TV and Monitor Deflection Systems


[Document Version: 1.13] [Last Updated: 05/25/1998]

Chapter 1) About the Author & Copyright

TV and Monitor Deflection Systems

Author: Samuel M. Goldwasser
Corrections/suggestions: | Email

Copyright (c) 1994, 1995, 1996, 1997, 1998
All Rights Reserved

Reproduction of this document in whole or in part is permitted if both of the following conditions are satisfied:

  1. This notice is included in its entirety at the beginning.
  2. There is no charge except to cover the costs of copying.

Chapter 2) Introduction

  2.1) Acknowledgement

Special thanks to Jeroen H. Stessen ( for
his extensive contributions to this document.


WARNING: Read, understand, and follow the recommendations in the document:
"Safety Guidelines for High Voltage and/or Line Powered Equipment" before
attempting any TV or monitor repairs.

  2.2) Scope of this document

TVs and most computer and video monitors depend on the use of similar (at
least in concept) circuit configurations to generate several outputs:

* Current waveform required in the deflection yoke coils of the CRT for linear
  sweep of the electron beam to create a high quality (geometry and linearity)
  picture.  This is close to a sawtooth but not quite.

* CRT High voltage (20 to 30 KV or more) required to accelerate the electron
  beam and provide high brightness and sharp focus, as well as other related
  voltages - focus and screen (G2).

* Various auxiliary power and signals for other subsystems of the equipment
  (low voltage, CRT filament, feedback, etc.).

This document addresses the basic principles of operation of these types of
deflection systems.  While most people with any familiarity with TV or monitor
opration or repair have some vague idea of how these circuits work (probably
just enough to be dangerous), many are incorrect or at least very incomplete.

Most of this information applies to the horizontal deflection which operates
at the higher frequency in a raster scan display (except for peculiar rotated
portrait formats where the functions of the horizontal and vertical scan are

Equipment which utilizes this circuitry includes TV (direct view as well
3-CRT and light valve projection types), computer and video monitors, tube
based video cameras (e.g., vidicon), and other magnetically deflected CRT

Vertical deflection circuits are much less complex due to the lower scan rate
(e.g., 50 to 120 Hz V as compared to 15.734 KHz H for an NTSC TV or up to 120
KHz or more for a high resolution computer monitor).  Most of the control and
output drive circuitry is contained in a special vertical chip in modern

Chapter 3) Horizontal Deflection System Fundametnals


  3.1) How does the horizontal deflection circuit work?

Although there are many variations, the basic operation of the horizontal
deflection/high voltage power supplies in most TVs, monitors, and other CRT
displays is very similar.

Since the flyback is constructed with an air gap in the core, it behaves more
like an inductor than a transformer as far as the primary drive is concerned.

There are scan rectifiers and the coupling factor with the primary is decent.
But they make no use of the stored magnetic energy, they load the primary
directly during the scan part. They do not cause an increase of the stored
magnetic energy so a heavy load is not a problem.  The flyback rectifiers
(especially the EHT) draw from the stored magnetic energy.  When the secondary
load increases, also the magnetisation current will increase.  Ultimately this
will cause saturation of the ferrite core. Excess beam current is a common
cause for this and should be avoided by the beam current limiter.  The
advantage of a flyback rectifier is that it provides 7 times more volts per
winding than a scan rectifier.

                 _  _
                  \/                                    _/\_
 B+ ------+     +----|>|-----+---o +V1  B+ ------+     +----|>|-----+---o +HV
         o )|:|( o  Scan     |                  o )|:|(   Flyback   |  
           )|:|(  Rectifier _|_                   )|:|(  Rectifier _|_ 
           )|:|(            ---                   )|:|(            ---
           )|:|(             |                    )|:|(             |
     _/\_  )|:|(             |              _/\_  )|:|( o           |
HOT ------+     +------------+--+      HOT ------+     +------------+--+
                               _|_                                    _|_
                                -                                      -

Here, V1 is just a typical example of an auxiliary supply derived from a scan
rectifier and HV is the best known example of the use of a flyback rectifier.

For understanding the working of the deflection circuit regard the flyback
transformer as an inductor.  The airgap stores energy, some of which may be
tapped off during flyback by secondary rectifiers (e.g., vertical deflection,
signal circuits, and high voltage supplies) and non-rectified loads (e.g.,
filament supply) but these have hardly any influence on the basic working

The scenario described below is only true in the steady state - the first few
scans are different because the picture tube capacitance is still discharged.
This represents a short-circuit at the secondary side of the flyback. It
prevents proper demagnetizing, hence the core will go into saturation (unless
special soft-start measures have been taken, like a VB supply that comes up
slowly). Generally, a hard start of the line deflection circuit represents a
very heavy load on the HOT.  This will happen after a picture tube flashover
or if the B+ is connected suddenly (due to intermittent contact) and can mean
instant death to the HOT due to secondary breakdown.

  3.2) Basic deflection circuit operation

The following description is only the basics.  For more information, see
the article by David Sharples in "Electronics World", June 1996.

A very simplified circuit is shown below - many components needed to create
a practical design have been omitted for clarity.  First concentrate only the
portion of the schematic shown below to the left of the yoke components:

                Part of T2  )|:|
                   Flyback  )|:|
                   Primary  )|:|
                           |                     <- Yoke components ->
                           |       |         |          |
          Horizontal       |       |         |          +
            Drive          | C     |         |           )|:|
              T1       B |/      __|__      _|_          )|:| L2
         --+     +-------|       _/_\_      ---          )|:| Horizontal yoke
    Driver  )|:|(        |\        |   D1    | C1       +
    Stage   )|:|(     Q1   | E     | Damper  | Snubber _|_ C2
    (not    )|:|(     HOT  |       |         |         --- S-correction
     shown  )|:|(          |       |         |          |
        +--+     +---------+-------+---------+----------+
       _|_                _|_
        -                  -
  Signal Ground      B+ Return (may not be signal ground)

The current in the flyback primary and collector of the HOT are *not* equal.
The horizontal deflection yoke, damper diode, HOT collector, snubber HV
capacitor(s), and flyback primary all connect to the same point.  We begin our
adventure at the end of the scan - retrace - when the flyback period begins:

1. At the end of scan, current is flowing through the flyback primary to the
   HOT, Q1.  At the start of the flyback period, Q1 turns off.  (This must be
   done in a controlled manner - not just a hard shutoff to minimize stresses
   on the HOT - but that is another story).  Since current in an inductor
   (the primary of the flyback has inductance) cannot change instantaneously,
   the current is diverted into the snubber capacitor, C1.  The inductance of
   the flyback primary (T2) and C1 forms a resonant circuit so that the voltage
   climbs on C1 as the current goes down.  At its peak, this voltage will be
   1000 V to 1500 V.

2. C1 now begins to discharge in reverse through the primary of T2 (back into
   the B+ supply - the filter capacitor will stabilize the B+ output) until
   its voltage (also C-E of the HOT) reaches 0.

3. If there were no damper diode (D1), this voltage would go negative and
   continue to oscillate as a damped sinusoid due to the resonant circuit
   formed by T2 and C1 (and the other components).  However, D1 turns on as
   the voltage goes negative and diverts the current through it clamping the
   voltage near 0 (-Vf for the diode).

   Note that the damper diode D1 may have been built into the HOT T2 in the
   case of an inexpensive or small screen TV that does not have any circuitry
   for E/W correction.

Steps (1) to (3) have accomplished the flyback function of quickly and cleanly
reversing the current in T2 (and, as we will see, the deflection yoke as well).
The full flyback (and yoke current) are now flowing through the forward biased
damper diode, D1.

4. At the beginning of scan, the damper diode (forward biased) carries the
   bulk of the current from the yoke and flyback.  The nearly constant
   voltage of the B+ across T2 results in a linear ramp of current now
   through the damper diode since it is still negative and decreasing
   in magnitude.

5. At approximately mid-scan, the current passes through zero and changes
   polarity from minus to plus.  As it does so, the damper diode cuts off
   and the HOT picks up the current (with a voltage drop of +VCEsat).  Current
   is now flowing out of the B+ supply.

   The base-drive to the HOT must have been switched on before this point!
   Timing is not very critical as long as it happens between the end of the
   flyback and the zero crossing of the summed current.  The location of the
   zero crossing depends on the secondary load, notably the beam current.
   Larger beam current requires that the HOT be switched on earlier. The
   designer has to do some optimizing here...

6. During the second half of the scan, the HOT current ramps up approximately
   linearly.  This is again due to the nearly constant voltage of B+ across
   the inductance of the flyback primary.

7. Near the end of scan, the HOT turns off and the cycle repeats.

   The HOT has a storage time between 3 us and 7 us, thus the base-drive is
   switched off earlier, in a controlled way to properly remove the charge
   carriers from the collector region in the HOT. The peak amplitude of the
   base current and the way it is decreased determine the ultimate dissipation
   in the HOT and are thus subject of heavy optimization. This is hampered
   by the fact that there is much spread in HOT parameters. 

Thus, the current in the flyback (ignoring the yoke components) is a nearly
perfect sawtooth.  The ramp portion is quite linear due to the essentially
constant B+ across the flyback primary inductance.  The current waveform can
be easily viewed on an oscilloscope with a high frequency current probe.  See
the section: "Probing TV and monitor yoke signals".

The voltage across the C-E of the HOT is a half sinusoid pulse during the
flyback (scan retrace) period and close to zero at all other times (-Vf of
the damper diode during the first half of scan; +VCEsat for the HOT during
second half of scan).

Caution: without a proper high frequency high voltage probe, it is not possible
or safe to observe this point on an oscilloscope with full B+.  However, where
the equipment can be run on a Variac, this clean pulse waveform can be observed
at very reduced B+.  Excessive ringing or other corruption would indicate a
problem in the flyback, yoke, or elsewhere.

Normally you would use a 100:1 probe suitable for 2 kV peak.  You could always
make your own voltage divider out of a couple of suitable resistors and use a
regular 10:1 or 1:1 probe.  Beware that also the capacitive division ratio
must be correct because the line frequency is high enough to make it relevant.
The current through Q1+D1 is several amperes peak-peak. There's a lot of
power circulating here, making this a *dangerous* circuit in every way!

  3.3) The deflection yoke connection

So, you ask: "Why can't the yoke just be placed in series or parallel with the
flyback primary?"

There are several reasons including:

* The desired yoke current is not quite a sawtooth but includes two major
  corrections: S and E/W (described below).  These cannot be applied easily
  with such a configuration.

* The flyback also generates the HV and secondary output voltages and the
  primary current might then be affected by these and change as a function
  of beam current (picture brightness) or audio level (although feeding the
  audio amplifiers from LOT windings is not common anymore).

Note that some TVs and monitors cut off power to the horizontal deflection
circuits if the yoke connector is removed.  This is a separate interlock
and not a result of the B+ flowing through the yoke.  Its purpose is to
protect the circuitry and the CRT.  With no deflection, the very bright spot
in the center of the screen would quickly turn into a very dark permanent
unsightly blemish.  With appropriate precautions to avoid this costly
situation, it is possible to power a monitor or TV with the yoke winding(s)
disconnected to determine if a defective yoke is messing up the deflection
system operation.  See the documents: "Notes on the Troubleshooting and Repair of Computer and Video Monitors" (or "...Television Sets") for
additional information.

The yoke is placed across the C-E of the HOT in series with a capacitor
(S-correction) and other components which in effect form a variable power
supply (analogous to the constant B+) which is used to compensate for the
various problems of scanning a nearly flat screen.

  3.4) What are S and E/W or N/S correction?

These terms actually refer to the various corrections to deal with what is
normally called scan linearity and pincushion distortion.  Most larger TVs
and nearly all high quality monitors will have various user and internal
controls to optimize the corrections for each scan rate (multiscan monitors).

Because the screen of most CRTs is relatively flat (even those not advertized
as flat) and the electron gun is relatively close, any picture tube will
naturally have sereous linearity problems and pincushion distortion if there
were no corrections applied.  Near the *edges* and *corners* of the screen the
spot will move faster because the same angular speed translates to a larger
linear speed.  This is simple trigonometry.

  3.5) S-correction circuit operation

The first correction to apply, in both directions, is S-correction.  By simply
putting a capacitor in series with each coil, the sawtooth waveform is modified
into a slightly sine-wave shape (the top and bottom are somewhat squashed).
This reduces the scanning speed near the edges. Linearity over the two main
axis should now be good.

When we add in the yoke components (only the horizontal deflection coil and
S-correction capacitor or S-cap are actually shown above) conditions are only
slightly more complex:

First, consider what would happen if instead of the S-cap, the yoke were
connected to B+ like the flyback.  In this case, the total current would
divide between the flyback primary and the yoke.  It would still be a sawtooth
as described above.  Of course, component values would need to be changed
to provide the proper resonant circuit behavior.

That's called 'tuning of the flyback capacitor', to achieve the proper
duration of the flyback pulse, matching the blanking time of the video
signal, and to achieve the proper peak flyback voltage, matching the
Vces specification of the HOT with a reserve of about 20%.
That's two conditions, requiring two degrees of design freedom.
There are 3 freedoms: supply voltage, flyback capacitor and yoke

With the S-cap and yoke wired as shown above, the inductance of the yoke and
S-cap form a low pass filter such that voltage on the S-cap will be a smoothed
version of the pulses on the HOT collector (similar in effect to the B+ feeding
the flyback but not a constant value).  The average value of the S-cap voltage
will be positive.

The S-capacitor together with the yoke inductance forms a resonant circuit
whose frequency is tuned lower than the line frequency. It has the effect
of modifying the sawtooth current into a sine-wave shape. This is called
'S-correction'. It reduces the scanning speed at the left and right edges
of the screen.

The value of the S-cap can be selected so that the voltage varies in such a
way as to squash the current sawtooth by the appropriate amount to largely
compensate for the fact that the electron beam scans a greater distance with
respect to deflection angle near the edges of the screen.

Think of it this way: When the scan begins, the yoke current is at the maximum
value in the direction to charge the S-cap.  The voltage across the S-cap
is causing the current to decrease but the S-cap is also gaining charge so
the rate of decrease is increasing.  At the time the current passes through
0, the S-cap is charged to its maximum.  The current now reverses direction
retracing its steps.  Got that! :-).  This is another example of a portion
of a resonant circuit.  The voltage on the S-cap is varying by just the right
amount to compensate for the geometry error.

For multiscan monitors, S-caps must be selected for each scan range since the
timing varies with scan rate.  These are only approximate corrections but
good enough for most purposes.  MOSFET or relay circuits take care that for
each range of scanning frequencies the correct combination of S-correction
capacitors is selected.

As an example, consider a multiscan monitor which supports VGA (31.4 KHz,
800x600 at 56 Kz (35 KHz), and 800x600 at 60 Hz (38 KHz):

For good geometry between 31, 35 and 38 kHz, two discrete values for the S-cap
is barely enough. Actually a 3rd value optimized for 35 kHz would be better. If
there is only one S-cap and it is optimized for 38 kHz, then at 31 kHz you will
be using a too large angle of the sine (of the resonant frequency between the
S-cap C and the deflection coil inductance). Possibly even > 180 degrees,
making the current fold back. Apart from an obvious geometric distortion,
there is also increased risk of HOT failure because you're operating too close
to the resonant frequency.

  3.6) N/S and E/W correction circuit operation

Then remain the N/S and E/W errors, meaning that near the corners the scanning
speed is still too large.

To a large extent the N/S errors can be corrected by a suitable coil design.
For smaller tubes (90 to 100 degrees types) this is also possible for E/W
errors. For larger tubes (110 degrees) or high quality tubes, electronic E/W
correction is required. This is the well-known pin-cushion.
E/W correction is modulation (which implies multiplication) as a function
of vertical beam position. The amplitude of the horizontal deflection current
is modulated with a parabola waveform which is derived from the vertical
deflection circuit. This squeezes the top and bottom lines back into the left
and right screen borders.
N/S correction (if any) is a method of injection - addition of a high
frequency waveform (harmonics of the line frequency) to the low-frequency
field-deflection waveform.

This requires costly nasty circuitry which is better avoided.

This is how the diode modulator for E/W correction works:

            | (Vb)
 Part of T2  )|:|
    Flyback  )|:|
    Primary  )|:|
            |                     <- Yoke components ->
            |       |         |         |          | (Vb)
            |       |         |         |          +
            |       |         |         |          )|:|
            |     __|__      _|_        |          )|:| L2
            |     _/_\_      ---        |          )|:| Horizontal yoke
            |       |   D1    | C1      |          +
            |       | Damper  | Snubber |         _|_ C2
            |       |         |         |         --- S-correction
            |       |         |         |          |
            |       +---------+---------)----------+ (Vm)
            |       |         |         |(cross)   |
            |       |         |         |          +
            | C     |         |         |       L3 )|:|   L4
        B |/      __|__      _|_       _|_  Bridge )|:|   E/W coil
     -----|       _/_\_      ---       ---    coil )|:|  ----    
          |\        |   D2    | C3      | C4       +     ----         (Vm) E
       Q1   | E     | Damper  | Snubber | Snubber  +----+^^^^+---+-----+-----
            |       |         |         |          |             |     |
       HOT  |       |         |         |      C5 _|_           _|_   E \| B
            |       |         |         |  Bridge ---     Elcap ---  Q2  |---
            |       |         |         |     cap  |        cap  |    C /|
            |       |         |         |          |             |     |
           _|_                                            E/W amplifier (PNP)
           B+ Return (may not be signal ground)

The deflection supply B+ gives a constant voltage Vb. At the output of
the E/W amplifier there is a variable voltage Vm. Because there can not
be an average voltage over any coil, the average voltage over the
deflection circuit (L2 + C2) is Vb-Vm. The scanning width is proportional
with this voltage Vb-Vm. By modulating Vm as a field-frequency parabola
which is higher for the top and bottom lines it is achieved that the
scanning width is reduced for the corners of the screen.

The required field-parabola waveform is derived from the field deflection
circuit. Amplitude and DC-level are adjustable, for correcting pin-cushion
distortion and setting the screen width. EHT is *not* influenced !

The E/W amplifier usually has a PNP emitter follower only, because it
must only sink current and dissipate a bit of power. The coils L3 and L4
take care that the E/W amplifier sees no line-frequency.
The "bridge" components L3 and C5 resemble the deflection coil with its
S-correction capacitor and carry the large amplitude alternating current.
L3+C5 are tuned to approximately the same frequency as L2+C2.

C1, C3 and C4 must be tuned so that EHT is independent of Vm and peak
voltage over D2+C2 is high enough but not too high when Q2 is off.
C4 is usually a small ceramic capacitor of approx. 1 nF mounted close
to the HOT Q1, it also suppresses EMI. Flyback capacitors are critical
components. Wrong types may overheat and burn. Bad contacts here or
elsewhere in the deflection circuit may arc and also cause fire.

There are many variants to this circuit, e.g. for dynamic S-correction.
Multi-sync monitors need added circuitry to make the EHT independent of
the line frequency (if there is not a separate EHT supply, that is).

If the E/W modulator fails then you will see that the top and bottom lines
will be much too wide. There are several parts that could have failed. It's
usually not too difficult to find why there's no parabola.  If you have
partial loss of E/W modulation, notably in the extreme corners, then you
should suspect the tuning of the 3 flyback capacitors that belong to the
diode modulator circuit. That's a specialist job...

  3.7) S-correction problems

* An open S-cap will result in no horizontal deflection - a vertical line.

* A shorted S-cap will likely load down the B+ possibly resulting in a blown
  fuse or other power supply components.

* An S-cap that changed value (or in the case of a multiscan monitor, selected
  to be the wrong value) will result in distortion at the left and right sides
  of the screen:

  - Too low: picture will be squashed towards edges.

  - Too high: picture will be stretched towards edges.

  Note that this is not the same as what is commonly called linearity which
  would likely affect only one side or gradually change across the screen.

  3.8) Horizontal linearity correction

Since there is non-zero resistance associated with the components (mainly
coil losses) in the yoke circuit (yoke winding, ESR of S-cap, etc.) the world
is not quite as ideal as one would hope.  Without compensation, this resistance
would result in non-linearity of the picture - it would tend to be squashed on
the  right side as the resistance saps energy from the yoke circuit.

The waveform becomes a damped sinewave, which will be 'undamped' by restoring
energy during the flyback.
One way to deal with this is to add a magnetically biased saturable inductor in
series with the horizontal deflection yoke.  This is called the linearity coil.
Its core is magnetically biased near the point of saturation such that the
inductance decreases with increasing current and this helps to stretch the
right hand side of the scan.  In other words, during the scan the coil
saturates so that the inductance decreases.  At the end of scan there is
practically no voltage left over the linearity coil so that the deflection
coil gets maximum voltage.

  3.9) E/W correction problems

The common name for the adjustments is likely to be 'Pincushion Amp' and
'Pincushion Phase'.  They are controlling the E/W correction circuits.

Pincushion Amp adjusts the amplitude of the correction signal.

Pincushion Phase adjusts where the correction is applied on the vertical scan.

* Failure of the E/W correction circuit will result in very noticeable
  pincushioning distortion of the vertical edges.

* Excessive E/W correction will result in barrel distortion of the vertical

* A bad power supply derived from the flyback could also result in similar
  symptoms due to ripple or lack of power to the pincushoin circuitry.

  3.10) Differences between N/S and E/W correction implementation

While the desired effects are symmetric - modulate the amplitude of one
component of the deflection circuit (H or V) by the other (V or H), the
implementations will differ substantially.  The reasons should be obvious:
The line frequency is much higher than the field frequency.

E/W correction is easy: the lower frequency modulates the higher frequency.
This reduces to simple amplitude modulation.  Well, simple in principle. The
line circuit is a high-energy circuit.  For this purpose the diode modulator
circuit has been invented.  It allows an energy exchange between the line
deflection  circuit and a pseudo deflection circuit.

N/S correction is difficult: the higher frequency modulates the lower
frequency. It *can* be done with sort-of amplitude modulation by using a
'transductor'. This is not a transformer but a component with 2 coils and a
saturable core where the (line frequency) current through 1 coil modulates
the inductance of the other coil. If there are tuned parts in the circuit
then the correction will be highly sensitive to line frequency variations.

It can also be done with a regular transformer, by injecting a strong signal
(from an amplifier) with line frequency components into the field deflection

Either way, it's an expensive solution which should be
avoided by designing the deflection coils in such a way
that the picture tube needs no active N/S correction.

  3.11) EHT (high voltage) generation

The EHT is generated from a flyback rectifier on a secondary winding
of the line transformer with many many turns of very thin wire.
Because the flyback pulse is so narrow, the rectifier diode will
conduct only a short time. Thus the peak current in the winding will
be quite high, resulting in a significant voltage drop when loaded.
The internal impedance of the EHT source is in the order of 1 MOhm,
so with a load of e.g. 1 mA the EHT will drop 1000 V = -3%.
Usually the EHT voltage is far from stable, 10% drop is quite normal.

If the EHT voltage drops, then the electrons will be accellerated less
and will move through the deflection field at a lower velocity. As a
result they will be easier to deflect by the magnetic field, and the
picture size will grow. Without special measures, brighter pictures
will be larger. The measure is to feed some EHT information or beam
current information to the deflection circuits, reducing the deflection
current amplitude a bit for bright pictures. For horizontal deflection
this is done by the E/W modulator. This is called anti-breathing.

Sets with raster correction free picture tubes don't have an E/W 
modulator. There the correction may be done by means of a power 
resistor in series with the B+ supply. A large beam current causes 
more power consumption, this lowers the B+ supply voltage and thus 
reduces the line deflection current. That also reduces the EHT even 
further, but the deflection current has a stronger effect on the 
picture width than the EHT. Better methods exist too.

The EHT information is also used to protect the flyback transformer
from overload. As the load increases, the average primary current
rises. Ultimately it may reach a level where the transformer core
may go into saturation. This causes large peak currents in the HOT
which might lead to destruction. To prevent this, some EHT information
is fed to the contrast controller, to automatically reduce the
picture brightness whenever the white content is too much. This is
called the average beam current limiter. 

A failure in the video path, like a video output amplifier stuck at
0 V, causes a high beam current that will not react to the contrast
controller. In that case the beam current limiter will not work and
the set should switch off automatically, usually within a few seconds
after applying power. When the cathodes heat up, you'll see an even
picture with diagonal retrace lines and then it will switch off.

  3.12) The difference between the ideal and the real

Don't expect to find the circuits shown above staring you in the face when you
get your Sams' Photofact or service manual.  There are a semi-infinite number
of variations on this basic theme.  Some of them will, to put it mildly, appear
quite obscure (or to put it more positively, creative) at first.

You may see all sorts of additional passive components as well as transformers
for generating additional voltages not provided by the flyback.  There may be
diodes in places you would think would be impossible.  Therefore, to really
understand even approximately how each design works may require some head
scratching but the basic operation of them all seems to be very similar.

Chapter 4) Horizontal Output Transistors

  4.1) Storage time of HOTs

Storage time differs between transistor types, there's parameter spread 
within a type (lower hFE gain gives shorter storage time and v.v.) and 
it depends on load (higher beam current or larger deflection amplitude 
due to E/W modulation gives shorter storage time).

Variations in storage time would translate into horizontal position
errors of the picture. That's why the base-drive to the HOT is 
generated by a PLL that measures the phase of the output of the HOT,
which is the flyback pulse at the collector. This PLL is called the 
PHI-2. It keeps a constant phase relation between the sync at the input 
and the flyback pulse. As a result, you will see the base-drive pulse
shifting in time as a function of HOT load, this is normal.

Most deflection processors generate a base-drive pulse with a constant 
duty-cycle. This means that also the switch-on moment of the HOT will 
vary with the load. This makes it extra difficult to optimize the 
base-drive because there is only a limited time interval where the HOT 
may be switched on and that interval is shorter with high beam current 
load. On-time is typically between 50% and 55%, depending on the IC.

The feedback of the flyback pulse to the PHI-2 PLL is not perfect
because the shape of the pulse distorts as a function of beam current.
This will give a dynamic geometry error. It is compensated by feeding
a certain amount of EHT information to the PLL.

  4.2) Why are nearly all horizontal drivers circuits transformer coupled?

One (probably secondary) reason is that this provides one of the
isolation barriers between a line-connected HOT and flyback primary
and the signal circuits of the TV.

A more important rational is that a transformer is nice easy way of
impedance matching the horizontal driver circuit (100s to 1000s of ohms)
to the few ohm input impedance of the horizontal output transistor
base which requires upwards of several amps for proper drive.  A typical
driver transformer may be in the 5-10:1 turns ratio representing 25-100:1
impedance ratio.

A byproduct of all this is that it is unlikely for a faulty driver stage
to kill the HOT.  Unlikely but not impossible.

  4.3) Additional information on horizontal driver circuits

Usually, the primary voltage is *constant* when the driver transistor is ON
and thus the HOT is OFF. Then when the driver switches OFF, the stored magnetic
energy switches the HOT to ON. 

This is called non-simultaneous base drive, which is most common.  The primary
voltage that you see then is mostly a transformed version of the secondary
voltage, over the series base impedance.  The voltage at HOT=ON is not forced
from the primary side.

Usually the "cold" side of the primary of the driver transformer is connected
via a power resistor and filtering electrolytic capacitor to the B+.  The R
determines the average voltage and thus the base drive current. Higher R means
less base drive and vice-versa.  Varying the value of this resistor is the
first choice for minimizing the power dissipated in the HOT.

The filtered DC voltage at the cold side is (typically) a little over
1/2 * V(B+). Then when the driver switches off the voltage at the hot side
(of the primary) will be higher, but typically not higher than V(B+). But how
high exactly is a coincidence.

  4.4) Horizontal output transistors keep blowing (or excessively hot)

Unfortunately, these sorts of problems are often difficult to definitively
diagnose and repair and will often involve expensive component swapping.

You have just replaced an obviously blown (shorted) horizontal output
transistor (HOT) and an hour (or a minute) later the same symptoms
appear.  Or, you notice that the new HOT is hotter than expected:

Would the next logical step be a new flyback (LOPT)?  Not necessarily.

If the set performed normally until it died, there are other possible
causes.  However, it could be the flyback failing under load or when it
warms up.  I would expect some warning though  - like the picture shrinks
for a few seconds before the poof.

Other possible causes:

1. Improper drive to horizontal output transistor (HOT). 

   * A too weak drive (or a HOT with a too low Hfe) causes Vce(sat) to be too
     large, giving conduction losses.

   * A too strong drive (or a HOT with a too high Hfe) causes it to switch off
     too slowly, giving switching losses.

   Base drive should be optimized to balance between these 2 losses.  Check
   driver and HOT base circuit components.  Dried up capacitors, open
   resistors or chokes, bad connections, or a driver transformer with shorted
   windings can all affect drive waveforms.

2. Excessive voltage on HOT collector - check low voltage regulator (and line
   voltage if this is a field repair), if any.

3. Defective safety/flyback capacitors or damper diode around HOT.  (Though
   this usually results in instant destruction with little heating).

4. New transistor not mounted properly to heat sink - probably needs mica
   washer and heat sink compound.

5. Replacement transistor not correct or inferior cross reference.
   Sometimes, the horizontal deflection is designed based on the quirks
   of a particular transistor.  Substitutes may not work reliably.

   Well, you can always *try* to optimize the base drive by changing the value
   of the power resistor that feeds the drive circuit at the primary of the
   drive transformer.  But you're on your own here!  Clearly label what you
   have done or else your name will be mud if the unit ever needs to be
   repaired by someone else in the future.

The HOT should not run hot if properly mounted to the heat sink (using
heatsink compound).  It should not be too hot to touch (CAREFUL - don't
touch with power on - it is at over a hundred volts with nasty multihundred
volt spikes and line connected - discharge power supply filter caps first
after unplugging).  If it is scorching hot after a few minutes, then you
need to check the other possibilities.

It is also possible that a defective flyback - perhaps one shorted turn - would
not cause an immediate failure and only affect the picture slightly.  This
would be unusual, however.  See the document: "Testing of Flyback (LOPT) Transformers".

Note that running the set with a series light bulb may allow the HOT
to survive long enough for you to gather some of the information needed
to identify the bad component.

  4.5) Why are there so many different HOTs?

I find it fascinating (ok, well at least interesting) that after 20 years
of designing totally solid state TVs and monitors, HOTs have not become
jelly bean parts.  Why does every new design insist on a unique HOT?  I
don't believe this is simply due to increased requirements like wider
deflection angles or (for computer monitors) higher scan rates.

Actually there is still some progress going on :-).

TV's for Europe DO get higher scan rates, you know.  Our entire high-end
range runs on 31250 Hz. Sets with VGA capability often run even higher.
100 Hz HDTV was supposed to run on 62500 Hz but that is a big technological
problem as you might imagine.

Larger screen sizes (32" 16:9, 33" 4:3) do tend to have less sensitive
deflection coils, so the peak-peak amps go up. Combined with the
higher scan rate this often means that some new transistor must be
found, even if it is only a higher rated selection from an existing 
type. There is also 1700 V Vce types next to the regular 1500 V.

And in USA it is known that setmakers (like ourselves) have
standard transistors marked with a different type number, to prevent
repairmen from putting in just any would-be replacement type.
For a fact, it IS risky to put in the wrong replacement, the faulty
drive conditions may destroy it early. So try and avoid this.

To comfort you, the BU508 is still widely used!

  4.6) Varieties of BU508 HOTs

(From: J. G. Simpson (

BU508 series seem to come in a number of variants, I haven't sorted out the
specification for each listed variant, but have foundthat it's worth trying
replacements by different manufacturers.  Also if the device is on a grounded
piece of metal as heatsink try adding another radiator (twisted vane is my
preference).  Some TV manufacturers introduce post production mods to change
resistance values to provide more drive into the base, this may be in the base
or emitter.  Scope the waveform for parasitic high frequency oscillation and
check that the waveform looks clean. Check voltage rails, supply and derived,
and that the set is not over scanning. Check all the components around the
horizontal output stage, it may be the manufacturer. had a duff batch of some 
component (often a capacitor that goes OC) that keeps failing
which then stresses the BU508.

  4.7) Is there a standard HOT transistor for monitors?

"Anyone like to comment on BU508's, they should be to the same spec. and born
 equal.  My experience has been that different manufacturers BU508's behave
 differently.  One make will fry and last about 3 weeks, put in a different
 make, no circuit change, and it runs cool, and is still running 3 years later.
 Price doesn't seem to be a guide, a $1 one may run cool and have no mfr. code,
 a branded one cost more and run hot."

Yes, here you have the problem exactly!

There is such a thing as component spread. Base drive must be optimized for
the whole range of gain within a type.  That range can be so large that at the
limits of the spread the dissipation can still be too large. The reason is
that the device with the largest gain will be overdriven, causing a tail in
the current at switch-off, whereas the device with the smallest gain will be
underdriven, causing it not too saturate enough. Each condition can be easily
viewed on the 'scope. By varying the base drive you can minimize dissipation.

Normally one basedrive is set for the entire population, accepting the
variation in dissipation and its upper limit.  Sometimes the variation is so
large that this will not be acceptable. But this is unlikely for a BU508 in a
16 kHz application. Substituting it with a similar type from a different brand
with different parameter spread may indeed cause it to dissipate too much and
thus fail early. This has nothing to do with price or quality, just with a
different optimum base drive. If base drive has been optimized for a brand
with low parameter spread then it can be that the heatsink may have correctly
been selected smaller...

Since you live in the UK, you should definitely read David Sharples' article
in Electronics World (was it June 1996 ?).  He literally optimizes base drive
for a living.

  4.8) Why do apparently similar or better HOTs sometimes run hot and blow?

It is often surprising that replacing a horizontal output transistor
with one that has overall better specifications does not work out - it
may run hot and fail.

There is more to characterizing a transistor than just maximum voltage,
current, and power dissipation.

One important parameter is current gain:  Too low a gain for a particular
operating point may result in incomplete turn-on during scan resulting in high
dissipation.  You want the transistor to be in the fully saturated state.  A
larger HOT is more likely to have a lower current gain.

If you read the app notes put out by the manufacturers like Motorla you
will also find that fast turn off based drive (negative step) is actually
not what you want since this traps excess carriers in the high resistivity
collector region which leads to continued conduction and heating.  The ideal
waveform also provide adequate drive during scan but not excessive overdrive
and is thus an increasing ramp to account for the increasing collector current
during scan.

Characteristics like this are not dealt with by the basic specs but can
differ substantially among otherwise similar transistors.

Also see the section: "Horizontal output transistor specs and substitution".

  4.9) Horizontal output transistors blowing at random intervals

The HOT may last a few months or years but then blow again.

These are among the hardest problems to locate. It could even be some peculiar
combination of user cockpit error - customer abuse - that you will never
identify.  Yes, this should not happen with a properly designed monitor.
However, a combination of mode switching, loss of sync during bootup, running
on the edge of acceptable scan rates, and frequent power cycles, could test
the monitor in ways never dreamed of by the designers.  It may take only one
scan line that is too long to blow the HOT.

  4.10) Horizontal output transistor specs and substitution

Refer the article on deflection circuit design by David Sharples
(Electronics World & Wireless World, June 1996).

Every line transistor has its own requirements for:

* Amount of base drive current, especially the Ib at end-of-scan.
* Waveform of base drive current (rising, steady, falling)
* Speed of reduction base drive current at switch-off.

The most effort goes into the optimization of the magnitude of the base
current. The problem is: gain spread. There used to be other spread factors
influencing the dynamic transistor parameters but these have been mostly
eliminated by better process control at P.S.. You have to find *one* optimum
drive so that neither the high-gain nor the low-gain type will dissipate too

Overdriving causes a slow switch-off behaviour, some collector current keeps
flowing during the beginning of the flyback and will cause dissipation.

Underdriving causes bad saturation, the collector voltage will start to rise
before the flyback should start. This too causes dissipation.  Either condition
is easily observed with an oscilloscope, a current probe and a 1:100 voltage
probe (be sure to calibrate it for HF !).

The dissipation as a function of the base drive current is a more-or-less
parabolic function with a global minimum. The minimum will be different for
high-gain and low-gain types. By measuring the curves for both extreme types
and combining them, an optimum drive for the random type will be found, with
a figure for the worst-case dissipation.

All this will only be true if you insert a device which is a member of the
population spread for which you optimized the base drive.  If you just insert
a random other device (different type, same type but different brand, same
type and brand but much older/newer batch) then all bets are off. Dissipation
may be way too high, with early failure as a result (and possibly a distorted
picture geometry due to excess damping of the waveform).

It is certainly not possible to substitute a standard HOT (BU508) in place of
a more advanced type (in >> 15 kHz applications like monitor).  It is also a
very bad idea to substitute a BU508 in place of a much lighter type like a
BUT11 (used in <= 17" sets). It will fail!

With horizontal output yransistors, it is *not* true that 'bigger is better'.
If you substitute a heavier transistor (more amps, more volts, more watts,
faster switching, whatever) for a lighter one, then there is a very big chance
that it will fail earlier, not later.  The reason is that the drive conditions
will now be wrong (most likely underdrive) and the transistor will overheat
from too high conduction losses (Ic * Vce,sat).  So do yourselves a favour and
get a correct replacement type.

  4.11) What is this diode across my HOT?

It is called a damper diode and is essential to proper operation of the TV's
or monitor's horizontal deflection as well as to the continued life and
happiness of the HOT.  Using an HOT with an internal damper is ok even if
there is a separate one in the circuit.  The other way around (leaving it out
entirely) will likely result in instant - i.e., single scan - destruction of
the HOT.  This is because in modern deflection system designs, the damper
carries the horizontal yoke current for a significant portion of the scan.
If it is not present, the HOT will be forced to try to eat this current - in
reverse - across C-E.

The damper is a special high voltage fast recovery type of diode - a 1N400x
type will not work in its place.

BTW, many of these HOTs have a D after the part number to indicate that
they have the internal damper and include a B-E resistor (which may confuse
transistor testing) of about 50 ohms.  However, the D is not a sure indication
of an internal damper - nor is its absense an indication of a lack thereof.
The entire part number must be checked to be wure.

  4.12) What is this funny capacitor (or capacitors) across my HOT?

These may go by the name flyback, high voltage, snubber, or deflection
capacitors.  When the HOT is shut off, the current flowing in the inductance
of the flyback primary and horizontal deflection yoke cannot be stopped
instantly.  These capacitors provide a place for this current to go and
is part of a tuned circuit (in combination with the flyback and yoke)
which needed to accomplish the flyback function.

If this capacitor is open or missing, excessive flyback voltage will result
probably killing the HOT.  If the HOT does not fail, the result will likely be
greatly increased high voltage.  Should the X-ray protection circuitry not
shut down the deflection, there could be internal or external arcing and/or
destruction of components like the flyback or tripler.

For proper operation and continued safety, only proper exact replacements
should be used for these parts.

  4.13) Brief comments on testing the HOT

For a TV with no blown fuses that will not start, here are two quicky
checks to see if the HOT is good and has power and drive:

* HOT tests - check across each pair of pins for shorts (preferably removed
  from the circuit board).  No junction should measure less than 50 ohms
  or so.  Lower readings almost certainly indicate a bad HOT.  If in-circuit,
  however, the reading between base and emitter will be near zero due to the
  secondary of the driver transformer.  See the document: "Testing Diodes and
  Bipolar Transistors with a DMM or VOM".  Don't be confused by internal
  damper diodes and B-E resistors.

* Power - measure across the collector to emitter with a multimeter (with
  the HOT removed or if there is no deflection, this is safe with it in
  place).  There should be solid B+ - typically about 100 to 160 V (115
  VAC sets - possibly higher for 220 VAC sets).  If this is missing, iether
  there is a problem with the power supply or the emitter fusable resistor
  has blown (probably in addition to the HOT) and there is no return.

* Drive: put an oscilloscope on the base - there should be pulses around
  .7 V for most of the scan (~50 microseconds) and probably going negative
  a couple volts at least for retrace (~12 microseconds).  If drive is weak
  or missing, determine how startup is implemented as there may be a problem
  in the startup power supply or deflection IC.

WARNING: use an isolation transformer for the oscilloscope tests (and whenever
you are probing a TV in general)!!!  This part of the circuit, in particular,
is usually line connected.  See the document: "Safety Guidelines for High Voltage and/or Line Powered Equipment".

  4.14) Saga on swapping of horizontal output transistors

(From: Malik (

The prologue:

I have in for repair a Tatung monitor model CM14UAE. Originally it was 
dead this fault was quickly remedied by replacing the line output 
transistor which was 'dead short', causing the power supply to shut down.

Now the transistor I replaced seems to get a hell of a lot hotter than it 
should, when it gets to a certain temperature after about half an hour the 
width begins to collapse at which point I quickly switched off. If I switched
off for a few seconds and back on again I would get the width back but again 
it quickly began to collapse.

Things you need to know:

The LOP stage is independent from the EHT transformer. i.e. there is a 
separate transformer for the EHT (and A1 supplies etc) with its own 
transistor and a separate line output only transformer with its own 

I have also tried substituting the efficiency diode, still no joy. 
I have also replaced the EW transistor and its driver.

To me it seems that the transistor is not being switched properly, 
although the waveforms do look ok.

Original line output device 2SC3893A, which I replaced with a BU508DF 
(European equivalent according to the books.  I also tried a 2SC3883 which
used  for the same job in other SVGA monitors. Both give the same problem,

I will be trying to get the original transistor in case this is the 
problem but I don't think it will make a difference. 

And the conclusion:

Anyway after all that headache I finally received through the post the
original 2SC3893A. I fitted this and left the monitor on test. Sigh of
relief, the problem has gone away. 

It seems the timing on this device differs from all the others I
tried. The transistor runs at a reasonable temperature even after a few
hours use.

The moral of this story is, use original line output transistors.
Unless you are 100% sure the replacement works in its place. Don't rely
too much on equivalent books, these should be used as a guide. The
book I looked at specified the BU508DF as a direct replacement for a
2SC3893A, but as I know now it doesn't work properly.

The HOT should not run too hot to touch.  If the replacement sizzles, it
won't last long and is probably deficient in some specification.  If the
monitor or TV appears to work normally otherwise, try an exact replacement
HOT if available before swapping other expensive parts like the flyback
(LOPT).  For testing, however, a substitute can usually be used - with a
series light bulb and Variac.

  4.15) S-correction circuits on multi-scan monitors

(Quoted text from a curious but frustrated tinkerer/technician.)

"Hi there. Recently, I've encountered this circuit (or ones similar to it)
 several times in VGA monitors and I'd like to know what it's purpose is.

               |D                  |                 |
              _+--+                |                 |
           G||  __|__             _|_              __|__
     S0 ----||  _/_\_ IRF620      ___ .64 uF       _\_/_
            ||_   |                |                 |
               +--+                |                 |
               |S        .64 uF    |B                |
     Ground o--+-----------||------+---o * See Text  +------o To +76 V
               |S                A |C                |
              _+--+                |                 |
           G||  __|__             _|_              __|__
     S1 ----||  _\_/_ IRF620      ___ .64 uF       _/_\_
            ||_   |                |                 |
               +--+                |                 |
               |D                  |                 |

     The IRF620's are N channel Enhancement Mode MOSFETs. 
     * The connection at this point is to the HOT collector via the horizontal
       deflection yoke (HDY), width, and linearity coils."

You will probably find that 0, 1, or 2 of the MOSFETs will be turned on
(S0, S1 set to ground or +15 V (typical) depending on the scan rate.  This
would introduce varying amounts of S-correction (E/W) depending on the
horizontal scan rate.
The diodes are probably for clamping and protection.  The MOSFETs are used
as switches.  You, in effect, get 1, 2, or 3 x .64 uF between the yoke return
(the point marked '*') and ground.
This circuit is from an auto-scan computer monitor where it switches the
effective S-correction capacitor to one of 3 possible values depending on
scan rate.

Note that this approach is used both in multi-scan and auto-scan monitors.
The multi-scan (largely obsolete technology though still used in some specific
applications) has two or more discrete line frequencies and the auto-scan has
a whole range of available frequencies. You would not guess this from the
discrete switching of the S-correction capacitors. Obviously, the S-correction
is optimized only for 3 frequencies and approximated for all others. This is
generally good enough (even) for auto-scan.
The internal diodes anti-parallel to the MOSFET (A to S, K to D) are important
to set the DC voltage over the extra S-capacitors to a reasonable value before
they are first switched in. At lower line frequencies the extra capacitors are
permanently switched on.

For each given scan frequency there is only one correct value for the
S-correction capacitor. If you limit your circuit to only 3 discrete values
(like 0.64, 1.28 and 1.92 uF) then the horizontal geometry can be optimized
for only 3 scan frequencies.

Somewhere between these 3 frequencies there will be 2 switch-over points where
more S-correction capacitance is added by the switches.  Generally, around
these border frequencies the geometry will be worst (because whatever the
position of the switch, there will be an error one way or the other) and it
would be wise to switch over near some little-used frequencies and thus
optimize for the much-used frequencies like 31, 38, 48 and 64 kHz.  For the
other scan frequencies, the geometry will then be only approximately correct.
Luckily, S-correction is not very critical.

For any given scan frequency, the position of the MOSFET switches is constant:
either on or off. These are not used in a switched mode.

There *will* be one or two other MOSFET switches that switch at a line scan
rate in order to take care of the correct dependency of the (average) supply
voltage and E/W voltage in function of the scan frequency.

There is a difference in the horizontal deflection circuit of a TV versus
an auto-scan monitor. In the diode bridge of a TV, the deflection circuit
is in the upper half of the bridge and the "bridge circuit" is in the
lower half of the circuit. The E/W amplifier *sinks* current from the
centre node (and dissipates). The VB supply voltage is constant.
In the diode bridge of a auto-scan monitor, the deflection circuit is in
the lower half of the bridge. This makes the S-correction capacitor grounded
and thus easier variable (without need for level shifters, opto-couplers,
etc).  The E/W amplifier *sources* current into the centre node. The VB supply
voltage is made variable, linearly proportional to the line frequency, by 
means of a switched-mode down-converter topology. The same topology is also 
used for the E/W amplifier, which now actually delivers power to the circuit.
With this topology you can have a constant (even regulated) EHT and a
constant deflection current amplitude, independent of the line frequency.

"In another similar monitor, this circuit had completely self-destructed
 and taken out a few other things too (fire).  The monitor shuts down
 immediately unless the HOT is disabled or connections A, B, and C are *all*
 broken (no other combination works).  It almost seems like it's used to
 accommodate or change the horizontal frequency for different video modes.
 For some reason, the monitor is now powering up so seems like it might be
 an intermittent short somewhere."

By disconnecting A,B,C you effectively take the deflection coil (HDY) out
of the circuit. By disabling the line transistor (HOT) there remains only
DC voltage over the bridge- and S-correction capacitors. Either way there is
no voltage over the HDY. I would suspect a short or arcing in the deflection
coil. Just a wild guess, though.
There is enough energy present in the deflection circuit (also in the HDY)
to create a very nice fire, let there be no doubt about that. A loose contact
in series with this circuit is potentially fatal.

Typical HOTs as used in monitors

(From an engineer at Philps).

14", SVGA (38 kHz)         A-types: BU2508AF, 2SC4830,  2SC5148.
                           D-types: BU2508DF, 2SC4762,  2SC4916,  2SC5149,
                                    2SC4291,  2SC5250.

15", XVGA (64 kHz)         A-types: BU2520AF, BU2522AF, 2SC3885A, 2SC3886A,
                                    2SC4757,  2SC4758,  2SC5129,  2SC4438,
                                    2SC4770,  2SC4743,  2SC5067,  2SC5207,
                                    2SC5251,  2SC5002.
                           D-types: BU2520DF, 2SC3892A, 2SC3893A, 2SC4531,
                                    2SC4763,  2SC4124,  2SC4769,  2SC5296,
                                    2SC4742,  2SC4744,  2SC4927,  2SC5003.

(A=no damper, D=built in damper diode).

For designs with larger screen sizes (and higher frequencies) the device
selection is not so straightforward as some designs split the horizontal
deflection and EHT stages.
In nearly all the above cases, the devices will plug-in substitute for
each other within category.

Continuous dissipation is hardly ever the cause of failure (never for
Philips devices). Failure is usually due to some infrequent transient
condition. For multi-frequency monitor designs 1991-1994 mode-change 
was/is a big killer. When repairs are made it is wise to cycle through 
a mode-change sequence. Delays of about 1 min. between changes should 
be used, shorter delays can cook the device.

* As a rule, once an engineer has a bad experience with mode change he
  takes greater care in the small-signal circuitry of his next design.
  This has led to mode change, in general, becoming more benign in the
  last couple of years. However, in Taiwan & Korea there is a high turn
  round of engineering staff and some, shall we say, less than perfect
  designs do still reach production.

* To "cook" a device by mode changing would take at least 30 mins. of
  continuous changing with a delay of about 20 seconds between each change.

* If a device fails during such a sequence the old, but rather uncooth,
  spit test is good indicator of why a device fails. That is, a drop of
  spit on the HOT immediately after failure can tell us a lot: if it
  sizzles then the device has probably cooked, if it doesn't then the
  device failed instantly after one stressful cycle. Frontiers of
  technology it is isn't but it is a useful technique.

* If you do get fails which haven't been caused by the HOT "cooking" I
  have no easy solution, sorry !

* Winfield Hill's comments reflect the experiences of many. School physics
  teaches that bipolar's are current driven and MOSFET's are voltage
  driven. In practice, of course, this is a gross over-simplification.
  A MOSFET in deflection would, as we say in the north of England, have
  to be a "bloody big bugger". For example, a 1500V MOSFET that behaved
  as a BU2508 would have to be nearly twice the size and at least twice
  the price. Getting 10V on all the gate cells of such a big chip requires
  a lot of charge to be injected (ie current) and then removed (reverse
  current); much like a bipolar drive.

  4.16) Why MOSFETs are not generally used for HOTs

One would think that with a MOSFET's high impedance voltage drive and other
desirable characteristics, they would have displaced bipolar transistors for
horizontal deflection circuits.  Why not?

True, a MOSFET is much much much easier to drive. In modern switched-mode power
supplies they reign, rightfully so. They are effective and rugged.

*But* in line deflection things are not so easy. I'll give you 4 reasons:

1. The flyback pulse is regularly 1200 V peak, allowing for margins you need
   a 1500 V device. There are few or no FETs available for that.  In a small
   set, a 1000 or 1200 V device may be usable.   With effort, the deflection
   circuit impedance may be re-scaled for greater current and lower voltage.

2. The conduction losses (Vds = Id * Rds) of a FET are quite high, this
   not only is wasteful dissipation but it also affects the linearity
   of the picture (it's a damped sine-wave, going slower and slower).

3. For the same power losses a FET needs a much bigger silicon area,
   also FETs are made in a more advanced process, with IC-elike features,
   this translates directly into greater cost!  They are several times more

4. A FET has no storage time, hence it has no storage-time modulation,
   which is a disadvantage because that would help to improve the natural
   stability of the control of the phase of the line deflection.

And these are just some major reasons, there are minors too.  We have looked
into it time and again and still the bipolar wins!

  4.17) Universal HOT replacement for TVs?

Warning: as noted elsewhere in this document, the following approach is much
less likely to work with long term (or even more than a few millisecond)
reliability in high performance computer monitors.

(From: Chris Jardine (

I shouldn't say this, but, the TV repair shop I worked for a number of
years ago stocked 1 universal Horizontal Output Transistor 2SC1308K or
NTE238. This worked in almost every set out there that used a
transistor and not SCR's (RCA, etc.). This may not be the best way of
substituting, but, these 2 part numbers seem to have fairly high gain,
power capability, voltage ratings, current ratings, etc. These
characteristics made it a good substitute and when you buy 100 at a
time you would get a really good price.

This may not work in your case, but, my $.02

  4.18) Optimizing base drive for HOTs

It is interesting that the problem of base drive optimization receives a fair
bit of attention, either for allowing a (non-compatible) replacement HOT to
survive or for rebuilding a fixed frequency monitor to a different scan
frequency. I wish there was an easy way to teach the average hobbyist a method
to do this himself.

The recipe itself isn't all to difficult: mostly we change the value of the
power resistor that feeds the line drive circuit until the temperature of the
HOT heatsink is at minimum. Too much basedrive current increases the switching
losses, too little basedrive current increases the conduction losses. The
optimum is somewhere in the middle. All you need is a handful of power
resistors and a thermometer on the heatsink.

If you need to optimize for a general HOT type (as opposed to one single
sample) then you kindly ask the manufacturer to provide some limit case
samples (lowest and highest hFE found) and find a basedrive that satisfies
both extremes. Of course you must select a slightly larger heatsink. So far so
good.  The difficult bit is when you find that you need to change other
components in the drive circuit as well, like the spread inductance of the
driver transformer, a damping resistor, a duty cycle of a drive signal etc.
And of course you may find that the HOT you planned to use is entirely
unsuitable for this application ...

Anyway, it's never a case of just 'drive it hard'.

  4.19) Testing of replacement HOTs

The following is useful both to confirm that a substitute replacement HOT is
suitable and that no other circuit problems are still present.  However,
single scan line anomalies (particularly when changing channels and/or where
reception is poor with a TV or when switching scan rates and/or when no or
incorrect sync is present with a monitor) resulting in excessive voltage
across the HOT and instant failure are still possible and will not result
in an HOT running excessively hot.

(From: Raymond Carlsen (

After installing a replacement HOT in a TV set or monitor, I like to check the
temperature for awhile to make sure the substitute is a good match and that
there are no other problems such as a weak H drive signal. The input current
is just not a good enough indicator.  I have been using a WCF (well calibrated
finger) for years. For me, the rule of thumb, quite literally, is: if you can
not hold your finger on it, it's running too hot, and will probably fail
prematurely. Touching the case of the transistor or heat sink is tricky....

Metal case transistors will be connected to the collector and have a healthy
pulse (>1,200 V peak!) and even with plastic case tab transistors, the tab will
be at this potential.  It is best to do this only after the power is off and
the B+ has discharged.  In addition, the HOT may be hot enough to burn you. 

A better method is the use of an indoor/outdoor thermometer. I bought one
recently from Radio Shack for about $15 (63-1009). It has a plastic 'probe' on
the end of a 10' cable as the outdoor sensor. With a large alligator clip, I
just clamp the sensor to the heat sink near the transistor and set up the
digital display near the TV set to monitor the temperature. The last TV I used
it on was a 27" Sanyo that had a shorted H.  output and an open B+ resistor.
Replacement parts brought the set back to life and the flyback pulse looked
OK, but the transistor was getting hot within 5 minutes... up to 130 degrees
before I shut it down and started looking for the cause. I found a 1 uF 160
volt cap in the driver circuit that was open. After replacing the cap, I
fired up the set again and monitored the heat sink as before. This time, the
temperature slowly rose to about 115 degrees and stayed there. I ran the set
all day and noticed little variation in the measurement. Test equipment doesn't
have to cost a fortune. 

  4.20) Oscillation or ringing at HOT base?

"Could this be happening because I'm using the wrong HOT?"

At these relatively low frequencies your scope probe is probably not suspect,
although you should keep loops small and not underestimate the effect of high
dV/dt inside the line transformer.

First, how do we know if the oscillations were already there with the
original HOT ? I suppose you have never measured that ...

Second, it is NEVER a good idea to replace a HOT with just any other
type because there can be significant differences which are not at all
visible in the spec. Even for specialists it's quite difficult to
optimize the drive conditions for a new HOT type. You're on your own here.

Third, some oscillations are normal because the inductance of the base
drive transformer does form part of a damped resonant circuit. Usually
these oscillations show after switching OFF the HOT and they can be a
problem if the negative Vbe is insufficient during the flyback pulse
and the transistor might be turned back on, which would kill it.
As a rule of thumb, Vbe should be -2 V during flyback, but there are
exceptions. Depending on the HOT type, some damping resistor may have 
to be applied. I've never heard of oscillations during HOT ON.

If the collector waveforms (V and I) seem OK and the HOT does not
overheat then maybe you shouldn't worry too much. But do check at high
beam current too (max. brightness on a white picture) because then the
HOT must be switched on earlier to provide the magnetizing current to
the line transformer too. 

(From: Alan McKinnon (

I've run into this kind of thing several times recently on different
sets, this is what I've found:

1. Do you have a damper resistor (about 30 or so ohms) across B-E of the HOT.
   It can go open circuit, causing weird stuff.

2. Check the supply feed resistor into the transformer of the line drive
   circuit. These can go hgh, especially if it's a high value resistor
   dropping a supply of 120 odd volts down to 30 or so.

3. Check all small resistors and caps in the line drive circuit, take then out
   and measure them, in circuit reading are funny in this area.
4. Try a new drive transformer.

  4.21) Why do some monitors fail if driven at the wrong horizontal frequency?

I think it *should* not have failed. It is the purpose of the deflection
processor to guard that the line deflection never runs at a forbidden
frequency. It is possible that such protection is not good enough,
that when the PLL is not in lock it might generate a very irregular
line drive. That can prove to be immediately fatal to a line transistor.
All it takes is one line length that is too long, followed by a flyback
pulse that is too high, and it's all over. Second breakdown, terminal.

Trying a too high line frequency is usually not harmful, generally your
EHT will be too low and your deflection current amplitude too.
The HOT might eventually die from overheating due to bad drive conditions,
but otherwise any decent monitor should be able to withstand it.
In the better cases it simply refuses to sync on an illegal frequency.

1. While increasing the frequency of the horizontal drive *all other factores
   being equal* should not result in HOT death, all other factore are not
   always equal.  The sync circuits may select an improper set of voltages,
   the drive waveform (as mentioned) may be insufficient, etc.

2. The sync may lock to a submultiple so the drive may be at 1/2 H resulting
   in a too long on-time and poof.

As noted, this can be a one shot affair - absolutely no warning.  My
recommendation remains to attempt if at all possible to obtain the
specs for any monitor you intend to use.

  4.22) Changing scan rate on fixed frequency monitor

Here is a discussion about modifying a monitor to run at other scan rates than
for which it was originally designed (in this case, a lower one):

"I would like to convert IBM XGA2 (39.4 KHz JH) monitors (9515, 9518) to work
 at VGA (31.4 KHz H) rates.

 I have the schematics, and with datasheets/application notes from chipmakers'
 websites altering the scan-frequencies and the mode-switching logic has been
 relatively straightforward.
 That's as regards the 'small-signal' bits (and frame output), to leave a
 39.4 KHz scan for 75 fps graphics and provide 31.5 KHz for the 400-line DOS
 text (VGA startup) mode, and 350-line graphics.
 *But* the HOTs blow up at 31.5 kHz: after a while in the first of a number
 of 9515s, almost instantly in 9518.  The corpses and heatsinks are very hot,
 as in 'high temperature' as well as function :-(("

Blowing up the HOT after minutes is due to bad base drive, or (much more rare)
due to LOT core saturation.  Blowing up a HOT immediately is due to too high
collector pulse.

"The base drive looks fine on a 'scope as regards holding the HOTs (or now an
 experimentally-substituted forward diode) firmly saturated for the longer 'on'
 time, and then sharply/cleanly turning them off."

There's more to a good base drive than just 'looking fine' on a scope. The aim
is to minimize the dissipation in the HOT.  This requires just the right amount
of base current and reducing it in a particular manner at the end of each line.
(A process that we call 'de-holing', removing all the holes from the
high-impedance collector region, this must not go too fast.)

* Too little base drive current will cause the collector voltage to start
  rising too early, before the beam has reached the right end of the screen.
  The HOT goes out of saturation too early, causing dissipation.

* Too much base drive current will cause the collector current to reduce too
  slowly during the beginning of flyback. The HOT goes out of saturation too
  late, also causing dissipation.

Usually for lower line frequency you need a bit more base drive.

"I have only a rather vague mental picture of how a LOPT works.  I think that,
 as a load on the HOT, it is almost purely inductive from turn-on (though not
 from the EHT-generating turn-off)."

If there are secondary scan-voltage rectifiers then they add to the load during
(beam) scan. Otherwise it is always inductive.  During flyback, the flyback
capacitor and the secondary flyback rectifiers cause the characteristic flyback
pulse shape.

"Hence I suppose that the longer 'on' period allows the current to
 increase to a higher (and destructive) level."

Not the higher current in itself is destructive, but the increased magnetic
energy is. During flyback, 1/2*L*I^2 is converted to 1/2*C*V^2 in the flyback
capacitor. More energy means more voltage over the HOT. Over 1500 V peak it
will be destroyed immediately.

"If magnetic saturation were to be reached, it would even look more like a dead
 short than an inductance."

That's true, but it occurs mainly due to excess beam current.  Secondary
flyback load causes the primary scan current to rise.

"My questions are twofold: (1) how correct is my understanding?"

Fair enough. You can get a fair idea by running simulations with ideal
circuit models. But to understand what happens inside a HOT you need very
advanced models. Which don't even exist (yet)!

"(2) More to the point, what solution is there?"

You're doing a good job guessing. I'll let you continue.

"One that springs to mind is to reduce the voltage of the HOT's supply,
 something like proportionally to the increase in 'on' time, so that the peak
 current is back to what it is at the higher frequency."

Exactly. Peak current and more importantly peak voltage, because then
also the EHT will be kept constant.
"This would be a non-trivial task, both to provide a second voltage (at quite
 high current) and to switch between them."

Maybe not trivial but a lot easier than you think.
"Though some multi-frequency monitors have such multiple (switched) supplies,
 I don't think all do: what alternatives can be adopted?  How does one
 calculate in this area, or is it (unfortunately) a matter of having LOPTs
 intended for the purpose?"

The LOPT must be optimized for the higher line frequency and the corresponding
short flyback duration.  For lower line frequencies, you want to reduce the B+
voltage so that it is proportional to the line time.  This is in fact very easy
if you use a switched-mode down-converter (is that called a 'buck'?) and you
make the on-time of the switch *constant*.  Now the line frequency may be
varied over a wide and continuous range. This gives near constant EHT too.

On other monitors, a voltage determined by the detected scan rate is returned
to the power supply in place of the usual zener reference to adjust its output.

You can use an extra filter coil to provide constant B+ to the line deflection
circuit or you can use the LOPT primary for that.  There are some subtle
problems with which I will not bother you.

It may be necessary to reduce base drive current for line frequencies beyond a
certain value.

The 'only' other thing left to do is to keep geometry correct for the various
display modes. Mainly by switching in more or less S-correction capacitance
(using FETs or even relays).

Don't assume that you will be able to re-invent the multi-scan monitor on your
own. I think you will find that you have insufficient resources to do that.


  4.23) Yoke and interlocks

"I was asked a question today that I never thought of much before. If 
 you remove the yoke (unplug it) from a TV or monitor is the high voltage 
 still present and if so can this cause damage to the set. Thanks"

The answer is an unqualified maybe on all counts.

Some have an interlock so you lose power to the deflection/set if the yoke
is unplugged - found that out on a Magnavox chassis that would not do anything
without the yoke connections.

Those that continue to have power may result in damage - at the very least,
if the HV is present, you will be drilling a hole in the center of the
screen since deflection will be absent.  The horizontal deflection will
be detuned possibly resulting in overheating, excess HV, etc.

If you do try this, I recommend using a Variac and carefully monitoring
the CRT for presense of a very bright spot (which will turn into a very
dark spot quite quickly) as well as an excessively hot HOT.

BTW, disconnecting the yoke may be desirable to troubleshoot a horizontal
deflection/high voltage if the yoke is suspect.  If the yoke is loading the
deflection output due to shorted turns, disconnecting it may allow high
voltage to come up - just go slow and nothing will blow.

(From: Stefan Huebner (

Some yoke plugs have a jumper wire to remove +B from the flyback when
pulling it. But this may still damage the SMPS since they are not all
build for running without a minimum load. Some yoke plugs only have the
four yoke wires connected; in those sets HV will build up when the plug
is pulled and you may even destroy something in the flyback area (most
likely the HOT) because the horizontal deflection yoke is an essential
part of the flyback circuitry (exception: some monitors, foe example Eizo,
which have separate HOTs for the flyback and the horizontal yoke).

  4.24) Why the yoke is needed to keep the horizontal deflection system happy

If you unplug the yoke (even if there is no interlock), while the system may
still work to some extent, performance will be poor.  High voltage will be
reduced and parts may overheat (and possibly blow up).

* One reason to do this at all is as a quick test to determine if the yoke is
  defective.  While the resulting waveforms will not be totally correct, they
  should be much better than with a shorted yoke installed and there may be
  enough HV to get a spot on the screen.  Caution: It may be quite bright.
  See the document: "Notes on the Troubleshooting and Repair of Television Sets" or "Notes on the Troubleshooting and Repair of Computer and Video Monitors" for more info and precautions.

* Another more fun reason is if it is desired to use the flyback subsystem of
  a defunct TV, monitor, or computer terminal as a stand-alone high voltage
  power supply - but details are left as an exercise for the highly motivated
  (and cautious!) student.  See the document: "Salvaging Interesting Gadgets, Components, and Subsystems" for more details.

WARNING: High voltages and possibly line voltage present at HOT, yoke, flyback
outputs, and elsewhere!

(From: Jeroen Stessen (

Of course, just removing the yoke doesn't work. The flyback capacitor is tuned
for the presence of both inductances: line transformer and deflection coil. If
you remove the deflection coil then the remaining primary transformer
inductance is about 5 times as large.  So, rule-of-thumb, you would have to
decrease the flyback capacitor by a factor of approximate 5.  But that's not

Without the deflection coil, a lot less current runs through the horizontal
output transistor. So, in all likelihood, it will now be overdriven. So you
need to reduce the basedrive. But that's not all:

If you remove the picture tube capacitance and the deflection coil then all
peak energy demand must be delivered from the primary winding of the line
transformer. Even the shortest peak load will cause saturation. The parallel
deflection coil will at least lend some temporary energy, and the picture tube
capacitance does an even better job. A good high-voltage source without the
benefit of a deflection coil is more expensive...

If you *must* get rid of the 'ugly' deflection coil, then you may want to
replace it with an equivalent 'pretty' coil.  But:

* It must be able to carry the peak current without saturation (a deflection
  coil has such a huge air gap that it can not possibly ever saturate, but a
  smaller coil can).

* It must have a low enough dissipation so you might have to wind it with
  litz-like wire (multi-stranded isolated), do not underestimate the losses in
  high-frequency coils,  mostly due to skin- and proximity-effect. 

* Yes, it can be done, good luck.

And you might want to add a discrete high-voltage capacitor.  How to isolate
the wiring (corona discharge!) is left as an exercise to the reader... (We
pot them in convenient blocks).

  4.25) Probing TV and monitor yoke signals

Due to the particular nature of the signals on the deflection yoke, special
techniques must be used to safely view them on a scope.  Note that the
following also applies to some extent to deflection signals in tube-type
video cameras and other similar devices as well.

You cannot put a ground clip on either pole of the line yoke, because one
side carries the flyback pulses which are approx. 1000 Vpp and the other side
carries the S-correction voltage which may be 100 V average with a (?) 50 Vpp
line parabola.  Both signals are at line frequency (15.75 kHz and up).  You
need a 100:1 oscilloscope probe for the 1000 Vpp signal!   The field coil
carries lower voltages (order of 50 Vpp max) and at field frequency (50 to
120 Hz).

Usually you connect the scope ground to a large heatsink, unless that carries
a warning label that it is not grounded...  ALWAYS check first to identify a
proper ground and use an isolation transformer for safety since even this may
not be the same as the power line ground.

If you have a current probe for your scope, this can be used to monitor the
various current waveforms.  I have used my Tektronix current probe to view the
yoke current on TVs.  The rendition of the horizontal deflection current
waveform is quite good.  However, the vertical suffers from severe distortion
due to the low frequency cutoff of this probe.

You can build a not-very-fantastic (but quite usable) current problem using a
split ferrite core of the type used on keyboard and monitor cables (preferably
one that snaps together).  The following will work:

* Wrap seven turns of insulated wire around one half of the core.
* Solder a 2.2 ohm resistor across the two leads to act as a load.
* Connect to the vertical input of your scope via a coaxial cable or probe.

You can experiment with the number of turns and load resistor value for best

To use your fabulous device, insert one and only one of the current carrying
wires inside the ferrite core and clamp the two halves together.

For a typical TV horizontal deflection yoke, this results in about a .3 V p-p
signal.  The shape was similar to that from my (originally) expensive Tektronix
current probe.  Enjoy the show!  Due to its uncompensated design, this simple
probe will not work well (if at all) for low frequency signals.

  4.26) Tweaking the deflection rates in a fixed frequency monitor

Pulling a fixed frequency monitor by more than a few percent will likely
be a problem.  I know this is not the answer you were looking for but
getting a new inexpensive video card may be a better solution.

If not, you are looking for an adjustment called horizontal oscillator,
horizontal frequency, or horizontal hold.  If you do tweak, mark everything
beforehand just in case you need to get back to the original settings.
There is some risk - changing it too far may result in damage either
immediate or down the road - I have no idea.

  4.27) Breathing compensation

Breathing is a change in picture size caused by a variation of the EHT voltage
due to varying beam current loading. The EHT has a typical output impedance of
(approx.) 1 M Ohm. Average beam current varies between 0 and 2 mA, so the
voltage drops by 2 kV (from 30 kV), or worse.  At lower EHT, the electron beam
is easier to deflect, so the picture size will increase.

Brighter picture -> higher beam current -> lower EHT -> larger picture.
That's the law of physics.

The breathing can be compensated by decreasing the deflection current
amplitude as a function of the beam current. A voltage called "EHT-info" is
generated from the beam current as delivered by the line transformer. This is
fed to the deflection circuit for feed-forward correction. For horizontal
deflection this means feeding it to the East-West modulator.

And here's the catch for you Americans: many USA sets don't have an East-West
modulator! Most picture tubes (for the USA) with 90 or 100 degrees deflection
don't need (much) pin-cushion correction, so there is no fundamental need for
the East-West modulator and so there is also no correction possibility for the
anti-breathing. There are other options for corrections, like creating a
deliberate higher output impedance for the V+ deflection supply etc., but
these are passive measures which will lack accuracy.

Sets with 110 degrees deflection angle (common in Europe) will need active
East-West correction and should also perform adequate anti-breathing. Although
that is still difficult enough, I should know...

The LD/DVD "Video Essentials" contains test pictures to check your breathing
performance. It advises you a simple counter- measure: reduce the beam current
by lowering the contrast setting ! This will also improve the sharpness (less
spot blooming) and increase the lifetime of the picture tube (less burn-in of
the phosphors). But you need to darken the room because the picture will
obviously be much darker...  ( )

Written by Samuel M. Goldwasser. | [mailto]. The most recent version is available on the WWW server [Copyright] [Disclaimer]