Ingenious Deflection Circuit, it's British.

[Argus25]

Recently on another thread I introduced a horizontal deflection circuit for electrostatic television that I discovered in a vintage Admiral TV back in the 1980's. It is quite something, the story is here:

http://worldphaco.com/uploads/Admiral_TV..pdf

Combing through a book by Puckle, I finally found it documented. It appears as though this amazing arrangement was patented in the UK but only commercialized in America.

[Nuvistor]

That restoration looks excellent, and what a career change. TV sets were magnetic deflection when I started fixing TV's and only ever saw scopes with electrostatic.

Frank

[kalee20]

Brilliant to get such a large amplitude linear sweep, I'm impressed with the waveforms!

[FERNSEH]

Let's find out if this Admiral 7" TV set has the special frame timebase circuit.
I believe it is a 19A11.

DFWB.

[Argus25]

The frame timebase is clever too, but it is conventional. It has a standard blocking oscillator (1/2 6SN7) feeding a twin triode para-phase amplifier based on a 6SL7 with 4.7Meg anode load resistors. To get the high HT to run it, about 900V, the penny dropped with the designers that since the two plate currents are equal and out of phase, the average current is a constant, so they simply dropped this output stage circuit into the ground leg of the 6kV EHT bleeder chain and got the supply voltage that way. Apparently though this method was not uncommon. The 6kV in this set is generated by an RF power supply based on a 6V6 and powered by the 250V DC supply rail.

[FERNSEH]

The first attachment is the circuit diagram of the line timebase of the Motorola 7VT1.

The second attachment shows the underside of the chassis of the Admiral 19A11.
The two large disc shaped capacitors are of 0.001 microfarad each and have a high voltage rating because the deflector plates are at near EHT potential.
The large orange capacitors perform the same function for the frame timebase.

DFWB.

[dominicbeesley]

Thanks Argus, I found it in Puckle too! I lost track of the other thread but I've built my own test rig for this circuit. It works well with 1/2 an ECC83 but I'm having bother with the same rig but with a 1/2 6j16b-v soviet valve...it appears to run into grid current a lot earlier and distorts the cathode waveform and the grid capacitor charging doesn't look abrupt enough....needs more experimentation

I'll try and get some pictures and circuits and a quick write up over the weekend.

It being patented makes sense - too tightfisted or difficult to pay the licencing fee. It's a shame as it is a very good and simple circuit. I was amazed that it worked first time when I tried it.

D

[Argus25]

D:

A 6SN7 seems ideal for it, a 6SL7 for example won't work well for this circuit. So I think it is important that the triode used has a low plate resistance so it can adequately change the charge on the 0.001uF tuning capacitors during the relatively short fly-back time. It would be interesting to compare the plate(anode) resistances of the tubes you are trying. I'm not familiar with a 6J16b-v.

[Neutrino]

Originally Posted by Argus25

Combing through a book by Puckle, I finally found it documented. It appears as though this amazing arrangement was patented in the UK but only commercialized in America.

Patents can be found at:
www.epo.org
Type GB471737, GB479275 or Faudell into the search box. Search by clicking on Patents, click on the underlined title and then on “Original document” which can then be downloaded. Start reading the patent from the heading COMPLETE SPECIFICATION to avoid a repeated sense of déjà vu.
The second patent refers to GB400976 which provides an early description of timebases from 1933.

[dominicbeesley]

Sorry lads, that was a typo they are 6N16b-v (that's 6Н16Б-В in Cyrillic).

From the attached characteristics, I'd say it's about ra~=5k, mu~=25, gm~=5mA/V (rough by-eye reckoning from the screen). (compares favourably with the 6SN7GT ra=8l, mu=20, gm=2.5mA/V)

I suspect something else is the problem - the flyback is fine and certainly enough to give no foldover on a 405line picture and room to spare. There is just some distortion of the waveform for the cathode voltage. I'll try and get some workshop time tonight and double check the circuit...

[Argus25]

Neutrino,

Yes the patents are still there. Also the only significant modification Admiral made to the Faudell circuit was to combine the two inductors of the low frequency resonant circuit on the one core - a very smart move. This was a sensible improvement of economy, I'll explain why:

Clearly the value of the tuning capacitors is the limiting factor during fly-back, they can't be too large, in conjunction with the tube's anode resistance, as it would take longer than the fly-back time to discharge them. So a substantial inductance is required to get the resonant frequency in down the range of 2 to 3kHz for each inductor in the original Faudell & White circuit.

However if the inductors are placed on the same core (as Admiral did) the practical effect of the tuning capacitance is doubled, so the resonant frequency drops by 1/root 2 for the same amount of wire and iron in the transformer and the same value tuning capacitors.

So there was an economy of design here, added by Admiral, to the original F & W circuit and to a circuit already incredibly ingenious for economy.

To convince yourself this is the case, imagine you have two tightly coupled identical coils on the same core, then connecting the tuning capacitor of value 2C uF across one of them has approximately the same effect as connecting two separate capacitors of value C uF across both of the coils individually.

There is also another way to get the same answer, using a principle of electrical equivalent circuits, where you can connect or disconnect any circuit link that always has the same potential on each side of the link, and therefore no current in the link at any time. Call a circuit link like this an Nlink (does nothing link)

If you draw two independent resonant circuits on the same core and create a single wire link between them, nothing changes with the other ends of the circuits floating and thats an Nlink. Then by re-drawing it you can cut a different Nlink between the junction of the two capacitors and the junction between the two inductors with out altering the electrical behavior of the circuit. Then you are left with a new resonant circuit, where the two capacitors are in series, so their value is halved, but the two coils are in series and therefore the inductance has increased by a factor of 4 because the number of turns have doubled and inductance is proportional to the square of the number of turns.

So the net effect of the circuit equivalent transformation with Nlinks is that the resonant frequency has dropped by a factor of 1/root 2 compared to what it would be for one resonant circuit on the core. These are some of the things I had to work out to re-create this circuit back in the early 1980's with no information on it and missing transformers.

Out of interest a good number of problems can be solved very simply and quickly with electrical equivalent circuits, here is one that avoids pages of algebra and get the solution very quickly and easily affectionately called a Bandsaw method:

http://worldphaco.com/uploads/A_SELF...SAW_METHOD.pdf


Hugo.

[Argus25]

Neutrino, the attached image shows the transformation I described in the last post as diagrams from 1 to 4 with the Nlinks in red, it might make it a bit clearer what I was getting at.

[dominicbeesley]

I think I worked out what was going on with my attempt at this circuit, increasing the frequency was a red-herring, it was decreasing the amplitude that was causing the fix. I'm guessing, but I suspect that the heater/cathode insulation is either breaking down or the heater is acting as an anode when the cathode goes too far negative, distorting the cathode waveform. I need to make another heater and HT supply and try moving that up and down to find out...

Here are some pictures of my scope and setup while I was trying to figure this out. I ended up with the resonant transformer being 2x1.8H and the tuning caps anywhere from 3.3n to 10, higher values giving a smaller scan but slightly better linearity

1st image is cathode (distorted)
2nd is anode undistorted
3rd is when I whack the frequency up the cathode goes linear but reduced
4th is grid voltage
5th is grid - cathode difference voltage

D

[dominicbeesley]

And here's the results...phone camera doesn't do the CRT picture any favours it is actually not quite so blue and is a lot clearer, for instance the text below the D is actually readable!

D

[Argus25]

D:

There are a few things wrong with the circuit currently:

For this circuit to work well the proportions of the blocking oscillator transformer are very important. The ratio is the usual ratio, about 1:3 to 1:4 (plate to grid winding ratio) but the self resonance of the secondary winding (which dominates the overall resonance and voltage seen on the plate winding) must be such that about 1/2 its period is equal to the flyback time at around 9 or 10uS or in that vicinity. This relates to the inductance of the secondary and the secondary winding's self capacitance.(The secondary has at least 9 to 16 times the inductance of the primary).

For a blocking oscillator the plate only conducts for the first 1/2 cycle of self resonance of the transformer, because the next half cycle takes the grid to cathode potential negative and cuts off the tube.

With early H blocking osc transformers they were even layer wound with paper between the layers and had an adequate self capacitance to get the correct flyback time. One way to remedy this is to add a tuning capacitor across the secondary if the transformer you have is resulting in a flyback time that is too narrow/short. In any case have a close look at the plate waveform in my Admiral article. The resonance is partly damped by the parallel resistors. Its important that the DC resistance of the primary is not too high, less than 80 ohms probably ok.

Also it appears as though the combination of inductance and tuning caps you have for the main resonant circuit have resulted in a ringing frequency that is a little high, so too many degrees if the sine wave cycle are seen and the waveform then goes non-linear (sinusoidal). Curiously I found for my transformer , it worked out best when each individual winding was around 3 to 4H. It is interesting that the original Motorola transformer measured around 1.6H on an inductance meter, but the meter could be incorrect measuring this iron cored inductor. As you know the the frequency is inversely proportional to the square root of the inductance, so although these values vary by a factor of 2 the frequency difference factor between them is only around 1.4.

My initial calculations for this circuit suggested that the resonant frequency should be about 2kHz for this application so that not too many degrees of the sine wave cycle appeared.

As noted the benefit of the two inductors being associated on the same core (Admiral circuit) is that the resonant frequency drops by a factor of about 70% of what it would be for the same value of capacitance and inductance on separate cores as Faudell had it.

I would suggest; get the blocking oscillator working correctly with the correct flyback time in isolation, then add the resonant circuit back later, and lower its frequency by increasing the inductance/turns (don't take the tuning caps above 0.001uF) Also, if there is any suggestion of asymmetry of the two output waveforms (when you invert one and overlay them on the scope) then make sure the HT feed to the width control is properly bypassed to ground with at least a 1uF capacitor.

[Argus25]

D:
PS: its easy to see from the scope traces, comparing the flyback time to the width of the H sync tip and H blanking that the flyback time is far far too short, probably less than 5uS. For this circuit it is best that it is close to the the whole H blanking interval. When you increase it to the correct value by modifying or tuning the blocking osc transformer, the amplitude of your output waveforms will be nearly double I would expect, because the change in charge on the 0.001uF caps, by the end of flyback, will nearly be double and the 2kHz resonant core will have acquired more energy at that point too.

[dominicbeesley]

Thanks Argus,

I will experiment with that...however I have more than enough deflection volts.

I'm now fairly convinced though that the non-linearity is down to the valve being used as this worked fine with a different valve. I'm sure I measured the flyback time as 14us without the resonant circuit but adding it sped up the flyback - I may however be misremembering.

I just did a quick simulation and adding a diode in series with 10k between the cathode and 0V gives the same symptoms as I'm getting.

I'll try and get some more workshop time this week.

Thanks for the all the guidance and the article!

D

[Argus25]

D:

In theory the tube should have absolutely no effect on the linearity of the scan, because during scan time the tube is; 1) initially reversed biased at the start of scan (anode voltage more negative than cathode) and 2) cut off during the rest of the scan time because the tube grid is more negative than the cathode. The tube only contributes during flyback time which is the only time it is conducting which is why the flyback interval is so important.

Your scope trace shows that the time from start to end of flyback is a slightly shorter duration than the width of the H sync pulse, which is only around 5uS. Aim for around a 10uS flyback is good for this circuit. Its a fun circuit to play with.

Of course in nearly all other scan circuits, eg all magnetic deflection, and most other types of electrostatic deflection, it is the opposite way around, the tube is conducting during scan time so the tube and its bias conditions have a big effect on scan linearity.

In any case if you were worried about the tube you could always build it first with the 6SN7 which is known to work perfectly, so as to perfect the transformer designs first, then substitute in other tubes later.

Hugo.

[Neutrino]

Argus,

Patent GB479275 (page 7 lines 44~74) describes how the inductance of T1 boosts the voltage on C1 and C2. The principle appears similar to that shown in Fig.4 and Fig.5 of Blumlein’s patent GB400976 and is described in Time Bases (O. S. Puckle) under the heading Blumlein’s Thyratron Time Base.

The photographs of the Plate voltage and Cathode voltage in your paper attached to the first post in this thread shows this action.

Considering that the original components were misssing, I think that you did very well with this restoration to get the rather clever time base working

[dominicbeesley]

In theory it shouldn't but if there is emissive cathode material on the inside of the cathode tube or the filament is near the cathode and the cathode goes very negative then the filament will act as the anode of a small diode. This would give the non-linearity observed.

I'll try some different valves, these ones are being abused quite heavily, overcurrent on the cathode during flyback, over voltage on the anode, the h-k insulation seems to say it is good for +100V but no idea the other way, my Russian isn't good enough to glean that from the datasheets.

The other option will be to put the 10k resistor in the bottom leg of the circuit as would lift the whole thing above ground...and save me having to make another 6.3V supply...