HV flyback design theory?

Hm, cool, it should pick high or low if it's overcoupled -- you know how IF coils become double-peaked? Or, if you've never done radio alignment, well, nevermind.

Could possibly try increasing leakage inductance, to make it double tuned, much like a Tesla coil.

This isn't the easiest thing to control, since it's determined by core geometry and stuff. Best I would recommend is:

- Make an autoformer on a nice big ferrite core, some turns bifilar. Treat this as the primary: CT to +V, drains to the ends.

- Connect the primary proper in parallel with this (drain to drain). It no longer needs to be a CT winding.

- Now you can control the oscillator inductance and capacitance (by connecting components in parallel between drains), and leakage inductance (by connecting some in series with the actual primary).

- You can tune the oscillator with the FBT disconnected. Target ~30kHz. You have a degree of freedom here: the ratio sqrt(L/C). The ideal value of this depends on circuit impedances -- one of the parameters you need to solve for, effectively. The target frequency means holding the product

To draw more power -- making more output voltage -- make the oscillator a lower impedance so it can draw more current from the same supply voltage. Tune the coupling (leakage / series inductance), and primary turns to a lesser extent, until it's spot on.

Note that load impedance depends on load current, obviously enough, so you have to tune it over all again if there's a load on it. Resonant supplies, right?

Tim

yes sure, critical coupling brings a flat top, undercoupled has still single peak but lower amplitude, overcoupled shows two peaks. Another experiment I've just done is changing the capacitor on the "new" primary winding, the one between the two drains. Well, it doesn't change much if it's 100nF or 47nF, with the 47n one, the drain waveform becomes more "ringy" on the skirts but DC output and oscillation frequency vary just a little (like 7.3kV instead of 7.55kV with the 100nF one).

hmm more parts, but worth a try. But there's really something I'm missing here, and it must be big: after one hour running, also this ferrite got quite hot, not too hot to touch but still it means that I'm running at or even more the rated flux density. Still on almost all these transformer I'm getting a little bit more than half the supposed DC output for the CRT they were designed for. I'm quite sure this particular one was from a BW 16" or 19" TV, meaning something like 15 kV or more. And I'm "easily" getting 7.5kV. The smaller one was supposed to give around 11 kV and I'm getting 6.6kV out of it.

They all have an additional "primary" winding connected in parallel to the "Leg-1" primary. This additional primary is wound on the HV leg "Leg-2". does this mean the two legs aren't very well coupled and they needed to make this strange trick to increase the coupling to the HV winding? What happens now that I'm not using anymore the original primary and there's no current flowing in this additional winding?

well, hopefully once I close the voltage loop, it should try to regulate it with the varying load unless I set it too high in the first place.

Thanks again! Frank

gh

w IF

well,

gle

ed,

reat

t no

ce

e of

e.

ouch

y.

ut

you

ies,

it

TV flybacks are designed to be used with a Deflection Yoke. Together, they are designed to create a sawtooth waveform. For System M (NTSC, USA) that i s 15,734.34 Hz (Color) to 15,750 Hz. (Mono) They are driven with a pulse, w hich is shaped by the two or three sets of coils. A Damper is used to quick ly return the sweep to it's lowest level, which returns the beam to the lef t side of the screen. A Deflection Yoke is a combination of inductors and c apacitors. Look at the schematic of a TV, and see what I'm talking about. P referably, the one for the TV these were intended for.

Yes, I had figured all this much alone already. As I said I think in the previous message, I could make the (so far) best circuit by simply replacing the deflection coil (horizontal one) with a big-ish inductor. However, the inductor was becoming very hot, and by using physically larger inductors (or best of all, another deflection yoke) I would be making a large "power supply" and one that's making a large stray magnetic field close to a CRT. I didn't think it was a good idea (and no, I haven't been able to find the schematic for the old TV I took the CRT from).

Now, I think that John Ross, the designer of the Vectrex console video circuit (another XY magnetically deflected CRT one), was facing the same problems when wanting to design a high voltage supply for the 9" console's CRT using a TV-designed flyback transformer. Interestingly, the service manual for this console, shows that the CRT's anode supply is 5.8kV, which is rather low for a black ad white 9" CRT. For example, measured anode supply on my 1979's green phosphors computer monitor is 10 kV.

John Ross seems living in a car now, and a couple of other co-designers of that console died years ago. So I can't really hope to contact him and have a small chat about his design. I suspect though that the 9" CRT was working enough well at the low anode DC supply and that allowed also to use less deflection current. If it was still possible to focus the beam, they should just play with cathode-grid voltage and reduced cutoff voltage probably. Just guessing, but it's possible.

Best regards Frank

Ok, I did experiment with this "additional primary". If I disconnect it the max DC output remains almost the same, but the resonance peaks are greatly widened (or so I think) as the max DC output is almost constant now from about 100 KHz down to 18 KHz (as before, the oscillation frequency the mosfet settle to can be varied by changing a parallel capacitor to the old primary winding, though it will jump abruptly from about 50 KHz down to 20 KHz). So another (useless?) discovery. Frank

For a vector display you don't need the same amount of anode current because the display is mostly black, and the black pixels aren't scanned. So you can run a lower voltage.

ing

er

pply

ve

DC

You just don't get it. You can't use just any old yoke. The flyback and yok e are impedance matched, You can't just slap an inductor across the flyback , it is a resonant load. Without the additional components, the flyback see s the SRF of the inductor, not a proper load. Yokes would catch on fire whe n they became detuned, and in some cases, the heat would break the neck of the CRT. Zenith had one design that destroyed thousands of CRTs, because of the horizontal sweep design. It resulted in the recall of hundreds of thou sands of their color TVs to replace a special 'safety capacitor'.

Without the proper waveshape of the drive to the flyback, it is not going t o produce the proper output. Even if you do cobble together a circuit, what will you do when the flyback fails? I replaced hundreds of failed flybacks , and many yokes over the decade that I serviced TVs. That was from the mid '60s to the mid '70s. Stancor and Thordarson made most of the third party yokes and flybacks. Their catalogs often provided schematics for their part s, and often required component changes to the chassis to reduce the failur e of new parts.

I'll discover this soon, As long as the beam can still be focused, it should work.

Frank

I think I've got also this. Since my only attempt to replace the horizontal yoke with a similar inductance ended quickly and that route was abandoned in that same moment.

and yes, I've quickly discovered also this aspect. The right output was never obtained with other circuits (well, I was close enough with the "replacement" inductor, but it wouldn't last probably and would be a poor solution).

find another one? :) So far all flybacks I've tried can produce a bit more than half of their rated output with a self resonant drive. I hope it's suitable to run the CRT in a XY vector display, but as soon as I rewind the vertical yoke's coils, I'll discover it.

If and when I run out of transformers, I can always buy a cheap chinese "static precipitator". The problem with this is that I then need to use another supply to generate the G2/G4 and cutoff biases.

Popular replacements in EU were made by HR Diemen.

Thanks Frank

Have you considered using a transformer, like this?

Oh! Thanks for the pointer indeed. Though the problem if I use something like that (that seems too small core-wise to me, also there're no magnetic/electric specs that I could find?) is that I don't get another low voltage (say 350 -

Frank

Hi all, let me post a "little" followup on my attempts at driving (succesfully) some (old-ish) BW TV flyback transformers. My original goal was obtaining the anode supply and G2, G4, K biases voltages to power a BW 11" CRT for XY display. My particular CRT's datasheet called for 11kV nominal anode supply. With a very clever circuit by Tim Williams, I could easily drive most of the old transformers I have, but they would output about 60% to 70% the "nominal" anode voltage, this is driving the transformer with a new center-tapped primary with a pair of push-pull mosfets, using whatever frequency they decide to run at. Also, there's a limit to how few turns you can drive or how high the supply can be, and the limit is obviously given by core saturation. A possible explanation for the much reduced HV value is that the push-pull drive unnecessarily swings the core between two extremes of the magnetization curve, but the HV secondary only rectifies one of the half-waves. In the original TV circuits, these transformers' HV secondaries only rectify the higher voltage and fast flyback pulse (which is nominally

Top waveform is the drain (50V/div), bottom is the gate drive (10V/div), horizontal is 20us/div.

I haven't added any capacitor in parallel to the mosfet, since I first wanted to see what's the shortest pulse I could get. Also the damping is done by the mosfet's body diode (I'm not sure it would help to have an external diode in this case?). I have delayed the gate turn-on to allow for a complete pulse to happen, this gives about 9.3kV DC output (this same transformer, with a push-pull drive topped at 6.6kV). First observation, the main resonance period is about 20us, so the primary's Q isn't as good as it was the horizontal yoke's coil one (expected). Second observation: the pulse repetition frequency is so low in this case that the whole transformer is a loudspeaker at about 6 KHz. If I delay the next gate turn-on to just right before the next positive pulse of the resonance waveform, the HV DC voltage increase of a few hundred volts (makes sense since the ringing waveform can go on for several cycles if I turn off the gate drive). Now, if I use a new primary (much less turns than the original), the resonance peaks become much shorter (less turn, less self-capacitance, less inductance).

Now, I believe this can be turned into a working PWM drive (probably using a new primary as the original one requires too slow drive frequency?), but I would like informed opinions from the experts :) What I still don't know is:

-) How to allow the drive circuit to "wait" enough after the flyback pulse ended before initiating the next gate drive?

-) Obviously in the tests I'm running in a "continuous magnetization mode" since when the next gate drive happens, the core has still some (high-ish?) magnetization.

-) I know it would probably benefit from a larger core gap, but I'm sure I first need to figure out a proper driving circuit.

Any hint is always welcome. I'm reinventing a lot of wheels for sure. Thanks Frank

If you're doing it with an European 15625 Hz transformer, the scan time is 53 us and flyback 11 us.

Did you have the horizontal deflction coils with the transfromer. The TV needed both. The coil system is tuned to third harmonic of the scan rate to give the necessary S-form to the deflection current.

I came pretty close with my measures, less than 10% error :)

Yes, I have figured out the third harmonic trick too, the designers did a very good job with these circuits really. And yes I still have the original coil, but as I wrote, it will be used for an XY monitor, so there will be no raster (= no scan). And by the way, the "perfect" S-form is obtained also with the S-correction capacitor, otherwise the image would be kind of compressed more on one side.

Thanks Frank

You may not get the full-size kickback from the transformer if the deflection part load is missing.

As a student of technology in the 1960's I got some extra income repairing tube-TV's often just before Batman was due.

Seems my post wasn't clear enough :) Yes, one can't really get the full-size kickback unless you replace the deflection coil with a suitable low-loss inductor. The inductance of the transformer primary winding, at least in my case, is 575uH and the original deflection coil is 83uH. They are in parallel (AC-wise). This has basicly two effect that I could observe:

So, of course I realized soon enough that I can't really get the nominal DC output, but still it seems I can get enough of it if I can figure out how to replace the HP-8116A that's supplying the gate drive :) And maybe without making the transformer sing at about 6 KHz would also be a plus (this could probably also be fixed with a new primary winding).

Thanks Frank

not much of my post is getting through, is it? But, hey, I'm not a native English writer/speaker, so that can sadly happen. I am running this transformer with a simple mosfet switch to ground and a positive supply of 12V limited to 3A. Mosfet's gate drive waveform is obtained with a pulse generator (HP-8116A) and the drive pulse width and repetition frequency has been tweaked to obtain the highest high voltage output and waiting for the flyback pulse to go to zero, then being swallowed by the mosfet body diode (you don't see the negative going wave on the scope) and only at this point, more or less, the mosfet gate drive turns it on again. This is: 100us (approximately) mosfet ON time and repetition rate < 7 KHz. This frequency can be heard by most of us.

With this simple circuit, the high voltage output from this transformer is 9.3kV. Since the CRT that was in the same TV set required a nominal

Frank

You should use the TV line frequency as well as you can, 64 us cycle. Most of the tube TV horizontal deflection parts are made to work on the line frequency only. There may be all kinds of surprises in the internal strays of the transformer and coils if you deviate from the design frequency.

English is not my first language, either.