What Oscillator is in This PS?

Here is a simplified schematic:

I think that circuit has a hang-up state. Both transistors on, beta-limited, D5 back-biased, zero volts across the inductor.

Can that circuit really make 600 volts?

John

It's suppose to be able to make close to 1kV. I've been playing with it, and it appears a lot depends on the choke. With 100mH the flyback is around 1kV. That would be too much for the MPSA42.

What I am really interested in is how to make it pulse faster. I'm only geting several pulses per second.

Fred

Looks like "Design" by simulator to me... GIGO ;-)

...Jim Thompson

I'm clueless about oscillators. I just taking this design from someone else and trying to understand it before I build it. It's suppose to power a Geiger Muller tube. I only need 450-500V with very little current. 500uA probably at most.

Both transistor appear to be on at the same time. It looks like the on time before the choke discharges can be shorted. I would like to at least double the cycle rate.

Fred

Bit of an idiot question, what Frequency does it run at? I don't see any small caps any where

martin

Like 5 pulses per second. I'm having trouble, probably from my complete ignorance of oscillators, I've lacked a need to study them until now, figuring out what controls the cycle rate.

There seems to be a rest period then a slight ramping of current in the collectors, then a sharp rise followed by a plateau period, then a high speed pulsing discharge at about 3Kpps that last about as long as the plateau period, then back to the rest period.

It looks like an adaptation of Charles Wenzel's (*) geiger-counter power supply circuits to me.

(*) The low-noise oscillator guru,

I went to check his excellent hobby website,

"Change

America has chosen Change! So Change is coming to techlib.com. I Hope you will like the Change. The period between November and January will determine just exactly what the Change will be; I'm not sure yet. But it will be Change filled with Hope and Fairness and Justice and other words.",

followed by links to U.S. Government websites.

Ouch.

James Arthur

I don't think it hangs if L1 has sufficient ESR. That'd keep the thing linear, then the positive feedback would kick in.

If L1 is super-stiff, then yes, I think it could hang.

James Arthur

Did I accidentally fall out of your kill file?

One problem here, aside from the control aspects, is that the energy stored in the inductor has to charge up all the various parasitic capacitances... in the inductor, the transistor, the connected diodes. So it's hard to get boost ratios of 200 or so, and this circuit needs over 300 to get to 1KV. Efficiency drops towards zero as you start to use up all the flyback boost ratio.

The fix is to use more c-w multiplier stages, and preferably use a transformer. Some of the Mouser/Digikey dual-winding inductors, used as autotransformers, help a lot.

I really like blocking oscillators, but they need custom magnetics.

John

I only killed the relay with "infinite" power gain thread :-(

I like current sense and dump. I don't like to rely on something saturating.

...Jim Thompson

I found the original in my archives; it *is* a Wenzel design, with changes. Wenzel used an auto-connected audio transformer for L1.

Also missing is 1.8k in series with the MPSA42's base.

James Arthur

It's a relaxation oscillator that uses the negative resistance (SCR-unijunction) connection of Q1 and Q2. The timing is determined by R1 plus R2 charging C2 until Q1 turns on which then discharges C2 and the cycle repeats. Flyback action in the inductor generates a high ring-off voltage when Q2 turns off.

The resulting pulses are voltage multiplied in D1, D2, and D3 with the caps C3, C4 and C5. Three 180 volt Zener diodes, D4, D6 and D7 in series regulate the output to 540 volts by causing Q3 to rob current from the oscillator as they turn on.

To speed up the period, reduce the size of C2 and/or lower the value of R1 and R2. This circuit may be a bit "fussy" because the negative resistance region is defined by the various resistors and diodes in the design. Play around with it and see what happens.

In series, or to ground?

There's nothing there to guarantee Q2 turns off expeditiously.

...Jim Thompson

Yes, the original schmatics had "Wenzel" marked on them. But, when I went to Wenzel Associates, all I found was crystal oscillator stuff.

This particular one has been modified to run off a 3V,instead of 9V supply, and a choke instead of the audio transformer.

Fred

That was it. I added some more serial resistance to the choke and things cleaned up neatly.

Thanks,

Fred

That did it. I added some more serial resistance to L1 and things cleaned up well. It's kiinda like MOSFET gate oscillations.

Thanks,

Fred

I just did a design with two supplies on there, one up to x30. That uses a CCFL transformer, the other lower x supply a Coiltronics Versa-Pac, those with the six independent windings. Works like a charm and both transformers are SMT and catalog items. Also with very low height so it'll fit nicely into a VME-style rack.

A blocking oscillator is dynamite if you can get the magnetics right.

Oh, the posted circuit probably suffers from the open base. It reduces Q2 breakdown voltage and allows a lot of Miller current to get multiplied by beta and make more losses. Overall, it's a D-.

I like to use a TinyLogic gate as a schmitt oscillator and drive a transistor or fet in fixed-width mode.

This one is open-loop, but it's not hard to add feedback.

ftp://66.117.156.8/Z206.pdf

John