Designer’s Guide Community

HdrChopper Community Fellow Offline Posts: 493	Re: constant gm bias goes unstable after modification Reply #15 - Nov 06^th, 2008, 2:46pm I agree with you buddypoor. In this particular case however, I think the overall stability could be evaluated by breaking the loop at the node that both loops share. This is a special case when a multi-loop system could be analyzed as a single loop one (the usual way) Regards Tosei
Back to top	Keep it simple IP Logged

Tlaloc

Community Member

Offline

Posts: 81
Dallas, TX

Re: constant gm bias goes unstable after modification
Reply #16 - Nov 6^th, 2008, 2:57pm

buddypoor wrote on Nov 6^th, 2008, 2:08pm:

I like to remind you on the Nyquist stability criterion in its most general form. This criterion allows us to prove the stability of a closed loop system also in case that the open loop is unstable (i. e. it has poles with a positive real part). There are a lot of examples that closing one specific loop can stabilize another loop which is unstable.
Regards

You are quite right. In fact I have designed such a system before, but it had slipped out of my mind. In that circuit, I had two feedback paths, one positive and one negative. I did run stability on both loops since I needed to verify that the negative path always had greater loop gain. Otherwise the positive feedback would become dominant, and I would have an oscillator.

I still think that it would be informative to run the three stability analyzes, though. Even though all are negative feedback, if one is flirting with instability, you should see it.

Back to top

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: constant gm bias goes unstable after modification
Reply #17 - Nov 6^th, 2008, 11:00pm

HdrChopper wrote on Nov 6^th, 2008, 2:46pm:

I agree with you buddypoor.
In this particular case however, I think the overall stability could be evaluated by breaking the loop at the node that both loops share. This is a special case when a multi-loop system could be analyzed as a single loop one (the usual way)
Regards
Tosei

Yes, Tosei, that�s correct.
A simple example is an opamp with positive and negative feedback (of course, the neg. feedback must dominate). In that special case the positive loop simply reduces the negative loop gain and, therefore, both loops must be broken at the same time.

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

Frank Wiedmann Community Fellow Offline Posts: 678 Munich, Germany	Re: constant gm bias goes unstable after modification Reply #18 - Nov 7^th, 2008, 1:02am I agree with both of you. If all loops can be broken at a single path, one should examine the circuit like this. The general criterion for stability of multi-loop circuits is rather complex, see http://www.designers-guide.org/Forum/YaBB.pl?num=1217822985.
Back to top	IP Logged