This is only a preview of the June 2022 issue of Silicon Chip. You can view 41 of the 112 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Items relevant to "Spectral Sound MIDI Synthesiser":
Items relevant to "Buck-Boost LED Driver":
Items relevant to "Arduino Programmable Load":
Items relevant to "500W Power Amplifier, Part 3":
Items relevant to "MOS Air Quality Sensors":
Items relevant to "Revised Battery Charge Controller":
Purchase a printed copy of this issue for $11.50. |
Vintage Television
The Admiral 19A11S TV and its unique
deflection circuit
By Dr Hugo Holden
This set features a unique horizontal deflection
circuit that has been sitting under everyone’s
noses for about 67 years, invented by Britons
Faudell & White. Over the last 35 years, I
have asked many veteran TV technicians if
they know about it but so far, none have been
familiar with the technique.
Early CRT TV sets like the Admiral 19A11S used electrostatic deflection rather than magnetic deflection,
which became the standard until
cathode ray tubes (CRTs) were made
essentially obsolete by LCD screens.
The difference is how the electrons
in the cathode ray are deflected to
land at the desired spot on the front
of the screen.
As you might guess from the names,
electrostatic deflection works by creating an electric field to deflect the
electrons, basically by charging capacitors placed on either side of the electron beam (left and right for horizontal deflection or top and bottom for
98
Silicon Chip
vertical). In contrast, magnetic deflection uses coils to create magnetic fields
that bend the electron stream.
The amount of deflection created by
the electrostatic method depends on
the applied voltage, while the magnetic deflection works on the current
through the coil. Either way, this voltage/current must be carefully controlled so that the electrons trace out
a zigzag to illuminate the phosphor(s)
on the face of the cathode ray tube.
The circuit concept used for the
horizontal deflection system in the
Admiral 19A11S has not been featured
or described in standard television
technology textbooks such as those
Australia's electronics magazine
by Fink, Grob or Von Ardenne. As far
as I am aware, the only two TV sets
which contained this “circuitry masterpiece” were the Admiral 19A11S
and the Motorola VT71; both use the
7JP4 electrostatic CRT.
Perhaps it was overlooked by people servicing the sets because ‘it just
worked’ and they put no further
thought into it; they needed to fix the
TV and get it back to the customer.
I recently posted this circuit on a
vintage TV internet forum, again seen
by many people with a long history
in TV repair, design & construction.
Nobody was familiar with it, and it
surprised most.
Before reading this article, imagine
you have studied all there is to know
about designing TV sets with valves.
You walk into an exam room and are
confronted with this question:
Design a circuit with a single triode
and any other R, C & L components
you wish which runs from a 250V DC
supply (ignoring the triode’s heater).
It must produce two anti-phase 450V
peak-to-peak linear sawtooth waveforms (one for each horizontal deflection plate) and be suited to television
horizontal scan and flyback timing.
It needs an adjustable frequency
of around 15,750Hz, and it will be
synchronised to the horizontal sync
pulses in the usual way.
I think most engineers familiar with
television scan stage design would
find this question too challenging. On
its face, it seems impossible. Conventional wisdom is that this task requires
a separate oscillator and a two-triode
para-phase amplifier running from
plate supply voltages of 700V or more
to allow enough linear amplification
for generating the 450V peak-to-peak
anti-phase sawtooth signals.
As argued in some texts, electrostatic deflection was abandoned in
favour of electromagnetic deflection because sets with larger CRTs
required very high voltages for the
siliconchip.com.au
deflection amplifiers. This is because
of the higher drive voltages required
for larger tubes.
Note that in electrostatic CRTs, the
amount of deflection is inversely proportional to the EHT voltage, so if you
double the EHT, you have to double
the sawtooth deflection voltages to get
back to the same picture size.
On the other hand, the amount of
magnetic deflection is inversely proportional to the square root of the EHT
voltage, so if you double the EHT, you
only need to increase the deflection
current by 41% to get back to the same
picture size.
How I noticed this circuit
I came into the possession of the
‘shell’ of a vintage television set, an
Admiral 19A11S, while I was in New
Zealand in the very early 1980s. I had
to bargain hard to get it. In the end, I
think I traded it for a fully working
26-inch colour television monitor.
Unfortunately, back then, I did
not see the wisdom in making pre-
restoration photographs. It consisted
of a rusted chassis with a tuner unit
on it. Most of the RF coils were there,
including the RF power supply. All of
the deflection oscillator parts, including two blocking oscillator transformers and one horizontal output transformer, were missing. No power transformer was present either.
There was no manual available
at the time (no internet either). As
TV broadcasting didn’t start in NZ
until 1958-1959, there was no service
The 1949 Admiral 19A11 set in a contemporary
advertisement is shown opposite, with my set shown above.
information on this 1949 model available from TV service shops.
Luckily, the fellow I got it from had
an old copy of the schematic. It looked
to me like an old treasure map; at least,
I thought of it that way. It was faded
and drawn over in Biro in places and
had tears repaired with clear tape, but
enough of it was there.
Nowadays, it is simply a matter
of searching the ‘net; the entire Rider’s manual for the 19A11S is online,
including everything one could ever
want for servicing (you can find it at
siliconchip.com.au/link/abd4). But
perhaps the lack of information back
then did me good. I had to carefully
study the design of the frame and line
deflection systems to work out how to
re-create the missing transformers and
make the set work again.
It was there that I discovered this
ingenious circuit. The line deflection
stage or horizontal deflection stage
generated two anti-phase linear 450V
peak-to-peak sawtooth waves, running
from a mere 250V DC supply. How
does it do that?
How it works
The circuit documented on the
faded schematic (or in the Rider manual) was drawn in a way that almost
concealed how it worked. After
The functional block diagram (above) and location of the
valves on the chassis (right), reproduced from the service
manual. The Admiral 19A11S shared a circuit with many
other models which are part of the “19A1” series, such as
the 19A11SN, 19A12S, 19A12SN, 19A15S and 19A15SN.
siliconchip.com.au
Australia's electronics magazine
June 2022 99
Scope 1: the voltage across C2 is essentially symmetrical
about GND as it is connected to the cathode of the 6SN7
valve.
Scope 2: the waveform across C1 is a mirror-image of that
shown in Scope 1; however, the average voltage is quite
a bit higher (around +138V in this case) due to C1 being
connected to the anode of the valve rather than the cathode.
re-drawing it, I realised what they had done. The designer
had nested a blocking oscillator inside a low-frequency
(2-3kHz) resonant circuit.
Due to the high-Q nature of the resonant circuit, when
it ‘rings’, there is voltage magnification above the applied
voltage. The really clever part is that since the first 20-30°
of a sinewave is pretty much linear, the blocking oscillator simply chops out about ±30° of the oscillation cycle to
produce a near-perfect linear wave.
The circuit, re-drawn, is shown in Fig.1. And depending
on the width adjustment, it can produce 350-450V peak-topeak sawtooth waves. In the set, they are generally about
400V peak-to-peak but adjusting the control can easily give
480V peak-to-peak.
The circuit is very efficient, and calculations show that
the 6SN7 is run well inside its maximum plate dissipation.
The peak-to-peak cathode voltage is just on the edge of
its maximum rating. In the blocking oscillator circuit, the
valve is not conducting most of the time; it only conducts
during flyback. When the valve conducts, it charges C2 to
about +200V at the end of the flyback period. By then, C1
is discharged to about -62V.
Scope 1 & Scope 2 show the voltages across C2 and C1:
As expected, since there is no average DC of any significance on the transformer, the cathode waveform (voltage
across C2) straddles zero volts. However, that is not the case
for C1, which swings from -62V to +338V here. The 400V
peak-to-peak amplitude is the same, but it has a +138V offset (approximately half the +250V HT).
Energy is imparted to transformer T2’s magnetic field
during the flyback period. When the 6SN7 comes out of
flyback (conduction), T2’s field begins to collapse at a rate
determined by the values of C1 and C2.
Due to the circuit Q, the voltages across C1 and C2 can
rise well above the power supply voltage; or at least, they
would do if the oscillations were not reset by the blocking oscillator conducting again about 25-30° into the sinewave cycle.
Even though the voltage on C1 falls to -62V and the cathode voltage on the valve climbs toward +200V at the end
Fig.1: the electrostatic horizontal deflection driver
circuit is brilliant in its simplicity. Using just one
active device, two transformers, six capacitors, three
resistors and a trimpot, it produces both 450V peakto-peak sawtooth waveforms using an HT of just
+250V. Transformer T2 acts as a phase splitter and
importantly, also forms a resonant circuit with the
two 1nF capacitors,
boosting the output
swing to the required
level.
100
Silicon Chip
Australia's electronics magazine
siliconchip.com.au
Faudell and White’s Time Base
Faudell and White have developed a time base which gives a
push-pull output. It consists of a transformer-coupled blocking
oscillator time base in which a large inductance is connected in
series with the H.T. supply to the time base. A condensor is connected between the valve side of the inductance and the negative
supply rail as shown in Fig.66.
Fig.66
Scope 3: this scope grab demonstrates how the valve’s plate
(anode) voltage stays above the cathode voltage during the
flyback conduction period. The valve is not conducting
the rest of the time, so it doesn’t matter that the anode and
cathode voltages cross over.
Fig.2: a recreation of an excerpt from Time Bases by
O.S Puckle, Chapman Hall 1951, attributing the clever
horizontal driver circuit to C. L. Faudell and E. L. C. White.
It names several British Patents; unfortunately, unlike US
patents, they are not viewable or searchable online.
of flyback, the valve’s anode voltage stays higher than the
cathode during conduction (flyback). This is due to the voltage on the primary of the blocking oscillator transformer.
You can see this effect in Scope 3.
So the plate voltage is always higher than the cathode
voltage at any moment during flyback. After flyback, the
valve is cut off, but as shown in Scope 3, the plate voltage
falls below the cathode voltage. Thus, the oscillations on
the plate from the blocking oscillator transformer do not
affect the scan as the valve cannot conduct with its plate
voltage below the cathode voltage.
Also, because it is a blocking oscillator, the grid-to-cathode
voltage is negative during the active scan time, and the valve
is not conducting. This is unlike magnetic deflection, where
the output stage valve conducts during the scan time and is
cut off during flyback. This is possible because the required
scan power for electrostatic deflection is minimal, only a
few milliwatts.
The ‘load’ for electrostatic deflection is merely the deflection plate tie resistors, which are in the order of 2-5MW.
The top of the chassis post-restoration. The only
components mounted on this side are the valves, CRT and
transformers, lending it a tidy appearance.
siliconchip.com.au
Rebuilding the set
Since I had no data on the missing transformers, I had to
guess at their parameters. I wound T2 as two independent
The underside is another story; while not exactly the worst
mess we’ve seen, the components are mounted in locations
based mainly on convenience for the point-to-point wiring
employed. The turret tuning mechanism is enclosed on five
sides by metal shielding.
Australia's electronics magazine
June 2022 101
102
Silicon Chip
Australia's electronics magazine
siliconchip.com.au
The entire circuit
of the Admiral
19A11S has been
reproduced here,
as the quality of
this scan is much
better than the
circuit that was
used to restore
the set.
siliconchip.com.au
Australia's electronics magazine
June 2022 103
windings on a small ferrite core (scavenged from a small transistor TV line
output transformer). The inductance
turned out to be 4H per winding, but
at the time, I was aiming for it to resonate at about 2kHz.
Later, I measured a transformer
from a VT71 set, and it was 1.62H per
winding. So the resonant frequency
of T2 in the original system (tuned
by two 1nF capacitors) was intended
to be about 2797Hz (probably 3kHz).
Mine resonated at 1780Hz, and it
worked fine.
Who was responsible for it?
Recently, browsing the textbook
Time Bases by O. S. Puckle, Chapman
Hall 1951 (the first edition was 1943), I
came across the circuit shown in Fig.2
by Faudell & White. Clearly this is
the same circuit, although, unlike the
Admiral circuit, the low-frequency
resonant circuit is split in two. However, the function is the same.
Restoring the rusty chassis
In the early 1980s, there was no electroplater in my locality that I could
trust it with. So the next best move was
to have it painted after I had removed
A rear shot of the Admiral
19A11S chassis after restoration.
the rust. Later, my preferred method
for chassis restoration was to use a
bead blaster with fine glass beads to
blow off the rust and have it plated
with electroless nickel.
I have completed several TV sets
with that method: an Andrea KTE-5,
RCA 621TS and HMV 904. You can see
the result on the HMV set in my article (November 2018; siliconchip.com.
au/Article/11314); there are before and
after photos on page 91.
Looking retrospectively, I chose an
interesting colour scheme of blue and
white, as you can see in the restored
chassis photos. The paint is two-pack
epoxy which is oven baked, so it is
extremely tough.
The two rectangular aluminium cans
on the chassis top contain transformers
Fig.3: I came up with these
modifications to the set to change the
CRT drive from AC-coupling to direct
coupling and add vertical retrace
blanking. While it worked, I did not
keep these changes as I preferred the
alternative shown in Fig.4.
104
Silicon Chip
Australia's electronics magazine
siliconchip.com.au
T1 and T2. Notice how on the centre
top of the chassis underside, to the
left of a smoothing choke, I added a
6AL5 valve (acting as a DC restorer).
It is mounted in a socket on two metal
posts and held in place by a coil spring.
The turret tuner assembly in these
sets is quite advanced for 1949,
although the 1946 RCA 621TS set had
a very good turret tuner using three
6J6 valves.
I needed a new CRT socket for this
set, as the one that came with it was
crumbling away. These are currently
available on eBay, but back in New
Zealand in the early 1980s, no such
part was available. So as shown in
the adjacent photo, I machined one
out of Nylon.
The pin retainer inside is a round
section of fibreglass PCB material
with the copper removed and countersunk holes with sharp edges. As a
result, when the socket is assembled,
it rotates into place and locks all of the
pins into position.
Upgrading the set
One failing of the design of this set
is that the video output stage is AC-
coupled to the CRT. This means that
As my CRT socket was
crumbling and I could not
source a replacement, I
had to machine this one out
of Nylon and fabricate a
retention mechanism using a
sheet of fibreglass taken from
a blank PCB.
the video signal’s DC component is
lost. The effect that this causes on
changing picture scenes is well known
to every television engineer.
I came up with two possible methods to remedy this:
1) Modify the circuit for direct coupling from the detector-video output
stage to the CRT (as shown in Fig.3).
2) Add a DC restorer (see Fig.4).
After trying both methods, I elected
to keep the added 6AL5 DC restorer.
With this circuit, the raster is black
or near blacked out when no signal is
being received.
I also added vertical retrace blanking and a 5.5MHz sound trap (suited
to NZ TV reception) to improve the
sound take-off gain from the video
output stage.
This method of wiring in the DC
restorer uses typical techniques developed by RCA to minimise the loss of
high-frequency signals in the video
output. It was interesting to look at
the notes I made at the time in the
neat handwriting I had back in the
early 1980s.
Shortly after this, I finished my
career in TV and VCR servicing and
then went to medical school. I became
an Ophthalmologist specialising in
cataract surgery which is my current
SC
line of work.
Fig.4: these changes
keep the AC-coupling
to the CRT but add a
DC restorer that alters
the DC CRT level to
match the average
DC signal level.
It’s a slightly more
complicated scheme,
but it provided a better
result, so I kept it.
siliconchip.com.au
Australia's electronics magazine
June 2022 105
|