The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Compensating with Miller Cap - which loop to break
https://designers-guide.org/forum/YaBB.pl?num=1395922515

Message started by scientificstick on Mar 27^th, 2014, 5:15am

Title: Compensating with Miller Cap - which loop to break
Post by scientificstick on Mar 27^th, 2014, 5:15am

I have a small question.

Imagine a 2-stage op-amp, with a miller compensation cap between the first and second (output) stage.

Everybody knows how to check the loop gain. In Cadence, with an stb analysis, you simply add an iprobe (to break the AC loop) and plot the loop-gain and phase margin. When I see this done in books, typically only the external loop (op-amp output to op-amp input) is broken. But the miller cap and the second stage makes an internal feedback loop as well. Why don't we check this? For example, why not break the loop at the output of the second stage BEFORE the miller capacitor connection?

I tried this, but then I measured much lower phase margins than when I took the conventional approach of only breaking the external loop. Intuitively, I think this is because the compensation capacitor doesn't see the Miller amplification in my case. Correct? But how much PM is sufficient for this case? After all, I understand EVERY loop must be stable, no?

So my question is - why isn't it common to break both loops? Especially for "indirect compensation", such as described in Baker, where the compensation capacitor goes to the cascode source, the compensation loop could also have more than 1 pole.

Any insight would be useful. Thanks

Title: Re: Compensating with Miller Cap - which loop to break
Post by wave on Mar 27^th, 2014, 2:44pm

Yes, every loop must be stable.

Picking the critical loops is not always easy. I often see people breaking loops that are at a poor place -- such as a TEE or between two intersecting loops.

In 2 stage miller amp, the last stage has loop with the cap - yes.
Are there two stages in THAT loop? (no). Are you likely to get 180 degrees shift (think ~ 90 deg per stage) ? Gain > 1 ? Likely to meet all Barkhausen criteria?
--- are you simulating an independent loop, or an intersecting one?

I've seen some people sim a diode connected current mirror. A high frequency loop yes -- but they probably have too much time on their hands.
I'm saying experience people take calculated risks rather than use a blanket approach to everything.

:)

Title: Re: Compensating with Miller Cap - which loop to break
Post by scientificstick on Mar 28^th, 2014, 1:20am

Thank you very much for your thoughtful reply.

Yes I agree with what you said. For a simple miller compensation, there is little chance of instability around the internal loop itself. After all, it doesn't have enough poles.

My question is - if I break the loop at the output of the second stage (before the miller cap connection), then I am breaking both loops simultaneously. Perhaps this is what you meant by "intersection". In the stability analysis I see a drastically reduced phase margin. How can I interpret the meaning of this?

For example, if I compensate the op-amp like normal (breaking on the external loop), I might design it for a PM of 75 degrees. But then, if I break it as I suggested above, that same op-amp might show only 25 degrees - which makes me nervous. How can I explain it.

Thanks again for your reply.

Title: Re: Compensating with Miller Cap - which loop to break
Post by Lex on Mar 28^th, 2014, 9:25am

The popular 'phase margin' approach only works well if you have some place in the loop that you can safely break. It might happen that your circuit doesn't have such a place.

When this makes you nervous, you should do a (closed loop) transient analysis with a step response and check for ringing/overshoot/etc.

Title: Re: Compensating with Miller Cap - which loop to break
Post by loose-electron on Apr 1^st, 2014, 12:44pm

this has been discussed here many times before, suggest you search the site.