Stage output, driver oscillating

[Rafael L]

I tried to triple darligton output, but depends on the driver has oscillation with capacitive load (330nF parallel 8 ohm)
The margin phase of the circuit is ~80° (resistive load).
In square wave the most has oscillation was with the transistors MJE15032/33, with MJE340/350 and BD130/140 the response was better.

This is the circuit output:


Driver BD139/140 (330nF parallel 8 ohm).
[img=http://img21.imageshack.us/img21/5844/gr1c.gif]

Driver MJE15032/33 (330nF parallel 8 ohm).
[img=http://img21.imageshack.us/img21/8072/gr2m.gif]

What is the best way to improve the oscillation

Thanks

[Rafael L]

I was simulating Zobel network without, noticed that the capacitance(330n) is very high without Zobel network.
My question is:
Which value of the maximum capacitive load, circuit does not oscillate without Zobel?
would be necessary base stopper in driver? (in simulator has not did difference)

Regards


Square wave:

[Mr Evil]

Try including one or more of the driver stages in the Miller loop, as in the attached schematic.

[jaycee]

way too much current through the predriver stage
also, 2n5401 etc is not going to be good enough. you want a high voltage high speed medium power (TO-126 ideally) device here such as 2SA1381/2SC3503

[Rafael L]

Originally posted by jaycee
way too much current through the predriver stage
also, 2n5401 etc is not going to be good enough. you want a high voltage high speed medium power (TO-126 ideally) device here such as 2SA1381/2SC3503

2n5401/5551 bias is 4 mA, these transistors are dissipating 135 mW. I do not think it is high for these transistors.
But I am thinking of lowering the bias in driver (22mA)

Originally posted by Mr Evil
Try including one or more of the driver stages in the Miller loop, as in the attached schematic.

I do not use this compensation miller, in my circuit is inverting diff pair.
That short-circuit capacitor in multiplier-ratio (Attached fig) , I should use ?

Thanks

[jaycee]

you still need miller compensation even with (especially with) this capacitor.

[wg_ski]

+1. The main effect of the miller cap is to install a dominant pole. Lowering the pole frequency naturally created by Cob of the vas has the side effect of pushing the poles created by the output followers higher in frequency. With a triple, that's not a luxury - it's a necessity as each stage has some phase shift. And of course, a capacitive load adds yet still one more pole to fight with.

[Rafael L]

Originally posted by jaycee
you still need miller compensation even with (especially with) this capacitor.

Simulated with this capacitor, no effect with low capacitance.
But for values higher of capacitance, makes some difference

Originally posted by wg_ski
+1. The main effect of the miller cap is to install a dominant pole. Lowering the pole frequency naturally created by Cob of the vas has the side effect of pushing the poles created by the output followers higher in frequency. With a triple, that's not a luxury - it's a necessity as each stage has some phase shift. And of course, a capacitive load adds yet still one more pole to fight with.

Which capacitor you say, Cdom or short-circuit Ratio?

Attached more simulations
Load 8 ohms/20nF
Frequency square 20Khz.
Base stopper output increased to 5 ohms (if I increased more, distortion increases )

[Mr Evil]

Originally posted by Rafael.luc
...Which capacitor you say, Cdom or short-circuit Ratio?..

Miller capacitor = Cdom.

You say you don't have one... is that because you are using some other type of frequency compensation, or is it a low-feedback design or something else that lets you get away without one?

Edit: Don't forget an output inductor!

[Rafael L]

Originally posted by Mr Evil
[B]
Miller capacitor = Cdom.

You say you don't have one... is that because you are using some other type of frequency compensation, or is it a low-feedback design or something else that lets you get away without one?

I tried cdom (shunt base-emitter VAS), not did difference for oscillation square wave, the capacitor inverting diff par does compensation Miller (better for the topology which I use)

Originally posted by Mr Evil

Don't forget an output inductor!

Yes, inductor solve the oscillation. But I want to know, which for value capacitive load I have on the output, 10nF/8R would be a good value, before adding the inductor ?
-Oscillation off, with inductor of 3uH in output.

Fig. Loop gain, phase margin with: Driver 2S-4793/1837, output 2S-1943/5200