New amp results

[CBS240]

Hello amp folks!

I have been working on a SS amp 'prototype' and finally get to where I can see some results. Most of the amp circuit is still laid out on a bread board, which may impede some high freq response, but that will be tested later. There are 4 of these boards, here are two, 1 per channel. Sometime in the somewhat near future, I will invest in some actual PCB's instead of this Radio Shack junk. I guess it works for now though...


BTW, this amp has an input Z nearly 100K and a low freq. rolloff around 14Hz. The feedback loop consists of a 2.2M resistor to the gate of -input transistor, and 27K in series with 1uF to GND, this gives a gain of around 80. Oddly when the gain is set lower, the amp isn't as fast and becomes more unstable.

[CBS240]

The amp drives +/- 16V into 4 Ohm at clipping with only a 2A transformer. This is driving a 4 Ohm resistor || 10 nf + 22 Ohm, to 2 Vp at 300Khz....

[CBS240]

Same load, 2 Vp at 200Khz...

[CBS240]

The square wave on the bottom is the input, and the top is the output at 5 Vp accross the same load...




BTW, this is at 50KHz.

[CBS240]

Now this square wave is 5 Vp at 10KHz accross same load, input on the bottom, output on the top...

[CBS240]

OK...last one...

This is 100KHz across same load at 15 Vp....







I think I may need some faster output transistors with less capacitances than these D44 & D45 I'm using here. Does anyone have any recomendations?

[ilimzn]

Your 200 and 300kHz shots show obvious signs of oscilation. Why are you using feedback resistors in the megaohm range? For FET input, bias current is non-existent, so you could just as well use nearly any practical value, as long as you kept it fairly low, so it does not make a filter with the (possibly substantial) input capacitance at too low a frequency, which only increases chances for instability!

[MikeB]

Ilimzn is right, the 1st 3 scope shots shows oscillation, the 50khz squarewave shows that these seem to be caused by slewing or anytime the error voltage is big enough.
You will need to finetune your feedback compensation, the most delicate work when designing an amp with gnfb.

As i mentioned already quite often, an amp not oscillating at idle does not mean that it is stable.

Mike

[CBS240]

Thanks for the replys guys. The values of nfb can be reduced, I only did this so that the DC blocking cap could be 1uF during testing. Obviously, this does not have to be so. They don't have to be that small because the input capacitance of the input FET's is around 3.5pF, but your right, 2.2M is too large. I'm sure this is reason for an occilation that is referenced to the input gain pot setting.
The occilation doesn't occur near as bad when the load resistor is removed. Of course an amp with no load is kinda worthless.

http://www.fairchildsemi.com/ds/KS/KSK595H.pdf

There is only 2 stages plus a darlington output in the nfb loop. The first is a cascoded j-fet diff, the second is a cascode BJT diff, biased by a Thompson mirror. Figuring out the compensation is certainly a task. There may be quite a bit of this being caused by the input filter/feedback. I will test this for any changes.

Thanks

[CBS240]

Hi

After some tweeking on the nfb loop impeadance and the compensation, there is some improvement. The breadboard probably affects the compensation scheme. I had to add a 10 Ohm resistor in series with 100nF to the output to make the amp stable with a speaker load, but it is stable on the 4 Ohm load resistor without the capacitor load, which seems more like typical amp behavior. However, this test was done across the 4 Ohm resistor || 10 Ohm + 100nF.



10 Vp @ 10KHz....