The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Definition and Determination of stability in high-Q filters
https://designers-guide.org/forum/YaBB.pl?num=1385391784

Message started by vivkr on Nov 25^th, 2013, 7:03am

Title: Definition and Determination of stability in high-Q filters
Post by vivkr on Nov 25^th, 2013, 7:03am

I am designing a high-Q analog filter. So, the output response has some ringing by definition and there are going to be high-Q complex conjugate poles.

Now, how does one define stability measures for such a circuit? I am sorry if the question sounds stupid, but there it is.

The one thing I could imagine was comparing the response of the ideal filter (using ideal opamps) to that of the actual realization with real opamps. I anyway need to do that as part of my sensitivity analysis. Here, I could look at the pole/zero locations, even do a Nyquist test by opening various loops one at a time, etc. But I can merely say that the actual transfer function matches the target transfer function with a certain amount of deviation in gain and phase.

However, all this still does not answer my original question viz., definition of stability for a system which is actually expected to ring a bit.

What am I missing?

Thanks
Vivek

Title: Re: Definition and Determination of stability in high-Q filters
Post by Frank Wiedmann on Nov 25^th, 2013, 7:30am

By definition, it's stable as long as it doesn't oscillate by itself. I suggest that you do some worst-case step response simulations (positive and negative steps, maximum step size and slope) for your corner cases and see if your circuit starts to oscillate (or rings more than you can tolerate).

Title: Re: Definition and Determination of stability in high-Q filters
Post by nrk1 on Nov 25^th, 2013, 8:25am

I guess what you are concerned with its stability margin. If you do a pz analysis, you can find the quality factor and natural frequency of all pools. There should be a set of poles close to the desired set in natural frequency and quality factor. The rest should have a low quality factor and high natural frequency compared to the desired set. That would mean that the response would be close to the desired one.