Opamp problem ... oscillation!

DSV · February 10, 2009, 09:52:56 AM

Hi all!

I am currently designing a hi-gain distortion. It is basically an opamp version of my BSIAB, which I need for a big project. The schematic of the first 2 gain stages follows:

The values are roughly the ones I am using on my current breadboard prototype (there are some slight variations because of the tests). Side note: the power supply is symmetrical, +/- 9V.

The first three opamps (of a TL074) are working nicely, and noise is acceptable. However, once I use the fourth one with a relatively high gain setting (>10), a high frequency oscillation (roughly 6-8kHz) appears. I checked the connections, relocated the components, replaced the pot with a fixed resistor, changed the value of the cap in the feedback loop ...

I really haven't got a clue about the problem. Does anyone have an idea?

Thanks,

DSV

Sir H C · February 10, 2009, 10:03:27 AM

Do you have power supply decoupling caps? Right by the ICs, usually 10-100 nF maybe a 10uF also.

How is the grounding? Star grounds are a must, expecially with the common connections for the positive inputs to the op-amps. High gain circuits are a pain to keep oscillation free.

DSV · February 10, 2009, 12:39:22 PM

My power supply unit is a separate 18V, 100mA one. I get the +/-9V through a virtual ground, buffered by an opamp (NE5532). The correspondig schematic is at page 5 of this document.
http://www.techniguitare.com/projets/ChaosPreamp/Preamp-Schems-V1.1-En.pdf

I have got a sort of star wiring (no GND loops), and adding decoupling caps does not have any effect.

I also tested opamps 3 and 4 separately, and there is no oscillation. Only when the 4 opamps are used together, oscillation arises ...

Sir H C · February 10, 2009, 01:34:25 PM

Total gain in that circuit is quite high, 10,000 if I am calculating right, so a very small pertibation will cause problems, you might have to have separate supplies for the two gains (rc off the common supply to kill coupling), and two virtual grounds as tiny amounts of signal at the front will cause the rest to scream.

DSV · February 10, 2009, 04:37:04 PM

Gain is indeed 100*120 = 12000. It may even go up to 20000 if I implement the boost control I'm planning to use. However this gain is only for high frequencies (above 1 - 2 kHz), since there is a lot of bass filtering in the 2 stages.

Could the slight perturbation also be the cause of the gating effects I sometimes experience? And would using two separate opamps (TL072) help? In fact I'd like to keep only one power supply, this time with "true" +/-7.5V, regulated, for a 4 channel project.

This bugs me a bit, because with the BSIAB distortion section I have done 3-4 layouts, and even with higher gains (roughly 400 and 300 for stages 1 and 2 respectively, in total 120000

), it never oscillated ...

slacker · February 10, 2009, 05:30:32 PM

I could be because with 4 sections the output is in phase with the input, where as with just the first 3 stages it's out of phase. On breadboards where you have wires running everywhere, this can cause oscillation, try making sure your in and output wires are as far apart as possible and see if that helps. I'd also make C14 a lot bigger to see if that cuts the oscillation.

It probably doesn't need it if you're using an opamp buffer to provide your virtual ground, but try sticking a big cap across it.

Sir H C · February 10, 2009, 10:48:05 PM

By two supplies, I mean have one come in and then split it with RCs to each. So you are using one quad op-amp? If so, I would go to two separate ones. Realize, the BSIAB is discrete gain and little negative feedback, the op-amps all have insane gain and lots of negative feedback. This makes stability a lot more of a pain. What kind of op-amps are you using?

DSV · February 11, 2009, 05:57:11 AM

A TL074. I tried with an LF353 for the last stage, a bit further from the others, and it improved a little (oscillation is weaker, and gating effects also tend to disappear). Also by increasing C14, only noise remains, but this tends to cut too much high frequency content. Anyway, your advice is proving useful indeed!

I will try to simplify the schematic by using only 2 op-amp stages (probably non inverting). It would be like the Sansamp Classic (the 1st 2 stages have got a gain of 1000, and the circuit seems pretty stable).

Quote from: Sir H C on February 10, 2009, 10:48:05 PM
Realize, the BSIAB is discrete gain and little negative feedback

By the way, what does "discrete gain" mean?

Sir H C · February 11, 2009, 10:09:48 AM

Discrete transistors, not a chip. Often seen when talking about pro audio gear with "discrete op-amps" where you roll your own with separate transistors and resistors.

Sir H C · February 11, 2009, 02:35:06 PM

One of my favorite apps notes on high speed amps, useful here on how to keep things stable.

http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1028,P1219,D4138

Jim Williams rules.

DSV · February 12, 2009, 02:28:10 PM

Thanks for the link, I'll read the document carefully.

Anyway I discovered that if the total gain does not exceed 80000-100000, oscillations disappear. I think I'll design my gain stages and set the clipping threshold accordingly.

News:

Opamp problem ... oscillation!