High voltage AC amplifier (quick, dirty circuit)

[c4757p]

I just thought I'd share this, in case anybody is interested. This isn't particularly good, but it was built quickly and served its purpose (providing a high voltage to compare multimeters on the AC range).

Post here



Yes, I know it's not really "high voltage", but almost 200VRMS (I actually did get above 200V, but it distorted quite a bit) is high for most of us!

[dannyf]

1. the feedback loop can be further simplified: all you need is a simple rectifier to drive the led;
2. unless you care about the fidelity of the output waveform, Q1/Q2 can be biased into class C;

[c4757p]

1. the feedback loop can be further simplified: all you need is a simple rectifier to drive the led;

Very good point!


As for class C - I tried. At some frequencies it worked quite well, and at others it distorted significantly. I don't mind some distortion, but I wanted around 5% between true RMS and average-responding, and it was way too much for that.

[Jay_Diddy_B]

Hi,

You can simplify this to:




I have removed the leveling loop, but added negative feedback so that the circuit has a fixed gain of 10. The phasing of the transformer is important or you will get positive feedback.

The transistors should be rated for the expected dissipation, I just used ones that are included with LTspice.

I have attached the LTspice model if you want to experiment.

Jay_Diddy_B

[dannyf]

You can simplify this to:

Your amplifier has no agc; and the OP's has agc.

[c4757p]

You can simplify this to:

Your amplifier has no agc; and the OP's has agc.

It is equivalent in function, though. What I'm wondering is whether it will work properly once you throw in all the nastiness of a mains transformer, particularly the hysteresis. I didn't want to waste time experimenting with that, since it's just a one-off (and really, I like to keep high voltages off my bench when not necessary). The benefit of the peak-detecting AGC loop is that any distortion/phase offset/anything else will not destabilize it.

One quick change I made to the SPICE file: the transistors' quiescent current is way too high; I added 1 ohm emitter resistors. (I made this mistake in mine; they slowly went into thermal runaway.)

[Jay_Diddy_B]

Hi c4757p,

You are absolutely right about the emitter resistors. It was an oversight on my part.

The transformer is inside the feedback loop. At frequencies less than 1kHz the circuit should work fine with a regular line frequency transformer.

Regards,

Jay_Diddy_B

[c4757p]

The transformer is inside the feedback loop. At frequencies less than 1kHz the circuit should work fine with a regular line frequency transformer.

Ah, OK. I wanted 1kHz and 10kHz!

[Jay_Diddy_B]

Hi,

I think you could try my circuit on your breadboard relatively easily.

Since the load on the circuit is fairly light, I believe that you using the circuit to test meters, the leakage inductance in the transformer will have little effect.

Do you have a meter with a good frequency response like the Fluke 8060A to check the accuracy?

Jay_Diddy_B

[c4757p]

It would be easy to try. I'm not doing it tonight, though - going to bed in a minute! I might play with this a bit more tomorrow, though I don't have much time for it.

I have a couple meters up to the task: Fluke 45 (0.5% to 20 kHz) and HP 3468A (0.5% to 50 kHz).

[SteveH]

Try a small toroidal as the inverter transformer, they have a much wider frequency response.

[megajocke]

Hi,

You can simplify this to:

With that configuration the DC operating point of the opamp output is undefined. You need some kind of DC feedback from before the coupling capacitor and transformer.

You probably need to limit the bandwidth of the feedback loop too. With the shown component values you'd be shooting for 1 MHz which is likely to be troublesome because of the leakage inductance and winding capacitance of the transformer.

[Jay_Diddy_B]

Hi,

With the constructive feedback I have adjust the LTspice model to incorporate necessary changes:




I have attached the revised LTspice model.

Jay_Diddy_B