What Oscillator is in This PS?

Excellent!

And, R5 controls the discharge period.

What a relief. I have this puppy purring like a kitten now at

The voltage is a little low. I might have to use the audio transformer. The article did say that was better for low voltage supplies.

Thank you everybody :)

Fred

Wow, Q2 is on for about 80 milliseconds! Since the inductor current flattens out in the first millisecond roughly, the rest of the on time is a pure waste of power. Efficiency must be a percent or two.

Horrible, horrible circuit.

John

Somebody else did mention that there was a lack of anything to expeditiously turn off Q2. There is a feedback circuit that turn prevents Q1 from turning on Q2 went the output is high enough, which is most of the time. Quiescent current draw from the batter should be

Actually now that I got it running well with a little help, Q2 is only on for 1.8ms.

In series. Q2 does not turn off expeditiously; it's inefficient.

James Arthur

You don't often see inductor current that's a square wave. Me-thinks that something in this simulation stinks.

RL

The audio transformer was used in forward and flyback, which has a number of advantages over the circuit as you've adapted it.

Q2 now sees an enormously larger voltage swing than in the original circuit, for one thing.

Here's an ASCII of the original:

C2 +3v >--o-----------. || 50nF/1kV | | .---------||-----. .-. | | || | R1 | | | T1 | | 1k | | | audio | 3 x 1n4007 | '-' | xfrmr | D2 D4 | D5 +500v | '--. ,-o-->|-o---o-->|--o->|---o--->

| o )|( | | | | D1 )|( | | | | 1n4007 )|( o | |C3 C4| .----------o----->|----o--' '-. | --- 50nF/ --- | | | | | --- 1kV --- | | o--------' | | | | .-. | D3 - | | | | | R4 | 1n914 ^ === === .-. | | 1k | (zener)| GND GND | | R2 '-' | (selected)| | |100k | | | '-' | | | | .-----o-----------)--------. | | | | | | | Charles Wenzel's | |< Q1 | | | | Geiger Supply o--| 2n4403| | \\| | circa 2002 | |\\ | ___ |/ Q2 |---' | '-----)--|___|--| MPSA42 .. | 2n4401 | |R3 | 1.8k | | | |150k --- | | '-' --- C1 === === | | 100uF GND GND | | === === GND GND

(created by AACircuit v1.28.4 beta 13/12/04

Wenzel had an improved version where Q3 was an emitter-follower from +3v to Q1(b), killing the oscillator to regulate the output voltage, but saving a couple mA over this version.

James Arthur

it looks like a type of relaxation oscillator, I don't know if that specific topology has a name.

Are you modeling it in LT-Spice ? or is that image from a third-party :)

are you using avalanche breakdown in the 1n5955s for feedback to regulate the output?

C1 seems kind of large, but then, L1 is huge!

I think the simulation's right--the oscillator's "on" time is much too long.

A 1.8k resistor in series with Q2's base has been omitted. That reduced the gain, and allowed Q2 to pop out of saturation when the desired collector current was exceeded. Kind of a poor-man's blocking oscillator.

Leaving out the resistor defeats that function, so the current quickly rises to max, then Q2 holds it there a looong time, wasting lots of power.

Cheers, James Arthur

cycle

regulate

I have the Wenzel page you mentioned. In one of his schematics he omits that the Q2 base resistor.

In any case I've been playing with different value, and it seems to have little effect on the sim. If it gets up to 5k or above the output voltage drops, but below that little change in operation.

The main problem I have noticed is when the choke/transformer flyback the positive pulse travels back through the diode D5 and cause the transistors to parasitically oscillate more or less severely depending on the other componet values during the time C2 is discharging.

I have found the by putting another small cap across the collector and base of Q1 or from the base to ground the paracitic oscillations can be prevented, relaxation operation is smooth, but output is lower than needed.

I'm also thinking I might have bad values for the choke transformer to begin with. It all seems to depend on that first. I have a coupling transformer, I measured it at 12mH and 150 ohms on both pri and sec windings. Those values in the sim fail to work well.

It depends, I can do it either way, and will do it the way that works best. Using the zeners for feedback rather than just voltage limiting, give the added benefit of reduced quiescent current, but at

cycle

regulate

That surprises me, but I'll take your word on it.

Whatever the cause, that "on" pulse is way too long for that inductor. I suspect Wenzel had a much higher inductance transformer, which he was also using in forward mode. That makes a difference.

Oh, that 100uF timing cap should be 10uF. I copied your value, but it's wrong. That explains a lot!

That's in part due to the huge voltage swing you're now producing at Q2(c). That gets coupled through the 1n4007's capacitance, upsetting Q1.

The transformer version would have a lot less dV and dV/dt, so it wouldn't suffer as badly.

You might do better with a signal diode here, if you can find one with a high enough breakdown voltage.

He used an audio interstage coupling transformer. That was before my time...maybe one of the old hands could tell you what winding ratios were common for those, but he wasn't using a 1:1 in this circuit, that's for sure.

Cheers, James Arthur

I found some spes XICON choke in my Mouser catalog. Plugging them in seems to work well with a 9V supply, but still have parasitic oscillation trouble with the 3V version. I had trouble finding any inductance figures for telephone coupling transformer, so I had to measure and figure out what inductance yield 600 ohms impedance.

cycle

caps

regulate

as

R1

Looking through my Mouser catolog, I find a variety of audio transformers with turns ratios of 1:1, over 10:1.

cycle

caps

regulate

as

R1

Here a schematic of the version he used a choke and omitted that base resistor.

Fred

R1

cycle

voltage

caps

regulate

oscillator as

of R1

resistance

Play

Any you know what, it fails to work with out that resistor when using a !0mH choke with 22 ohms DCR :)

Fred

cycle

caps

regulate

as

R1

Very interesting how Q2 has to have a significant base resistor in order to maintain maximum conduction for the full C2 discharge time.

The original author is obviously not to be trusted in conveying his own work accurately. This schematic simply fuses the MPSA18 'regulator'.

Lord knows what other transcription errors are present.

Suggest you just use the 'concept' to construct your own circuit, using basic principals and sensible component values.

RL

Sorry, misread the drawing - no fusing present. Need coffee.....

RL