes are still capable of acceptable output power. Since the overall size of the glass envelope is only slightly smaller than that of the Long tubes, the gas reservoir is similar and life expectancy should be as well. Short tubes have now (as of 2019) been found in virtually all Agilent 5517 lasers including the 5517A, which has the lowest REF frequency of any common Agilent laser in current production.

    Some other changes appear to have been made in the short tube including setting the divergence closer to the diffraction-limited value rather than the much wider divergence of the previous tubes (though this may simply have been a side-benefit of the shorter cavity as the same OC mirror RoC is used). The smaller divergence (or mostly the smaller beam diameter that comes with it) means that a new beam expander is required in the short tube lasers to achieve the same beam diameter.

    Gone are the funky springs and backing disks, so the bore is rigidly mounted and fused to the HR mirror mount assembly, with metal "cage" stuck on the other end to hold the OC mirror in place against the mirror spacing rod.

    The tube voltage of Short tubes when new is about 50 V lower than for the Long-HV tubes, but 50 V higher than the Long-LV tubes. With the same 100K ohm ballast, they should run happily on the same power supplies. And it is believed that all HP/Agilent HeNe laser power supplies are compatible with all tube designs. Late model lasers do have either a VMI PS 504 (which looks identical to the VMI PS 373) or a VMI PS 253, which is in a shielded case, but these don't appear to be specifically tube-related.

    The cold heater resistance of the short tubes in production may be a bit higher than that of the long tubes - 9 ohms versus 8 ohms, but that may just be normal process variation. (The 5517E I have with no serial number had a lower heater resistance - around 4 ohms - and an external resistor to make up the difference, but this may have been a prototype or early production version.)

    The outer glass envelope is shorter for the Short tube :), but not nearly as short as it could be. So, the OC is recessed inside by roughly 2 inches. This was probably done to maintain an adequate gas reservoir volume and similar life expectancy compared to the Long tubes. However, it also provides an opportunity to lengthen the cavity if needed to achieve higher output power for low REF lasers like the N1211A without any major changes.

    It is likely that all of the common laser models in production now (2014 and beyond) use the Short tube, but that there are probably at least two versions. These would be physically identical but differ in OC mirror reflectivity to cover the range of REF frequencies from approximately 1.5 MHz to over 7 MHz. The required REF/split frequencies could then be achieved solely by selecting a Short tube with the required OC mirror and the appropriate magnet field strength. This would certainly simplify inventory control. (It is not known whether the N1211A has converted to a Short tube. Since it is designed for high power, the Long tube may still provide advantages.)

    Based on tests of the single healthy bare Short tube I current have available, the FSR of the ~100 mm cavity (around 1.5 GHz) is so large that when run without a magnet, the output may actually go to 0.00 mW during a small part of mode sweep. This would be when the (at most) 2 longitudinal modes are approximately equidistant on either side of the 1.5 to 1.6 GHz neon gain curve. This behavior is not seen on any other modern commercial HeNe laser (except the Zygo 7705) since none have nearly as short a cavity. The cute little SP-007/Melles Griot 05-LHR-007 with a cavity length of 110 mm is the shortest I'm aware of. When installed in its magnet, the gain curve splits and the power variation becomes more in line with that of other HP/Agilent tubes.

    Possible Evolution of the Agilent Short Tube

    The following is speculation based on physical evidence, but that can be compelling.

    Around when Agilent was spun off from Hewlett Packard in 1999, there must have been an effort to extend the 5517 laser to higher REF frequencies, in part to compete with Zygo's 20 MHz REF. Zeeman technology had no chance of matching that, but there were still a few tricks available to extend REF to around 8 MHz. And that would be sufficient for most high-end applications. (The much more complex Agilent N1211A Fiber AOM Laser could provide almost arbitrary REF frequencies to handle anything else.) For the standard Long-HV tube, the useful upper limit for REF is around 4.0 MHz. 5.0 MHz or a bit more could be squeezed out of a Long tube but at an output power of less than 100 µW! More below. A greater OC mirror transmission (%T) as well as a stronger magnetic field will boost REF. And a shorter cavity will also boost REF with the other parameters unchanged, but at the expense of gain due to a shorter discharge length. However, if the OC %T is too large or the gain is too low, there will be no lasing at all. And too strong a magnetic field will result in rogue modes.

    Around 2002, based on examination of a prototype for the 5517E (minimum REF of 5.8 MHz), Long-HV tubes were constructed with a mirror spacing of 100 mm instead of 126 or 127 mm. That may have been the only change to the lasing parameters, though the internal heater resistance dropped from around 8.0 ohms to 6.1 ohms, no doubt due to reduced space available for it. Whether anything else was changed is not yet known. But for a given OC %T, the length change alone would boost REF by 25 percent and permit a magnetic field 25 percent higher before rogue modes set in, for an overall benefit of over 50%. Thus going from 4.0 MHz to around 6.0 MHz would be possible IFF the thing lased at all with the shorter bore discharge length available. ;-) A diagram based on a test for mode spacing with a Scanning Fabry Perot Interferometer (SFPI) and external visual examination is shown in Internal Structure of Agilent Prototype 5517E Laser Tube Assembly. Compare this to Internal Structure of Hewlett Packard 5517B/C/D Laser Tube Assemblies. The principal changes are that the mirror spacing rod is shorter with more backing disks. But I've also taken the liberty of moving the cathode discharge escape holes a bit closer to the end of the rod to maximize bore discharge length. So some minor details may not be entirely accurate, but I doubt anyone on the Planet could identify them. ;-)

    The prototype I have does work - barely - with an output power of around 50 µW when locked. It is not known whether this tube has been run for a life test and thus declined in power, or is simply a failed experiment. It starts reasonably quickly and runs stably at the default current of 3.5 mA. The label simply says: 5517E-C01, prototype_000006, nothing about a life test. The laser it was in also has the Type III Control PCB, which will lock at very low output power. The Agilent Type II Control PCB was not in production in 2002 and the HP Type I Control PCB might not lock at such low power. Whether that means anything is also not known. Based on other information, the tube part number 5080-0183 was actually used in some production 5517E-C01s with a spec'd minimum output power of 65 µW. There may also have been some high-REF 5517Ds around the same time, as well as high power 5517As and N1211A, using Long-HV tubes with modified mirror spacing rods, but possibly the changes were other than cavity length. An SFPI test of a 5517D-N07 (4.25 to 4.60 MHz REF) and N1211A (700 to 800 µW) showed the same 1.18-1.19 GHz FSR as other 5517s, so perhaps the cathode-end discharge escape hole was moved to extend the bore discharge length slightly to increase power. While that would result in a small increase in tube voltage, it would probably not be detectable given the normal variation of tube voltage with use. Other changes like bore diameter can't be totally ruled out but are probably much less likely. However, a careful comparison of mirror spacing rods removed from a 5517C and N1211A failed to detect *any* physical differences including the location of the discharge escape holes and bore diameter. So perhaps there really is no difference.

    Given that these 5517Es were prototypes with serial numbers in the single digits, only minor structural changes were probably made to the tube, limited to the length of the mirror spacing rod at the OC-end, and possibly the location of the discharge escape holes in an attempt to maximize bore discharge length. A visual inspection of the HR-end indicates that it is similar to that of a normal Long-HV tube. Only destructive disassembly or an X-ray can determine for sure what else has changed.

    At that point, it's quite possible the engineers who had been frustrated for many years where only minimal changes were allowed to the 5501B, and then the 5517s, started salivating at the possibility of a more radical approach. So they finally convinced Management that minor changes to the existing tube design would not result in a robust solution for the 5517E and other lasers with even higher REF, demanded by higher performance applications. What was needed was an optimal design to boost gain - and thus output power - significantly while maintaining the same ~100 mm spacing between mirrors to achieve the higher REF. It would thus be beneficial in more than one way. By using a narrower bore as in modern conventional HeNe laser tubes and putting the HV connection behind the HR mirror to increase the bore discharge length, the gain could be increased significantly without sacrificing the dual benefits of the 100 mm cavity length. It would be difficult to match the performance of the Short tube with any reasonable modifications to the design of a conventional internal mirror HeNe laser tube.

    The tube design is also simpler in several ways and is likely less expensive to manufacture. Specifically, the front and back sections of the glass envelope no longer need to be precisely aligned and fused while at the same time hoping that everything doesn't fly apart due to the springs at both ends. Everything is attached to the rear section. And the mirror spacing rod itself no longer determines mirror alignment as the HR is on a metal support and the OC is on a deformable metal cage affair, both of which can be tweaked during passive alignment (probably using some sort of interferometer). And in the case of the HR, tweaked even after the tube is sealed. Therefore, the rod doesn't need to be ground to the same precision. On the downside, with mirror alignment not being determined by precision ground surfaces, it can change over time and thermal cycles as may occur with HeNe laser tubes of conventional design, though there is little evidence to suggest that this is a significant issue.

    Only a close examination of a dissected Short tube would also reveal that there is a metal plate with a small hole attached directly to the OC-end of the mirror spacing rod between it and the OC mirror. The hole is slightly smaller than the bore diameter so probably serves as a controlled limiting aperture for the lasing mode. Agilent "Short" HeNe Laser Tube Cavity Parameters.

    And thus the Short tube was born. Versions are now used in all of the Keysight 5517, 5519, N1211A/Z4203 lasers. These may differ only in the OC mirror %T. But the mirror spacing rod length could also be easily modified as needed. For example, it could be made longer to boost power for low REF lasers like the N1211A as there is plenty of extra space available, though the magnet would also need to be longer if the discharge region approches or exceeds the 4 inch length of the current magnet because the field reverses polarity beyond the ends. A 5 inch magnet could easily be constructed from a standard 4 inch magnet and one segment of the original HP segmented magnet. ;-) And a "spider" would be desirable to support the end of the longer rod, but of course it should not interfere with the normal thermal expansion required for stabilization. The length of the heated region could also be extended. See: Internal Structure of Possible High Power "Short" Tube. The glass envelope is identical to the normal Short tube but the discharge length is around around 25 percent greater and the cavity length is about 20 percent greater, though not quite as long as that of the "Long" tubes, but the gain is higher. To preserve benefits of the standard Short tube including efficiency and low divergence, the bore diameter and aperture may need to be adjusted and the OC mirror RoC would need to be larger than the 13.6 cm that's been used for the past 35-40 years. Sorry. :-) The high power Short tube would fit the same size frame as the normal one but would require the frame and front optics section to be redesigned to accomodate the extra 1 inch of magnet. Having said all that, there is really little need for such a tube. Lively samples of the normal Short tube do well over 1 mW locked in a laser. :) It just seemed a shame to waste the effort put into drawing the silly diagram. :-) Going the other way, it might be possible to shorten the rod to eke out a bit higher split frequency. However, so far there only appears to be a single rod length.

    See Types of HP/Agilent Thermally Tuned HeNe Laser Tubes. The Long-LV and Long-HV tubes are virtually identical but span some 25 years, while the Short tube at the bottom differs significantly.

    However, the useful life of Short tube appears to be limited by power decline rather than gas depletion issues as with Long tubes. This conclusion is based on the following:

    My hypothesis is that this is caused by sputtering of the discahrge at the anode electrode in close proxiity to the HR mirror. The result is the deposit of a metallic film over time on the mirror decreasing reflectivity and increasing losses. The result is a decline in the output power and increase in the REF frequency. Suggested Improvements to Keysight "Short" HeNe Laser Tube" shows what might be a simple solution, or at least a test to confirm the cause. The top diagram is of what is believed to be the present Short tube design in 2024; the lower diagram shows the suggested modifications to the anode-end mirror assembly at the left. Basically, a glass sleeve is added to act as an baffle to redirect the discharge so it avoids terminating where there is a direct line-of-slight path to the mirror. This would cost next to nothing to implement and could potentially add years of life to these tubes. Oh, but perhaps that isn't what is desired. ;-)

    HP/Agilent 5517 Laser Control PCBs and Locking Sequence

    The stabilization technique used with 5517 lasers is really dirt simple: Start by heating the "mirror spacing rod" to a set-point temperature, fine tune the temperature, then switch to optical feedback which forces the two orthogonal Zeeman-split polarized modes to be equal. However, the details vary slightly depending on the specific Control PCB used in the laser.

    The 5517 laser Control PCBs are known as the "A3 Control/Reference Board" in HP/Agilent manuals. The locking sequence for each of the three distinct types are described below.

    Locking sequence with Type I Control PCB:

    The Type I Control PCB, know as the "A3 Control/Reference Board" in HP/Agilent manuals has part number 05517-68003 This also applies to the 5518A and 5519A/B since they all use the 5517A Type I Control PCB. It is also generally applicable to the 5501B though some details differ slightly.

    From power-on to READY takes around 4 minutes for most 5517 lasers - all those NOT using the Type III Control PCB. Even on the original Type I Control PCB, a state machine based on counters, flip-flops, and gates determines the timing. This may be true of the Type II Control PCB as well, except that the state machine would be inside a Xilinx FPGA. Who knows how the Type III Control PCB with its SHARC CPU implements this algorithm (which tends to take much longer than 4 minutes, reason unknown)! The following is paraphrased from the 5517A manual, which assumes the Type I Control PCB implementation. (All timing is approximate as the main clock is a 555 timer on the Type I Control PCB!):

    Thus, under normal conditions, the laser will be locked and ready to make a measurement (approximately) 150 seconds after the READY LED starts flashing. Note that the only check to make sure the laser is locked is that the REF signal is present. Since this only occurs for a small percentage of the entire longitudinal mode sweep cycle, REF will not remain on for long without active feedback, so this is a reliable test. The laser will in fact continue to repeat the above sequence forever if REFON is not detected. Typically, this will occur when the output power from the laser tube has declined to below the REF detection threshold of the internal optical receiver after years of hard work. However, some marginal lasers will go through the sequence several times when powered up as the output power from the laser tube gradually increases with warmup until the amplitude of the difference frequency signal exceeds the REF detection threshold.

    Locking sequence with Type II Control PCB:

    The Type II Control PCB has 3 yellow state LEDs near the top right corner of the large Xilinx chip. These provide some information about where the controller is in the warmup process and they have a 1:1 correspondance with the major modes of the Type I Control PCB. I'm not sure the times in each state are identical for the two but they are close. Here is a rough chart of their behavior for a normal 5517C laser:

       Time    READY   State    Comments
       0:00             000     Power on (WARMUP Mode)
       0:01             001
       0:02             000
       0:07             00X     001-000-001-000 in three seconds.
       0:10             000     Remain here 3 seconds.
                The previous two entries repeat approximately 14 times,
                 dependant on time to reach set-point temperature.
       1:26   Blinking  010     (HEATER QUALIFIED Mode)
       1:32   Blinking  01X     011-010-011-010 in three seconds.
       1:35   Blinking  010     Remain here 3 seconds,
                The previous two entries repeat approximately 16 times.
       3:03   Blinking  110     (OPTICAL Mode)
       3:09   Blinking  11X     111-110-111-110 in three seconds.
       3:12   Blinking  110     Remain here 3 seconds.
                The previous two entries repeat approximately 9 times.
       3:58      ON     000     (LASER READY)

    The blink rate for READY is about 1.5 Hz.

    The REF signal for the Type II Control PCB is trasformer-coupled rather than capacitor-coupled as in the Type I Control PCB for small-case lasers. (Both types have been found in 5517/18/19 lasers.) This results in a smaller open-circuit p-p signal but one that is still sufficient for the 5508A.

    Locking sequence with Type III Control PCB:

    The Type III Control PCB seems to go through many more gyrations during warmup than either of the others, including several times where READY flashes multiple times separated by a period of inactivity, and then finally flashing READY continuously for two minutes until it locks - the latter being similar to what the Type I Control PCB does. The entire process consistently takes much longer than the 4 minutes typical of a laser with the Type I Control PCB, up to 10 minutes or more. The behavior is not obviously different whether a weak (but functional) laser tube, or one that greatly exceeds minimun output power specs is used, though it may take slightly longer with a below-spec tube. After all this, the end result seems to be exactly the same.

    Here is a chart of the typical startup behavior for a very healthy 5517B tube installed in a (previously) 5517D laser with this Type III Control PCB:

       Time    READY   State    Comments
       0:00             1111    Power on
       0:01             0000
       0:02.0           0001    0001-0010-0100 sequence in less than 0.5 sec.
       0:02.1           0010
       0:02.2           0100
       0:03    Flash    1100    MSB LED and READY LED flash briefly.
       0:04.0           0001    0001-0010-0100 sequence in less than 0.5 s.
       0:04.1           0010
       0:04.2           0100
       0:05    Flash    1100    MSB LED and READY LED flash briefly.
       0:06.0           0001    0001-0010-0100 sequence in less than 0.5 s.
       0:06.1           0010
       0:06.2           0100
       0:07    Flash    1100    MSB LED and READY LED flash briefly.
       0:08             0100    3x(Step 0, Step 3, Step 5) sent on RS323 port.
       0:15             0101    "LASER" sent out RS232 port.
       1:00             0110
       1:35   Flash 5   X110    1110,0000,5x(1110,0110).
       1:40             0110
       2:14             0000
       2:15   Flash 5   X110    1110,0000,5x(1110,0110).
       2:20             0110
       2:55   Flash 6   X110    1110,0000,6x(1110,0110).
       3:01             0110
       3:35   Flash 7   X110    1110,0000,7x(1110,0110).
       3:42             0110
       4:12   Flash 8   X110    1110,0000,8x(1110,0110).
       4:20             0110
       5:02   Flash 8   X110    1110,0000,8x(1110,0110).
       5:12             0110
       5:40   Flash 10  X110    1110,0000,10x(1110,0110).
       5:50             0110
       6:45   Flash 32  X110    1110,0000,32x(1110,0110).
       7:17   Flash 96  X111    96x(1111,0111).
       8:53      ON     X000    1000,0000,1000,0000,...

    The State refers to the 4 SMT LEDs above the upper left corner of the Lattice chip near the center of the PCB. The MSB is green while the three LSBs are red. All times are approximate. "Flash" is just the briefest pulse of light. "Flash n" denotes "n" flashes at a 1 Hz rate with a 50 percent duty cycle The duration for the 1110,0000 state changes in each "Flash n" sequence is relatively short (perhaps 100 ms for each of the two states). The minimum value for "n" seems to be 5, but it tends to increase as the laser warms up. (I'm not positive it's monotonically increasing though.) Once the REF frequency can be sustained by the feedback loop, it continues flashing for 32 seconds, and then switches to state 0111 for 96 seconds prior to becoming READY. Until that time, the READY LED and the MSB state bit track each other almost perfectly. But then, the MSB state bit (1000) continues to flash (but now at about a 1.2 Hz rate) while the READY LED remains on solid, And there is just a hint of the 0100 state bit flickering dimly, possibly the actual feedback loop in operation. :)

    Multiple runs from a cold start may differ slightly in the number of "Flash n" sequences and the values of n, as well as other details, but always take much more than 4 minutes (typical of the Type I Control PCB). The very healthy tube will lock in 7 to 9 minutes while a weak but usable one might take 11 minutes or more. In all cases where the laser successfully locks, the last two minutes will be identical in behavior to that of the other two Control PCBs with READY flashing continuously until it stays on solid. A tube that is very weak or with no detectable beat (REF) frequency will result in only occasional very short abortive flashes, and no conclusion (at least not in 15 or 20 minutes).

    One annoying difference between this Control PCB and the others is that the signal level for REF and ~REF seems to be much lower - about 2 V p-p open circuit and less than 1 V (maybe as low as 0.5 V p-p) terminated, instead of more than 5 V p-p, and the 5508A display apparently doesn't accept this as a valid signal. So even if the laser comes READY within 10 minutes (the maximum allowed by the 5508A), it still comes up as "Laser Fail", which isn't recoverable without power cycling the 5508A (which means the laser as well if it receives DC power from the 5508A). However, my home-built SG-MD1 display has no problemss. :) I wouldn't be at all surprised to learn that the signal levels are programmable - somehow.

    Here is a chart of the typical startup behavior for the 5517E with its similar Type III Control PCB. The tube is probably below the Agilent spec for minimum power, but locks without problems so the sequence of event is probably not affected singificantly:

       Time    READY   State    Comments
       0:00             1111    Power on.
       0:01             0000
       0:02.0           0001    0001-0010-0100 sequence in less than 0.5 s.
       0:02.1           0010
       0:02.2           0100
       0:03    Flash    1100    MSB LED and READY LED flash briefly.
       (Repeat the previous 4 events 27 times.)
       1:15.0           0001    0001-0010-0100 sequence in less than 0.5 s.
       1:15.1           0010
       1:15.2           0100
       1:16    Flash    1100    MSB LED and READY LED flash briefly.
       1:17             0100    3x(Step 0, Step 3, Step 5) sent on RS323 port.   
       1:27             0101    LASER sent on RS232 port.
       1:52             0110
       2:25             0000
       2:26   Mode 16   XXX0    16x(1110,0000,....,0000).
       3:35   Mode 12   XXX0    12x(1110,0000,1110,0000,....,0000).
       5:10   Flash 15  X110    1110,0000,15x(1110,0110).
       5:40             0110
       5:48   Flash 32  X110    1110,0000,32x(1110,0110).
       6:10   Flash 96  X111    96x(1111,0111).
       7:46      ON     X000    1000,0000,1000,0000,...

    The last part of the sequence is essentially identical to that of the other laser, but the initial behavior differs significantly. This one appears to keep track of the mode cycles, or at least flash the State LEDs in response to them! "Mode n" denotes "n" times where the Zeeman beat is on, at least momentarily. Also note that the READY LED only tracks the MSB State bit near the end. I assume that the AM29F040B (4 Mbit flash memory) is the firmware NVRAM but there is no version number so I don't know that they differ, Bbut they must as everything else on the two Type III Control PCBs appears identical including the DIP-switch settings.

    For more on the 5517 laser Control PCBs, see the section: HP/Agilent 5517 Laser Construction.

    HP/Agilent 5517/8/9 and 5501B Temperature Set-Point Adjustment

    Setting the lock temperature set-point is the only user adjustment on HP/Agilent lasers.

    The following applies to all of the small HP/Agilent 5517 lasers using the Type I or Type II Control PCB but NOT to those having the Type III Control PCB (which may not require a temperature set-point adjustment at all). It also applies to the 5501B and to the larger 5517A, 5518A, and 5519A/B lasers which have a modified version of the Type I Control PCB. (The locations of jumpers and test-points may differ but they have the same labels. Any discrepencies are noted.) This is a very non-critical setting and the laser will operate normally over a rather wide range. But it is worth doing when testing any laser and necessary if installing a replacement tube. While the laser may appear to work fine without performing this adjustment, doing so will assure that lock will be maintained over the spec'd temperature range for the laser. This is particularly important where the tube type has changed (Long to Short or vice-versa).

    A DMM (preferably with a clip lead on the negative probe) set to measure 200 to 400 mV, a medium flat blade screwdriver to remove the cover on the small (rectangular) lasers or Philips or Torx for the large lasers, and small flat blade screwdriver to turn the trim-pot will be required. Note: This default procedure is what would be found in the 5517 operation and service manual. It's possible that to "fudge" the specifications, slight modifications may have been made during factory testing. I've seen several lasers where the set-point is significantly higher than would be accounted for by this procedure. I doubt any of these lasers had been tampered with, or that the setting drifted that far on its own. A higher temperature set-point may have been used to slightly increase output power, slightly reduce the REF/split frequency, or to accomodate specific environmental conditions.

    The laser should be unpowered for at least 2 hours prior to performing the temperature set-point adjustment:

    1. On the small (rectangular) lasers, remove the cover by rotating the shutter wheel (if present) so the large hole is at the bottom. Use a medium flat blade screwdriver to push in and rotate the 1/4 turn fastener 90 degrees counter-clockwise. (This may require considerable force on newer lasers with the sheet metal shroud, which has a compliant conductive gasket all around.) This will release the front plate. Then the side panels can be swung out and removed, or the shroud can be slid off, taking care not to catch components on the Control PCB. On the large lasers, remove the 4 screws on top and use a thin blade to lift the lid at each end until it pops off.

    2. With power still OFF, move the "HTR" jumper (J7) on the Control PCB from "NORM" (left and center pins) to "OFF" (center and right pins). This is the far right jumper on the Type I Control PCB (including lasers in the large case) and second from the far right on Type II Control PCBs for lasers in the small case. I do not know where it is located on the Type II Control PCB for lasers in the large case but it should be labeled.

    3. Unplug the HeNe laser power supply cable (3 pin header with red and black wires) from the Connector PCB. (DO NOT remove the 2 pin header with red and purple wires.) This avoids heating of the mirror spacing rod from the HeNe discharge. With nothing at all heating the mirror spacing rod, there's much more time to make the measurements and adjustment though doing it as quickly as possible is still recommended. (Older operation and service manuals omit this step.)

    4. Clip the DMM's negative probe to one of the GND test points on the Control PCB, or to the chassis.

    5. Apply power to the laser.

    6. Confirm that the ±15 VDC power supplies are within spec: 15V±0.3V. There are labeled test points on the Control PCB. Adjust if necessary.

    7. Measure and record the voltage at TP11 on the Control PCB.

      • On the Type I Control PCB in the smaller lasers (including the version for the 5501B), TP11 is near the bottom of the board to the left of the pair of white film capacitors in the middle of the board.

      • On the Type I Control PCB in the 5517A, 5518A, and 5519A/B, TP11 is at the top near the center of the board.

      • On the Type II Control PCB in the smaller lasers, TP11 is about 3/4" above the status LEDs, above the FPGA.

      • On the Type II Control PCB in the 5517A and 5519A/B, I don't know where TP11 is. ;-) Hopefully, it's labeled.

      The multiplicative factor used in the next step depends on the tube type:

      • For lasers with "Long" tubes, multiply the voltage on TP11 by 1.285 and add 0.005 V and record that value. (The 0.005 V was found in an on-line Keysight document and is not present in any manual I've seen.) The Long tubes have rubbery potting compound visible at the back of the laser tube assembly, and are in virtually all the lasers you're likely to encounter. :)

      • For lasers with "Short" tubes, multiply the voltage on TP11 by 1.4 and record that value. (I assume 0.005 V should be added to this value also but have not seen it documented.) The Short tube may be identified by the plastic cover secured with three screws to the back of the laser tube assembly. Until 2010 or so, Short tubes were only found in the very rare 5517E/F/G lasers and some (probably high-REF) versions of 5517Ds. Starting around 2012, most or all 5517 lasers are made with Short tubes.

      That correction of 0.005 V won't make a huge difference either way, but adding it won't hurt. The tube will run a bit warmer with a slightly lower REF frequency and possibly slightly higher output power.

      The type of tube may be determined by inspecting the back of the tube assembly. The Long tube has a glass stem sticking out of the rubber potting. The Short tube has a translucent or black plastic cover secured with three screws.

      The factor (1.285 or 1.4) is independent of the controller. So, if installing a different type tube, the controller should be adjusted using the appropriate value for the tube type - Long (rubbery back) or Short (plastic back cover).

      Note that there may be differences between Type II Control PCBs used with high-REF lasers, so it's best to transfer the PCB with the tube.

      An Agilent operation and service manual for the 5517B/BL/C/D/DL/FL replaced the simple factors with elaborate tables for each tube type and controller (even though the controller makes no difference). I suppose that the authors/editors figured that gadget-addicted human beings are no longer able to do simple multiplication! But, the first two columns in each table had the factors (1.285 and 1.4) swapped! While using the smaller value with the Short tube would not damage anything, the laser may lose lock, or the temperature range over which it remains locked would be reduced. However, using the larger value with the Long tube would result in the tube and laser hotter which could be detrimental. There is also a troubleshooting flowchart for the lasers with the Type II Control PCB but some of the entries are plain screwy and less than useful. :) I have not been able to find a corrected version of this manual. It is Agilent PN 05517-90077, © 2009.

    8. Measure the voltage on TP15.

      • On the Type I Control PCB (including the version for the 5501B), TP15 is near the top of the board to the left of the long blue trim-pot, R16.

      • On the Type I Control PCB in the 5517A, 5518A, and 5519A/B, TP15 is at the top of the board above the large white film capacitor.

      For the Type I Control PCBs, the voltage on TP15 is reasonably accurate regardless of the position of the HTR jumper (J7).

      • On the Type II Control PCB for lasers in the small case, TP15 is to the upper right of the square tan trim-pot, R14.

      • On the Type II Control PCB for lasers in the large case, TP15 should be near the square tan trim-pot, R14.

      For the Type II Control PCB, the voltage on TP15 may not be at all valid unless the HTR jumper (J7) is in the OFF position.

    9. Adjust the trim-pot (R16 or unlabeled for the Type I PCB; R14 for the Type II Control PCB) so that the voltage on TP15 is within ±1 mV of the value calculated in step 6, above.

    10. Power down the laser, wait 15 seconds for power supply voltages to decay, move the HTR jumper back to the NORM position, and plug the HeNe laser power supply back in.

    11. Power up the laser and wait for it to lock (READY on solid).

    12. Measure the voltage on TP11. The value will depend on the tube type:

      • Long tube: 5 to 7.5 V.
      • Short tube: 6.5 to 9 V.

      If the voltage on TP11 is NOT within the appropriate range, adjust the set-point up or down slightly to bring it within range, preferably closer to the lower end. Clockwise will increase the operating voltage (and heater temperature) and vice-versa. One half turn in the apprpriate direction is probably a good start. There is no need to power down the laser or change the HTR jumper to do this, but the laser must be forced to go through the locking process by blocking the beam between the laser tube and beam sampler until the READY LED goes off. Repeat as necessary.

      Even if it is in spec but near the upper end of the range, it may be worthwhile to reduce it slightly. A cool laser is a happy laser. :) Then to confirm, power down for at least two hours and confirm that the TP11 voltage is within spec after normal warmup.

      However, note that the voltage on TP11 - and thus heater power - will decline as the laser reaches thermal equilibrium especially with the cover in place. The manuals make no mention of this. The bottom line is that I generally just set it according to the tube type and do not measure the tube voltage at all. There has so far never been any problem with original HP/Agilent/Keysight tubes. It's only when doing a rebuild with another type of tube that this may be critical.

      Note that where the laser current has been increased to accommodate a tube with a high dropout current, external ballast has been added, or anything else that increases power dissipation inside the laser, the temperature set-point may need to increased so the voltage on TP11 remains within range after full warm-up. Otherwise it will tend to be too low. And your mileage may vary. ;-)

    13. Replace the cover or shroud and front-plate.

    Note that this voltage will decrease somewhat once the entire laser reaches thermal equilibrium, so it is assumed these measurements are made shortly after the laser locks from a cold start.

    Improving the Frequency Stability of HP/Agilent 5517 Lasers

    The following also applies directly to the 5518A and 5519A/B, which are virtually identical to the 5517s. It could also be used with the 5501B, but additional modifications would be needed since these have other issues.

    (Any circuit modifications below apply directly only to the common Type I Control PCB. They would be more difficult, if even possible, for the Type II Control PCB, and probably not at all for Type III Control PCBs. At the very least, cuts and jumpers would be much more difficult on the dense surface mount PCBs. And, since there are so many of the older Type I Control PCBs available, why would you want to!)

    While HP/Agilent lasers are very good for their intended metrology applications, they can't compare to the best stabilized HeNe lasers like those from Laboratory For Science, Spectra-Physics, and others. There are issues with both short term variation in optical frequency as well as long term frequency drift. The 3 most significant are probably:

    1. HeNe laser power supply ripple: This shows up as a frequency modulation of the optical frequency at 30 to 50 kHz (and harmonics) with a deviation of up to 1 MHz or more. This may or may not be significant depending the power supply version and where on the power-versus-current curve the tube is operating.

      Replace the HeNe laser power supply with a low noise/low ripple type or add an external ripple reducing circuit to its output. The older VMI 148 had particularly high ripple, but even the VMI 217 can stand improvement. I have not tested the older Laser Drive 111-ADJ-1 or the newer VMI 373, PS 253, or PS 503 for ripple, but they should be at least as good as the PS 217. The VMI PS 504 should be even better. See the section: Reducing Ripple and Noise in a HeNe Laser Power Supply with a Switchmode Regulator.

    2. LCD beam sampler and sample-and-hold voltage decay: This shows up as a sort of "hunting" behavior with a period of around 2.56 seconds, resulting in a variation in optical frequency of 50 kHz to 1 MHz or more.

      Remove the LCD panel and its photodiode. Drill a hole in the beam sampler PCB and mount a polarizing beam sampler (e.g., polarizing beam splitter cube and a pair of silicon photodiodes) on top of the PCB. (It might even be possible to build this into the plastic housing instead.)

      The schematic for one possible modification is shown in Upgraded Electronics for HP-5517 Lasers 1. This references the part numbers found on the 5517A/B/C/D Type I Control PCB, and probably the 5518A and 5519A/B as well.

      The dual trans-impedance preamp for the photodiodes generates separate S and P mode signals. These feed the "Subtracting-Sample-and-Hold" circuit modified so that when in "Optical Mode" under feedback control, it passes both straight through - no holds allowed! During "Warmup Mode", it must pass the normal heater drive signal. The added preamp can be made from any stable dual op-amp mounted on a little circuit board perhaps stuck on top of U12, the LF13331D quad JFET analog switch, and attached to the photodiodes via twisted pairs. A 1M ohm pot in parallel with a 1 nF capacitor should suffice for the op-amp feedback, providing enough gain for all but the weakest laser tubes. The op-amp, U101, isn't critical - something like an LT082 would suffice. A few cuts and jumpers will be required, but on the wide open through-hole layout of the Type I Control PCBs, that shouldn't be too difficult. An alternative would be to remove the LF13331D and install an IC socket in its place. Then, build a little PCB that plugs into that with the LF13331D and preamp circuitry on it. Add an offset pot and it will then be possible to fine tune the optical frequency. It may not end up pretty, but should work great! It may be easiest to do the modifications in two stages: First replace the LCD and its PD with the polarizing beam sampler and preamp, and confirm that the correct polarizations are selected - the system should lock normally. Then disable the LCD selection logic so that both signals are passed at all times.

      I later implemented a simpler set of modifications as shown in Upgraded Electronics for HP-5517 Lasers 2. This should produce similar results but with a wee bit less flexibility:

      • The LCD and the original photodiode PCB were removed and replaced with a polarizing beam-splitter cube and a pair of photodiodes. However, rather than adding separate pre-amps for each PD, they were wired in parallel with opposite polarity and connected to the same pins as the original PD. The resulting signal is then proportional to the difference in photocurrent. A two pin connector was installed so that the polarity of the PD signal could be determined by experiment rather than trying to figure it out based on theory. :) This would also enable a 5501B tube assembly to be easily tested, as it would require opposite polarity due to the swap of the F1/F2 axes.

      • Three cuts and jumpers reconfigured the subtracting sample-and-hold. All could be done without removing any ICs:

        1. The upper PD input switch (U12D) was disabled (pin 16 tied to ground).
        2. The upper grounding switch (U12C) was permanently enabled (pin 9 tied high through a 10K ohm resistor).
        3. The lower PD input switch (U12A) was enabled by having pin 1 tied to "-WARM UP" (U3-5).

        Wiring of the lower "POWER AMP DRIVE" switch (U12B) was unchanged (enabled by "DISABLE").

      • A circuit for offset was added so the mode balance could be fine tuned. This can vary the net photocurrent by approximately ±7.5 uA. If I'm ambitious, I'll even drill a hole in the cover so the pot can be twiddled without affecting the thermal environment. As it turned out, the response of the two photodiodes was close enough that the laser locked fine and was very close to optimal with zero offset. But the control would enable the optical frequency to be moved a few 10s of MHz one way or the other.

      • Pin 4, the RESET input of U10, (LM555, main clock) was isolated from its direct connection to +5 VDC and jumpered to "-REF ON" (U1-1). This turns off all the digital circuitry once the laser is locked (READY on solid) to eliminate Control PCB-originated digital noise.

      See Modified Beam Sampler and Offset Adjust Circuit for HP/Agilent 5517 Laser for a photo of these modifications.

      The first two sets of changes were implemented first. These worked fine with the locking characteristic after warmup, total time-to-lock, uncertainty in REF frequency, and slow oscillation in REF frequency appearing very similar to the behavior of an unmodified laser. This is actually a rather surprising and unexpected result, so more study will be required. :) A discrete time system has been converted into a continuous time system without doing anything to the loop parameters and there were no dramatic changes to the system response. Interesting.... However, later I did confirm that actual locking to the modes occurred almost immediately after the laser entered "Optical" mode (about 100 seconds after READY starts flashing). I also confirmed that if the photodiode polarity was incorrect, it would repetitively pass through the lock point at a rapid rate but never stabilize there. I had expected it to lock to the opposite crossover point of the two modes, but apparently the slope there is so much different that it never latched on, so to speak. Or, possibly it would have locked there eventually but I did not wait long enough.

      For the record, the laser first tested with these modifications was a somewhat high mileage 5517C with a power output of around 240 µW and a REF frequency of around 3.3 MHz, the latter being outside the spec'd range for the 5517C (2.4 to 3.0 MHz). The uncertainty in REF frequency may be 200 Hz or more. The variation starts out with a period of around 16 seconds and deviation of around 0.003 MHz. Over a few hours, it slows down and finally stops (or becomes so long as to not be obvious).

      Some tests:

      • The "Power Amp" jumper was pulled off which prevents the optical signal from affecting the short term stability. The integrator is good enough that the lock-point will remain nearly constant for many seconds. Both the randomness and slow frequency variation in REF frequency were essentially unchanged. The heater driver (Power Amp) could still be acting in a peculiar way though this doesn't seem all that likely.

      • The main clock was disabled after the laser locked by grounding pin 2 of U10, the 555. This made no detectable difference in anything, indicating that these anomalies aren't an issue of either some digital signal still affecting the feedback directly or digital noise being picked up by the analog electronics. Even though there was no detectable change in behavior, modification (4) was installed just for insurance as it was easy enough.

      • A 2M ohm resistor was added in series with the Power Amp input. (The normal input is approximately 2.15M in parallel with 0.22 uF). This made no difference (not surprising given the previous result).

        Then a 220K ohm resistor was installed in parallel with the 2.15M ohm resistor. This also had no detectable effect once locked. But, while the time-to-lock didn't change that much, the locking behavior was more rapid after the initial warmup.

      • A low ripple low noise linear HeNe laser power supply (modified SP-248) was substituted for the original VMI-148. This also made no significant difference.

      • An adjustable HeNe laser power supply was also installed to see if running at slightly higher current would have any effect. There may have been a very slight decrease in the randomness but nothing conclusive. At the normal current, this tube does have some amplitude ripple that disappears at higher current, so that may be related.

      • The PBS cube was mounted at a slight angle to reduce the possibility of back-reflections but this also had no obvious effect. The remaining (original) optics of the beam sampler still have perpendicular surfaces that could produce back-reflections, but these are still present when used in the normal way with the LCD and may be worse, since the LCD has several additional perpendicular reflective surfaces.

      Later, I installed the modified Control PCB and beam sampler in a certifiably healthy low mileage 5517B (510 µW, 2.32 MHz). Locking was fine and both the randomness and periodic variation in REF were still present, though subtly different. The amplitude of the randomness was slightly lower - perhaps averaging 50 Hz compared to 200 Hz. The period started at about 10 seconds and the deviation was about 0.0045 MHz. However, running this laser with its original Type II Control PCB resulted in essentially identical behavior. The deviation as well as the amplitude of the randomness also appear to be affected by exactly which longitudinal mode pair (i.e., exact temperature) at which the laser locks. A later power-on cycle resulted in a deviation of almost 0.01 MHz. Turning my offset control too far (accidentally!) resulted in the laser losing lock and then reaquiring it after the offset was turned back toward center. But the behavior had changed! The deviation in particular had dropped from 0.01 MHz to 0.004 MHz or less. Nothing else was different other than (presumably) where it locked!

      The detailed character of these artifacts remains a mystery. The randomness may in fact be a faster but lower amplitude oscillation in REF frequency superimposed on the larger slower one but it's hard to tell without actually recording it, which I'm not sure I am eager to do. :) Since other evidence suggests that there isn't a corresponding variation in optical frequency to go along with the variation in REF frequency, this peculiarity may be a fundamental characteristic of the Zeeman laser and have nothing to do with the stabilization at all. Or, they may be the result of some sort of etalon effect. The time constant of the slow down in the periodic variation in REF frequency is too long to be anything but thermal in origin. HP/Agilent laser tubes have at least 4 planar uncoated glass surfaces outside the laser cavity and these are not wedged or set at an obvious angle to minimize back-reflections. In addition, the optics of the beam expander telescope and beam sampler have several optical surfaces. Since the structures these are mounted on are all mostly temperature independent of the controlled thermal environment of the mirror spacing rod, it's possible that one or more are forming some sort of external resonator with its longitudinal modes interfering with the normal lasing process very slightly. Maybe.

      And eventually, I will have to set up the dual laser setup to check the optical frequency stability.

    3. Thermal effects due to ambient conditions: While the laser cavity itself is well thermally isolated from the environment, the outside of the tube and electronics are not. Variations in the temperature of the overall tube assembly can affect gas pressure inside the tube, and thus the optical frequency. Changes in the temperature of the electronic components can also affect the location where the optical frequency is locked. One simple way to remedy this might be to add a small fan forcing air through the case, with a sensor to control its speed based on interior temperature. Then, put the entire laser in a box to minimize air currents. A more sophisticated scheme could use a heater or TEC(s) to actively control the case temperature.

    A second order effect is external magnetic fields, but this really shouldn't be significant unless other Zeeman lasers are living nearby, or you want to run this inside an MRI machine. :) And for the purist, air pressure and seismic activity also affect optical frequency, but the three modifications described above should reduce the short and medium term (up to days, probably not years) variation by more than an order of magnitude. Long term drift of optical frequency will be dominated by changes in the laser tube gas pressure and fill ratio from use, and this can't be easily controlled. But periodic diddling with the offset pot can compensate for those. :)

    Adding a Mode Balance Adjustment to HP 5517 Lasers

    The F1/F2 power balance should never really need to be changed except for experimentation :) since it's determined by circuit parameters that should not drift. There actually no specification for the acceptable range but a few percent is normal with H power usually larger than V power. Any serious imbalance is probably due to a bad LCD device or dried up electrolytic capacitors.

    So, there should really never be any need for this, but if you have nothing better to do, it's quite trivial to modify the Type I Control PCB to provide an adjustment. Replace R10 (forth 10K ohm resistor to the left of U11) with a series combination of a 7.5K ohm resistor and 5K ohm 10 or 25 turn trim-pot. This should provide a range of <1:1.1 to >1.1:1 in H/V power balance. A single turn is typically a shift of several MHz in optical frequency. In the center (total value of 10K ohms), operation will be unchanged. Note that the circuit for adjusting power balance shown in the previous section will NOT work without performing the other modifications since it will add the same offset into both F1 and F2.

    Agilent 5517 Laser RS232 Communications

    The following likely only applies to the Type III Control PCB. Although there are several unidentified connectors on the Type II Control PCB, tests have not located anything that behaves like an RS232 port, though that hasn't entirely been ruled out.

    And, no, there is nothing labeled "RS232 Port", even on the Type III Control PCB. But, there was a header a with a suspiciously appropriate number of pins (10) near the SHARC chip, so I started looking at voltages. Sure enough, pin 3 on the header had -9 VDC on it, and was occasionally pulsing to 0 V. So, I made up a cable to my ancient Kiwi laptop, and there was ASCII being spit out at 9600 baud! :)

       Header Pin   DB9 Pin   Signal
           3           2      Data from 5517 (transmit)
           5           3      Data to 5517 (receive)
           9           5      Ground

    The "DB9 Pin" is the result of using an IDC cable wired directly to a DB9 connector. These may be salvaged from old PCs as they are often used to attach the mainboard to the rear panel. The pin numbers will be the same on the PC (not swapped). It's 9600 baud and full duplex (the laser echos characters typed). I have no idea about start and stop bits and parity, but suspect they don't much matter.

    The few interesting things I've discovered so far are:

    Perhaps flipping one of those DIP-switches will put it into Verbose mode, but picking the wrong one might erase the Universe, so I'm not willing to risk that - just yet. :)

    At the very least, the runing time is probably maintained in NVRAM and it would be nice to know how to access that!

    If anyone has more information on these digital Control PCBs and their RS232 or other diagnostic port, please contact me via the Sci.Electronics.Repair FAQ Email Links Page.

    More to follow, maybe, but these "First Contacts" are encouraging. :)

    HP/Agilent 5517 Connectors

    All versions of the 5517 lasers have the same rear panel connector for power and REF. See HP/Agilent 5517 Laser Rear Panel Connector for the physical arrangement of the pins. (The connector on the 5518A also has ~MEAS, MEAS, and MEAS signal level, but is otherwise compatible. The pinouts for the corresponding Excel lasers are the same except that they don't require -15 VDC.)

    HP-5517 power/reference connector J2

       Pin      Function
        A       No Connection on 5517 (MEAS signal level on 5518A) (1)
        B       ~MEAS (Not used on 5517) (2)
        C        MEAS    "          "
        D       Signal Return (MEAS)
        E       ~REF (Zeeman beat signal from internal optical
        F        REF  receiver's differential line driver)
       G,H      Ground
        J       +15 VDC Sense
        K       +15 VDC
        L       -15 VDC
        M       +15 VDC
       N,P      Cable Shield
        R       Signal Return (REF)
        S       Ground
        T       +15 VDC
        U       Cable Shield


    1. Pin A goes to the signal level meter on the 5508A display which provides a reading proportional to the current flowing from an internal +5 VDC source to Ground, approximately -2 mA full scale. This is normally generated by the 5518A laser head.

    2. Pins B and C are the ~MEAS and MEAS differential signals, along with the associated return, D. These go from the built-in optical receiver of the 5518A to the 5508A display, but used only for test purposes on 5517s: If JMP8 is in the TEST position on the A3 control PCB, ~MEAS/MEAS track ~REF/REF. Normally they are driven to the de-asserted state by the line driver.

    The 5517/5518A laser head cable connector looks like a standard Amphenol MIL-style bayonet lock type but does not have the default keying orientation but has it rotated 90 degrees. One confirmed part number is PT06A-14-18PX. These are available from various electronics distributors and even (!!) for $20-$50 in singles. The least expensive supplier I've found so far is Powell Electronics. It's essentially the same connector as used on the 5500C laser head cable (PT06A-14-18P) but that has the default keying. One mating connector for the 5517/5518A version from an original HP cable is labeled: 97 USA/CTI 26SOU 851-06P14-18PX50-44. Searching for this part number only seems to result in non-stock items with no listed prices - "Ask for Quote". You know you're in trouble when this is involved! :) Used cables for the 5517/5518A are available for $100 to $300 from various surplus dealers and often on eBay. But the standard cables may be 10 or 20 feet long and much more than is needed if the other parts of the interferometry system aren't being used. For power and reference signals, the mating connector and a few wires should suffice. It would be a pity to chop up an expensive high quality cable simply for the connector. However, it is possible to modify a standard PT06A-14-18P to rotate the shell. And these tend to be somewhat less expensive. See the section: Making HP Interferometer Cables.

    The pinout for HP 10881A/B/C cables with a 5 pin DIN connector (looks like an old PC AT keyboard connector) for power is as follows:

       Pin      Function                 5 Pin DIN Female
     ------------------------                  2 o
        1       Ground                     4 o     5 o
        2       Ground        
        3       No Connect?               1 o       3 o      
        4       -15 VDC                         ___
        5       +15 VDC                        [KEY]

    The color coding of the power wires for the 10881D/E/F cable (which has only spade lugs) is strange:

    HP/Agilent 5517 Mounting

    There are only two styles of mounting arrangements all of these lasers:

    Testing the HeNe Laser Tube in HP/Agilent 5517 and 5501B Lasers

    The following procedure can be used to test the tube in any 5517 laser as well as the 5501B to determine if it is likely to be useful, regardless of the state of the rest of the laser (if one exists). (The tubes in the 5501A and 5500A/B/C cannot be tested in this manner.)

    A working HP/Agilent HeNe laser power supply brick, 15 VDC power supply rated at 2 amps or more, and a laser power meter will be required. The power supply doesn't need to be well regulated for these tests, but it should be well filtered and not go much above 15 V even unloaded. A simple wiring harness should be constructed with mating headers for the 2 and 3 pin connectors on the HeNe laser tube and HeNe laser power supply, respectively:

    Both wires to the 15 VDC power supply should be #18 AWG or larger.

    Note: If the connector on the HeNe laser power supply brick has a white wire in the center position, it requires a control input and cannot be used for testing when not attached to the (rear) Connector PCB using this simple procedure. Most do not, so find a different one. :-)

    Now for the testing:

    1. Attach the 15 VDC power supply to the positive and negative wires of the wiring harness. Make sure these connections are secure using either binding posts or screw terminals on the power supply, wire nuts, or something else.

    2. Plug the power connector from the HeNe laser power supply brick into the larger connector of the wiring harness.

      Both these connectors are keyed and only go one way, don't force. :)

    3. Insert the high voltage connector from the HeNe laser power supply into the mating HV connector of the tube and secure the lock nut.

    4. Apply power. A new tube will usually - but not always - come on instantly. However, some older/high mileage tubes may require a few seconds or longer to start. The tube is lit if the exposed glass section at the back of the tube glows (Long tubes) or from the plastic cover if it's not opaque. Sometimes, shining a flashlight or other bright light on the back of the tube or even through the front will convince it to start sooner.

    5. Assuming there is a beam and it is stable (not flickering or sputtering), use a laser power meter to monitor it as the tube warms up. The power (and brightness) may vary noticeably (especially on lower power high mileage tubes) as the laser cavity "mode sweeps" while it expands.

      • When the tube is installed in the laser and the laser locks normally, the output power is generally at or near the minimum value on the power meter so it's safe to use that as the expected output power.

      • Allowing the tube to warm up for an hour or more will give a better indication of actual usable power as it often increases somewhat. But if the output starts flickering or sputtering, the tube cannot be used at the normal current of 3.5 mA. Even if it is stable for an hour, when run at normal operating temperature in the laser, it may not behave. In either of these cases, it may still be possible to install an HeNe laser power supply with a higher current to get some more life out of the tube. See the next section.

      • If the tube is NOT installed in a laser but is complete (waveplates and beam expander), the usable output power will be about 10 percent lower than what was measured due to losses in the beam sampler at the front of the laser.

      • The raw output from the tube (no beam expander, waveplates, or beam sampler) may be 30 percent or more higher than when complete and installed in the laser.

    6. When the measurements are complete, turn off the 15 VDC power supply and allow a few seconds for its voltage to decay to 0 before disconnecting it. DO NOT CHANGE CONNECTIONS WITH POWER ON.

      WARNING: The high voltage on the HeNe laser tube and power supply takes some time to discharge, up to a few minutes or even more depending on the version of the HeNe laser power supply and other factors. Take care when disconnecting the tube. It's not dangerous but could be a slightly shocking experience.

    Adjusting the Waveplates in HP/Agilent Lasers

    The outputs of the HeNe laser tube inside the Zeeman magnet are (ideally) left and right circularly polarized lasing modes at the optical frequencies F1 and F2. The interferometer requires F1 and F2 to be linearly polarized and perfectly aligned with the horizontal (H) and vertical (V) axes of the laser. Most of the time, these are also the H and V axes of the tube assembly. And perfection here means within a few degrees. A pair of waveplates do the conversion.

    All HP/Agilent lasers from day one have had waveplate assemblies similar to those shown in HP/Agilent Waveplate Assemblies. The type on the left is used in the 5517A, 5518A, and 5519A/B while the one on the right is found in all other 5517s with a 3 mm or 6 mm beam, and the 5501B. 5517s with a 9 mm beam have a waveplate assembly with a slightly larger barrel and aperture. Waveplates in the older 5501A and 5500C lasers use the same black barrels but the mounting differs slightly as do those in the original 5500A/B lasers, where they are part of the built-in interferometer. The remainder of this section is written specifically for 5517s and 5501Bs, but applies with minor changes to the 5500C and 5501A.

    While incorrect settings of the waveplates can result in symptoms from a laser that never locks to poor performance in an interferometer, doing anything to the waveplates is virtually never required unless they have been tampered with. If that is unlikely, first look elsewhere as it's very easy to make things worse. If evidence of tampering is present, both misadjustment as well as damage are quite possible.

    While the physical waveplate assemblies for most small 5517s and 5501Bs are identical (except for size in the case of 9 mm beams), they are NOT directly interchangeable without adjustment because the F1/F2 axes on the two types of lasers are swapped. And a 5501B with an unmodified 5517 waveplate assembly (or vice-versa) will not even lock properly. However, transferring waveplate assemblies among 5517s (any model subject to having a suitable beam size) or among 5501Bs should result in the laser locking without problems and probably being usable, though performance may not be totally optimal.

    The waveplates themselves are made of a thin pellicle of something like optical quality mica glued to the mount. All except the oldest are AR-coated for 633 nm. The pellicles are VERY FRAGILE. Contacting their surfaces for any reason should be avoided at almost all costs. Even the slightest deformation will result in stress lines degrading the transmitted beam, likely making them unusable. And sometimes the glue isn't very strong, so the pellicle could fall off. If cleaning is absolutely necessary, first try blowing off any dust with an air bulb. If that isn't enough, cleaning procedures for delicate laser cavity mirrors must be employed. And there may still be unavoidable damage. Only if contamination is so severe that there is obvious scatter in the transmitted beam, should anything beyond blowing off dust be considered.

    The one closest to the tube is (nominally) a 1/4 waveplate (QWP) which converts the left and right circular polarized Zeeman modes to F1/F2 orthogonal linear polarization. The second one is (nominally) a 1/2 waveplate (HWP) that rotates F1/F2 so that they are aligned with the H and V axes of the laser. If the outputs of the laser tube were truly pure circularly polarized, then the HWP would be unnecessary as the orientation of the QWP could be set to do this. However, the lasing modes from real lasers are often, if not always, slightly elliptical, in which case only four specific orientations of the QWP will produce outputs that are linearly polarized. But they almost certainly won't be aligned with the H and V axes, so subsequent rotation by the HWP is required. And even using both the QWP and HWP may be a compromise depending on the specific character of the elliptical modes - perfection would only be possible under special conditions including the requirement that the major axes of the elliptical polarization states be orthogonal with the same ratio of major and minor axis amplitude.

    A single waveplate with fully adjustable retardation and orientation can convert any single input polarization state to any output polarization state. This also applies to the conversion of a pair of orthogonal input states to a pair of orthogonal output states (subject to certain restrictions) as in the HP/Agilent lasers. But wide range variable waveplates - typically based on a pair of optical wedges or tilt of a thick slab of birefringent material - tend to be large, costly, and may introduce problems of their own. The QWP and HWP used in HP/Agilent lasers can perform the same transformation, are very compact, relatively inexpensive, and adjustment is somewhat easier with a separate QWP and HWP than if combined into one.

    My first comment on doing any adjustments of the waveplates is: "if it ain't broke, don't fix it.". :) This would also be my 2nd, 3rd, and nth comments. :-) I have *never* seen an unmodified HP/Agilent laser whose performance wouldn't be acceptable without waveplate adjustments, even one that's over 30 years old. As the tube is run and degrades over time, some subtle aspects of the polarization may change but this is probably not worth worrying much about. And some lasers may even come from the factory less than totally optimal, but passing all tests and thus good enough.

    If the waveplate assembly needs to be removed, label both the orientation and front-back direction as the two barrels appear similar. Take closeip photos before removal if possible! There is *usually* a red dot on the QWP and an orange dot on the HWP, but not always. If the waveplate assembly is installed backwards, there will still be orientations of the individual waveplates where the result will be correct, but the behavior will be even more confusing during adjustment. And it is almost certain that none of the screw holes will line up. In-depth analysis on this topic utilizes esoteric stuff like Jones Calculus, Poincare spheres, and polarization ellipses. This is left as an exercise for waveplate theorist types or techno-masochists. ;-) Fortunately, there are really only three things to know when adjusting the waveplates in real as opposed to ideal Zeeman lasers:

    1. The QWP converts the elliptical Zeeman modes to linearly polarized modes at four specific angles of the QWP that are 90 degrees apart.

    2. The HWP rotates the linearly polarized modes to line up with the H and V axes of the laser.

    3. Adjustment of the tilt of one of the waveplates is used to fine tune the pair to produce a result similar to that from a perfect QWP and HWP.

    One implication of (1) is that it is only necessary to go through 90 degrees of QWP rotation to have covered the entire range. And (3) means that if originally set close to QWP and HWP, adjustment of only one of the tilts is required but generally none at all.

    The only actual waveplate adjustment procedure I've ever seen was for the earliest HP-5500A laser with built-in interferometer optics, found in the HP document: 05525-9000, "5525 Preliminary Operation and Service Manual". After paraphrasing and simplification, it can be summarized as:

    And of course, there is no explanation as to why this procedure is the way it! But it does make sense in the context of the three points, above. In essense, this is not that different than what needs to be done for newer laser with no interferometer built in.

    However - and this may be critical for your sanity - the original 5500A/B locked the laser using the waste beam out the back of the tube, so one could twiddle the waveplates in the output beam without worrying about the laser losing lock. This is not true of ALL subsequent lasers. Thus if the waveplates are far out of adjustment to begin with, the laser will fail to lock. Or if they are moved too far on a laser that is locked, lock will be lost. Either way, there could be much frustration. The full remedy may be more trouble than it is worth to adjust a single laser. More below. But it can be avoided at least on a laser that already locks by (1) marking and photographing the waveplates before touching them so the original settings can be restored if necessary and (2) going in very very small increments so lock is never lost.

    And there is a simpler procedure (at least in principle) if the waveplates are to be adjusted on a bare tube if it can be rotated through at least 90 degrees. More on this below.

    Basic adjustments

    The small black barrels with front rings and holes around their periphery are compact cleverly designed mounts providing for rotation and tilt. For rotation, the entire barrel is held in place against a rubber O-ring and is free to turn (though originally locked by the genuine certified HP blue paint) and sometimes considerable effort is required. Tilt adjustment is provided by a cam and shaft inside the barrel when the ring at the front is turned relative to the barrel. The tilt axis is aligned with an optic axis of the waveplate and can slightly vary the retardation. Since the mica or whatever is composed of several thin layers, it can't be cleaved to the exact thickness required for a QWP or HWP. The total range of retardation adjustment is rather limited (and much smaller than would be required to convert a QWP into a HWP or vice-versa). But it is enough to enable the combination to be fine tuned rather than requiring precise and expansive waveplates.

    Adjustments are performed with blunt thin tools stuck in any of the 12 holes surrounding the inner barrel (rotate) and outer ring relative to the inner barrel (tilt). I use a pair of straight dental picks ground down to just fit the holes. HP offered a special 0.05" tool, for which they probably charged an arm and a leg. :)

    Initial settings of waveplates

    Where the blue paint has been obliterated and it's obvious that someone has totally messed up all four degrees of freedom, it's straightforward to identify orientations that should allow the laser to lock and to restore them to pure 1/4 and 1/2 wave if necessary. If the blue paint is intact and/or the laser locks properly, skip this step!

    The QWP is supposed to be closest to the laser tube and should have a red dot. The HWP should have an orange dot. (If the dots are reversed, someone may have installed the waveplates backwards. Perhaps that's your main problem!) Identify and label the optic axis of each waveplate by locating the where the tilt shaft pokes through the inner barrel close to the mounting plate. (If the shaft end is not visible, remove barrels from the overall waveplate assembly and examine their interior. The optic axis is identified by the two dots or scratches on the pellicle that line up with the tilt axis and shaft. Label them on the outside and put the waveplate assembly back together.)

    A procedure to more precisely locate the optic axes and set up tilt for the QWP and HWP can be found in the section: Adjustable QWP and HWP Waveplate Setup: but it's almost certain this is unnecessary unless the waveplates are totally mislabeled. So, first try the "Laser lock test". Only if that fails, consider the complete setup, though some other cause for an inability to lock is more likely.

    Laser lock test

    Install the waveplate assembly with the QWP closest to the tube. Orient the QWP with its optic axis at -45° (CCW as vewed from the front of the laser) for a 5517 or +45° (CW) for a 5501B. Orient the HWP with its optic axis vertical. Under ideal conditions, the QWP will do all the work of converting circular to linear polarization and the HWP will simply pass the H and V modes without alteration. Even if not ideal, these settings should allow the laser to lock.

    These don't have to be precise down to 1 second of arc. :) Within a few degrees should be close enough for the laser to lock, but not for use in an interferometer. If the laser does not lock, HP/Agilent labeling of the waveplate optic axes may not be consistent and they are set for the wrong laser type (5501B or 5517). Readjust the HWP and try again. Once the laser locks, fine tune using the adjustment procedure below. If it doesn't lock with either setting, one or both tilt settings may be too far off, or more likely there is some other problem with the laser.

    The laser tube assembly must be secured to the laser baseplate and the entire laser must be prevented from moving during adjustments to avoid loss of lock and confusing the measurements.

    Setting up the laser for waveplate adjustments:

    All lasers except the original 5500A/B will lock and remain locked only over a limited range of orientations of the QWP and HWP because they use the linearly polarized output of the waveplates for feedback. If the waveplates are too far out of adjustment or get too far out of adjustment, lock will be lost, READY will go off and then start flashing, and about 2-1/2 minutes will elapse before lock is reacquired - and only if the setting is restored to within that limited range. Otherwise, the laser will continue trying and failing FOREVER. If a "good" setting wasn't documented, this can end up being a frustrating task. The easiest (if not quickest) way to get back to a locked state would be to rotate the HWP in 10 to 15 degree increments, allowing the laser to attempt to lock after each change with READY on solid. This should restore lock in less than 90 degrees of rotation.

    However, for the 5501B and all 5517s, 5518A, and 5519A/B, setting the REF jumper on the (Type I or II) Control PCB to "LO" (on, second position from the right) after READY starts flashing will force the state machine to think there is a valid reference signal and remain in analog control once it tries to lock. While adjusting the waveplates may still cause lock to be lost, it will be reacquired within a few seconds when the waveplates are back within range. However, READY will always be on, so REF must be monitored with an oscilloscope or frequency counter to know when its present and stable.

    Adjustment procedure for complete waveplate assembly

    If the waveplates are already set so the laser locks, make scratch marks (the blue paint isn't reliable) and document the exact positions with photos to be able to get back there if necessary!

    While there are many equivalent ways to do this. The following requires only a linear polarizer on a rotation mount, an HP/Agilent 10780 optical receiver (A, B, or C preferred), and a fast responding analog meter or oscilloscope. These 10780s have a focusing lens and built-in polarizer oriented at 45 degrees. (The 10780F/U lacks these and is thus less convenient.) With the polarizer aligned with the H axis, the AC component of the signal will be zero when the waveplates are adjusted perfectly. This is because only F1 (5517) or F2 (5501B) will be present and no beat is possible. The use of the 10780 is nice because it has high gain and a large dynamic range. (The H axis is specified for consistency but the V axis could also be used.)

    During adjustment, the Signal Strength test-point of the 10780 is monitored on an oscilloscope or an *analog* Volt Ohm Meter (VOM). (Digital multimeters are too slow to catch the fluctuations during final adjustments.) Mount the 10780 in the output beam horizontally or vertically. With its built in polarizer at 45 degrees, there should be a strong signal at the REF frequency once the laser is locked since both F1 and F2 will be present. Place a linear polarizer and orient for minimum signal near the H axis. If the waveplates are performing perfectly, the signal will go to (nearly) 0 V when the polarizer is precisely horizontal since only F1 (5517) or F2 (5501B) will then be present in the beam. If the dip is NOT aligned with the H axis, a slight rotation of the HWP by half the discrepancy in the opposite direction will place it there. Very small incremental rotations of the QWP at twice the rate of the HWP in the same direction will result in the dip remaining aligned with the axis.

    As adjustments of the QWP and HWP orientation are done in very small increments, the depth of the dip will either increase or decrease. Go in the direction that minimizes the signal at the bottom of the dip. And "small" means just about the least change that can be made. As the optimal settings are approached, the changes need to be even smaller - on the order of 1 degree or less. There's stiction due to the rubber O-rings so it's real easy to slip past the optimal orientation. Note that the level on the test-point of the 10780 receiver is very non-linear with high values compressed. So, even a small dip may already be pretty good. This is also affected by the laser output power and overlap of the beams on the optical receiver. See the section: HP-10780 Optical Receivers for sample data.

    You're probably thinking that these directions for adjusting the relative orientations of the waveplates are incorrect. They are at least counterintuitive. Since the orthogonal F1/F2 linear polarizations rotate in lock-step with the QWP, it would be more natural to assume that the HWP would need to be rotated in the opposite direction to maintain the polarization axis unchanged. However, there's no requirement for the laws of birefringent optics to be intuitive to humans. :) A linearly polarized input to a HWP at an angle of +a° relative to its optical axis results in an output polarization at -a°. So, for example, suppose the initial conditions are that the QWP is oriented to produce polarizations aligned with the H and V axes of the laser, then the HWP should be aligned so its optical axes are at H and V as well so it passes the beam without any change. Now rotate the QWP by -10°. The output of the HWP will then need to be at +10° relative to H and V and it will be necessary to rotate it by -5° to restore the polarizations so they are again aligned with H and V.

    I the best settings are found using only rotation of the QWP and HWP don't result in a really low minimum signal, should adjustment of tilt be considered. Tilt adjustment can be done on either the QWP or HWP, though for reasons that aren't entirely clear, it seems to work better on the QWP. While holding the barrel fixed, turn the outer ring in the direction that reduces the signal strength at the bottom of the dip. Go back and forth between tilt and the rotation adjustments to further optimize it.

    In summary:

    1. With the laser and its tube rigidly mounted so they won't move around during adjustments, allow it to warmup and lock. Set up a 10780 (A, B, or C) optical receiver in the beam and monitor the voltage on the 10780 test-point with a fast responding meter or oscilloscope. It should be maxed out. Install a polarizer aligned precisely with the H axis between the laser and 10780.

    2. Alternately adjust the QWP and HWP orientation in very small increments to minimize the voltage.

    3. Adjust the QWP tilt without changing its orientation to minimize the voltage. Goto (2) until no further improvement is possible. ;-)

    4. If available, confirm proper operation in an interferometer.

    Not surprisingly, this is essentially similar to the old HP procedure for the antique HP-5500A!

    Perfect orthogonality may not be achievable (or at least detectable) with all tubes or lasers. Amplitude ripple in the laser output due to HeNe power supply ripple (mostly older bricks), or in the case of the 5501B, from the heater PWM as well, will also be detected by the optical receiver and increase the monitor voltage. If the output from the 10780 is displayed on an oscilloscope, the normal waveform will nearly or totally disappear leaving behind a fuzzy one at the ripple frequencie(s). And, a laser outputting 600 µW will probably not result in as low a minimum signal as one at 150 µW!

    Also, the depth of the dip can vary significantly in a slow periodic manner even when not touching anything or even breathing. So you're not imagining it! While the explanation is not known, it is surmised that reflections from the various surfaces inside and outside the tube result in interference which changes due to thermal expansion and slightly affects the Zeeman modes. Insert more hand waving here. :)

    Adjustment in an interferometer

    However, since I do a fair amount of testing of HP/Agilent and other metrology lasers, I have built a permanent setup including a plane mirror interferometer with 10780C optical receiver, an oscilloscope (and frequency counter) for monitoring the MEAS signal and an analog Volt Ohm Meter (VOM) for monitoring the 10780C optical receiver signal strength test-point. My moving "stage" is a plane mirror glued to a loudspeaker driven with a triangle waveform from a function generator. See: Diagram of Two-Frequency Interferometer Laser Tester. Adjusting the waveplates is then a matter of minimizing the MEAS signal strength when the stage is stationary with a polarizer inserted in the beam at the output of the laser (beyond the waveplates) oriented along the H axis. The signal should approach zero because (ideally) only F1 is present and a 10780 optical receiver set at maximum sensitivity will drop out (signal detect LED goes off and produce no output) except possibly with a very high power laser.

    Almost the same effect can be achieved by blocking the MEAS beam path. Then the interferometer will only return REF, which will ideally have only a single component (F1 or F2 depending on the orientation of the interferometer and laser type) with the waveplates optimized. However, this can't distinguish between impurity and misalignment, and never seems quite as sensitive to them. Thus I prefer to use the polarizer on a rotation mount.

    With the stage in motion (no polarizer or beam block) and optimal waveplate settings, the MEAS signal waveform should remain clean and only change in frequency, though there may be some edge jitter ("fuzz") from various sources including imperfect interferometer alignment and current ripple from the HeNe laser power supply. If the adjustment is slightly off either way, there will be what might be described as leading or trailing "feathery tails" along the tops and bottoms of the signal but the frequency will still be observed to change. If further off, the edges of the signal will be fuzzy but not change frequency at all. If still further off (but the laser may not remain locked), the signal will be clean and unchanging.

    As a sort of confirmation, if the polarizer is replaced in the output beam, (still aligned with the H and V axes), and there is a signal at all with the stage in motion, its quality will be horrible under these conditions when the waveplate setting is optimal.

    Adjustment procedure for QWP only

    Having said all that, the QWP alone (removing the HWP entirely) may result in acceptable performance, at least for hobbyist use. :) The HWP can be added later if necessary. And in some cases, it could be optimal. Where the HWP is dirty or damaged, this may be the only choice! Two degrees of freedom (and the temptation to fiddle with them!) are eliminated. However, due to tilt's limited range and its non-monotonic behavior, adjustments are less predictable than using only rotation of both waveplates.

    Using a photodiode with load resistor and oscilloscope may be better for the QWP-only setup than the 10780 because if the purity isn't quite as good, the signal level may end up in the range where the 10780's response is quite compressed.

    Remove the entire HWP barrel - NOT the pellicle itself! - and replace the ring that was securing it so the QWP can be adjusted normally. Set the tilt about half way or leave it alone if it doesn't appear to have been tampered with. Install the QWP with its optic axis at -45° (CCW, 5517B) or +45° (CW, 5501B) relative to the vertical. This should allow the laser to lock. If it does not, rotate the entire QWP by 90 degrees (the labeling of the optic axis is not consistent). Once locked, alternately adjust the orientation and tilt of the QWP to minimize the AC component passed by a polarizer aligned with the H axis as above so that (ideally) only F1 (5517) or F2 (5501B) is present. Then rotate the polarizer to the V axis. There should be a similar dip, though it may not be quite as deep since the settings were tweaked for the H axis.

    With only the QWP, there will be cases where mode purity and alignment cannot be optimized as well as with the HWP present. On a test with a healthy tube, using only the QWP resulted in nearly the same performance as with both waveplates. But on a high mileage one, there was 2 to 3 percent impurity and tilt was at one end of its range. But this was still good enough that the MEAS waveform appeared perfectly clean in an interferometer with a moving plane mirror. However, the impurity was way under 1 percent and undetectable on an oscilloscope with both waveplates. No real conclusions can be drawn from two data-points except that using only a QWP may be worth trying.

    Beam sampler modification to prevent loss of lock

    Even setting the REF jumper to LO isn't ideal since lock can still be easily lost, even if for a few seconds. A way to maintain the lock point while still allowing arbitrary adjustment of the waveplates would be more convenient. So a spare 5517 beam sampler was modified such that the waveplate assembly could be removed from the tube and mounted out front where any adjustments only affect the output beam and not locking. However, constructing a rig like this doesn't make sense to adjust a single laser.

    The simplest approach is to add only a QWP (which should be adequate for locking) *inside* the beam sampler just beyond the first 45 degree beam-splitter mirror. So a QWP pellicle was popped out of the waveplate assembly from a dead 5501A laser tube and stuck in a slot cut in the plastic beam sampler housing. The waveplate axis (marked as a pair of dots or scratches on the pellicle) must be at a +45° or -45° orientation (depending on the laser type, flip a coin). It must have been my lucky day because lock was successful on the first attempt using a 5517 controller. :) As noted above, the pellicle is extremely fragile. Even a slight bend will delaminate it and introduce a permanent unsightly blemish, though the one I did this to still works well enough.

    The result can be seen in: HP/Agilent 5517B/C/D and 5501B Waveplate Adjustment Adapter. The 3 hex standoffs are mounted so the waveplate assembly is a snug fit and only a single screw at most is needed to secure it.

    A similar approach could in principle be used with the 5517A, 5518A, or 5519A/B except that removing the waveplate assembly from one of these laser tubes requires a combination of a jack hammer and TNT as they are attached with 5 ton adhesive! The remaining details are left as an exercise for the highly motivated student. :-)

    Once modified, the laser can't be used with a normal (complete) laser tube assembly (with its waveplate assembly installed) since there is already a QWP inside the special beam sampler. But the modified beam sampler assembly with waveplate adjustment adapter attached can easily be swapped with a normal one when testing complete tube assemblies.

    Note that with the QWP inside the beam sampler set up so it will lock with a 5517 controller, no change is required to adjust waveplates for 5501B tubes as the lock point is the same. The waveplate assembly will be adjusted as appropriate for the specific laser model. Aside from the REF frequency difference, the waveplate orientations and label on the magnet :) are what distinguish a 5517B/C/D from a 5501B tube assembly. However, since the laser remains locked regardless of what is done with the waveplates, there is a risk of going too far and end up optimizing for the wrong laser type. So this must be confirmed by noting how the MEAS frequency changes with the direction of motion in an interferometer. For a 5517 laser, MEAS decreases when F1 is the measurement beam and the path length is decreasing; opposite of a 5501B. Or reinstall in the tube assembly, install that in the laser, and check if it locks!

    With this scheme, it became straightforward to adjust a bunch of waveplate assemblies for close to optimal performance. But fine tuning would still be required when actually installed on a tube assembly.

    Waveplate setting for tube in magnet

    This is probably more useful for non-HP/Agilent/Keysight tubes such as those from barcode scanners and the like but can in principle be used for them as well.

    Where it's possible to power the tube alone and adjust its orientation over at least 90 degrees, there is a very simple procedure that should quickly location the optimal orientations for the QWP and HWP. This doesn't address waveplate tilt but that can be dealt with separately if desired.

    A polarization on a rotation mount, fast photodiode, and oscilloscope will be required. The tube can be placed on V-blocks. Ideally, the polarizer should be motor driven so it doesn't need to be touched, but if its mount is clamped down, that should suffice.

    So here goes:

    Power the tube and watch for the beat on the scope. It will appear for only a portion of the mode sweep cycle.

    Incrementally rotate the tube in steps of a few degrees and stop. When the beat appears, rotate the polarizer back and forth by at least 90 degrees and note the minimum beat signal amplitude. Repeat this procedure over a range of tube orientations of 90 degrees to locate the minimum of the minimums. There will be a specific orientation of the tube where that is very close to zero.

    If it is necessary for the tube to be in a particular orientation, rotate the tube along with both waveplates by the same amount to make it so.

    Rotate the HWP to line up the polarized f1/f2 components with the X and Y axes, and for the desired motion polarity, i.e., 5501 or 5517.

    This assumes that both the QWP and HWP are perfect. In reality they may not be, so your mileage will vary.

    Sam's Lab Rat 5517/5501B Laser

    This is a small 5517 laser chassis that has had a variety modifications and additions done over the years to facilitate testing of 5517, 5501B, or other HeNe laser tubes for Zeeman testing, experiments, and general hacking.

    Changes to the Type I Control PCB (see Modified Type I Control PCB in Sam's Lab Rat 5517/5501B Test Laser):

    1. Installed switch (in place of jumper) for HEATER: Normal and Off. This plugs into the HEATER jumper block JMP7. (Top-right.)

      This is more convenient than moving the jumper (except when it is accidentally left in the wrong position).

    2. Added jumper to select 5517 or 5501B tube locking. This repurposes the LIQUID CRYSTAL jumper block JMP3 to select the polarity of the feedback error signal. (Top-center.)

      The only difference between 5501B and 5517 tubes is the orientation of the F1/F2 polarized components in the output beam, so this flips the polarity of the feedback signal so either will lock.

    3. Added mode balance adjustment. (Center.)

      Initially this was a 10K ohm single-turn trim-pot in series with a 5K ohm resistor that replaced R10 (10K) in sample-and-hold circuit (shown). But it was found that the sensitivity of the setting was too high so it was changed to a 5K ohm 25 turn trim-pot and 7.5K ohm series resistor. With those, the sensitivity is several MHz/turn.

      This is mostly for experiments with non-HP/Agilent/Keysight tubes for measuring and matching optical frequency. Adding a switch to select adjustable or fixed (for the default) mode position would be desirable.

    4. Added adjustment 500K ohm trim-pot in place of R17 (316K) to reduce gain of photodiode preamp to accommodate laser tubes with raw power above about 1 mW. (Center.)

      It was found that the preamp for the PD following the LCD beam sampler was saturating resulting in an inability to lock if a tube with a power of more than 1 mw was installed. This could apply to lively N1211As as well as non-HP/Agilent/Keysight tubes.

    5. Added jumper to select use of separate temperature sensor and sensor connector for non-HP/Agilent tubes with heater having a temperature coefficient of resistance close to zero resulting in little or no increase in resistance during warmup. (Bottom-center.)

      This isolates the temperature feedback input at R21 and provides a small current through a PTC temperature sensor to generate a voltage based on temperature that has similar behavior as the normal tube heater. This sensor is typically 20 feet of #36 copper magnet wire smushed together to wrap ~180 degrees around the tube.

    6. Added "heartbeat" LED + resistor between pins 1 and 16 (T1.28 and +5V) of thet Test IC socket (U9). (Upper-center.)

      This was installed only because there was an issue once with the main clock not running resulting in the laser getting stuck in the warmup state, possibly due to a metal sliver causing a short, but it has disappeared So this doesn't really count as an enhancement. :)

    Other changes:

    And no doubt there will be other enhancements as time goes on. ;-)

    Replacing the 5517 Control (A3) PCB with µSLC1

    The Control PCBs in 5517 lasers (called A3 by HP/Agilent/Keysight) are very reliable and well engineered (see the info on this above). So "if it ain't broke, there is probably little reason to fix it", unless it really is broke. ;-) The HP through-hole version is well documented with service information readily available and generally straightforward to repair unless a proprietary part is bad. But the more modern SMD version from Agilent, and now Keysight, uses nearly all surface mount pars and publicly available documentation is next to non-existent. Repair is therefore much more difficult.

    Alternatively, one might have acquired a 5517 laser with the A3 PCB missing, or simply want to their hand at implementing a modern digital replacement.

    Doing so is very straightforward. What follows is only a suggestion using Micro Stabilized Laser Controller 1 (µSLC1).

    (The following also applies to the 5501B, though some details may differ.)

    Here is the general procedure:

    1. Remove the existing A3 PCB if present. Nothing on it will be required except possibly Q3, the PNP Darlington heater driver transistor..

    2. Remove the Beam Sampler PCB, LCD panel. and "Zebra Stripe" elastomer contact strips, but don't disturb anything else inside. The beam exiting the top will have the orthogonal X and Y polarized components while the beam exiting the side will be varying in intensity at the REF frequency.

    3. Mount a Polarizing Beam Splitter Cube (PBSC) on top and install a pair of photodiodes to "view" the separated polarizations. A 10-15 mm PBSC is recommended so it intercepts the entire beam regardless of diameter and imperfect alignment. (The original Beam Sampler PCB with its photodiode can be mounted on the side if using SG-OR3 for REF. More below.)

      Note that the original Beam Sampler and REF photodiodes have lenses that help somewhat with beams that are not centered and/or are larger than the PD area. Without lenses, alignment may be more finicky. Using larger photodiodes here may be worthwhile.

    4. Provide mating connectors. A replacement for the main connector (A3J1) may be difficult to identify, but a common 12x2 pin female to male header can be used in its place. The Beam Sampler connector is just a 4 pin header, or just remove it and solder directly to the PCB.

      Main Connector (to Connector PCB):

        Pin  Function        Pin  Function
         1   NC              13   NC
         2   HEATER SENSE    14*  +5 VDC (From A3)
         3   MEAS            15   -MEAS
         5   NC              17*  -READY
         6   -15 VDC         18   -15 VDC
         7*  +15 VDC         19*  +15 VDC
         8*  +15 VDC         20*  +15 VDC
        12*  -REF            24*  REF

      The "*" denote the 8 required signals/connections.

      Beam Sampler Connector:

        Pin  Function
         1*  PD Cathode
         2*  PD Anode
         3   -CRYSTAL (LCD)
         4   +CRYSTAL (LCD)

      The "*" denote the 2 required signals.

    5. Since &miccro;SLC1 is designed to run on a maximum of 12 VDC, either install 4 silicon diodes in series with the 15 VDC supply, or a 12 V linear regulator or switch-mode step-down converter set to 12 V. The latter would minimize heat dissipation.

    6. The heater in the 5517 laser tube is connected to its cathode internally and GND in the laser. Therefore µSLC1 cannot drive it directly since it needs the heater to be floating or have one end connected to +12 VDC. One approach is to add a PNP Darlington power transistor like the MJE700G to do the conversion. (This is the same one that may be Q3 on the A3 PCB.) The only other required part would be a resistor. Details will be forthcoming. ;-)

    7. If it is desired to be as compatible with the original 5517 operation as possible, add an optical receiver to monitor the output from the side of the Beam Sampler. This can be any version of a 10780, but a 45 degree mirror will required if it is desired NOT to stick out of the case. SG-OR3 can also be used with a short twisted pair to the photodiode, which can be the one the Beam Sampler PCB, stuck in the side. Nothing is needed in front of either.

    8. To be fully compatible with a stock 5517, REF should be present only when the laser is READY. With SG-SLC1, this means when the LOCK signal is asserted. For a 10780, where this can be done may depend on the specific version. There are several places on the SG-OR3 PCB where the output can be inhibited. In both cases, it doesn't matter if the REF/-REF signals are high or low as long as they aren't changing.

      And if it is desired for the LEDs on the back of the laser to behave normally, some additional logic may be required - hardware and/or changes to the SG-OR3 firmware.

      These are left as an exercise for the student. Contact me for assistance.

    Everything else is basically the same as constructing any stabilized HeNe laser and is similar to what is in my kits for these using µSLC1. A 5517 version may even be available soon. ;-)

    Common Problems with HP/Agilent 5517 Lasers

    Here are some of the faults that typically occur with the 5517A/B/C/D lasers. Those found in the 5501B, and the 5518A/B and 5519A/B should be similar since the tube design and electronics are virtually identical. The comments with respect to tube issues also apply to the 5501A and 5500C lasers, though the details of the remedies will be slightly different.

    The following are causes that may produce a variety of symptoms: