Hunting down opamp oscillation

[iampoor]

Hi

I am currently working on a basic baxandall eq project. Of course, the circuit is oscillating above 20khz 

What is your usual process for working out oscillation issues?  I think the problem is my PCB layout........but I am wondering what your methodology usually looks like when trying to solve issues like this? What do you usually try and in what order?

Thanks!

[LA7SJA]

Too little information fore someone to give a good answer.
But you can try reading this http://www.linear.com/docs/45597

Johan-Fredrik

[iampoor]

Too little information fore someone to give a good answer.
But you can try reading this http://www.linear.com/docs/45597

Johan-Fredrik

Maybe my question isn't clear, Im wondering what people initially check when they notice opamp oscillation

Thanks for the article link

[T3sl4co1l]

"Your circuit is bad, and you should feel bad"? 

Oscillation is either a topological problem (e.g., you put feedback to +in when you meant -in), or a feedback problem (too much junk in the feedback loop and it's phase shifted to hell).  Either way, you have to start with the circuit.

There isn't anything general that can really be said, aside from recitations of Nyquist stability criterion.  As for what's violating the criterion... you'd have to look.

Don't forget to inspect everything related.  Supply pins don't have infinite PSRR at just any frequency.  They are inputs, just not very good ones!  They are also quite good outputs, since any current drawn from the output pin is also drawn from +V or -V.  For both these reasons, bypass caps are a good idea.

Don't ignore the reference voltage.  Usually this is GND.  Sometimes it's a floating level between -V and +V.  Sometimes it's explicit in the circuit.  Whatever it is, it needs good bypass too, otherwise all the impedances driving it give a place for AC signals to roam, and you get feedback loops between stages.

Tim

[iampoor]

"Your circuit is bad, and you should feel bad"? 

Oscillation is either a topological problem (e.g., you put feedback to +in when you meant -in), or a feedback problem (too much junk in the feedback loop and it's phase shifted to hell).  Either way, you have to start with the circuit.

There isn't anything general that can really be said, aside from recitations of Nyquist stability criterion.  As for what's violating the criterion... you'd have to look.

Don't forget to inspect everything related.  Supply pins don't have infinite PSRR at just any frequency.  They are inputs, just not very good ones!  They are also quite good outputs, since any current drawn from the output pin is also drawn from +V or -V.  For both these reasons, bypass caps are a good idea.

Don't ignore the reference voltage.  Usually this is GND.  Sometimes it's a floating level between -V and +V.  Sometimes it's explicit in the circuit.  Whatever it is, it needs good bypass too, otherwise all the impedances driving it give a place for AC signals to roam, and you get feedback loops between stages.

Tim

Hey Tim, thanks for the informative post as always! 

Im fairly positive (hah get it!  ) that the issue is in the feedback loop. My circuit routing was pretty awful! In fact....its almost embarrassing to post! Phase shift is of particular concern in active equalizer design right?

Love the points about the supply pins, I added some decoupling, but they look pretty good on the scope. The ref voltage Is a virtual earth, just two 10k resistors, a 10uf and a .1uf bypass cap. I have noticed oscillation on the reference point before, but it seems pretty clean again. Im thinking my routing is just causing some brutal phase shifting. I need to redo it anyways as the layout was lost in a hard drive crash..