Open Loop response, Gain Margin. PM is 65+degrees

Hello All,

I am designing a three stage amplifer NMF and am running into Gain Margin issues. I can't find anywhere in books or on the web how much is a good Gain Margin. Everyone always just talking about PM which I totally understand. Does anyone have any papers or places to look?

My Phase margin in 65+ but I am still getting ringing in the time domain. I am seeing a Gain Margin greater then -9dB solves my problems.

Is there a rule of thumb? is -9dB the min? What do people usually design with?

Johnk

A phase margin of about 65 deg looks good and certainly is sufficient.
However, a value of 65 deg should cause only minor ringing in the step reponse, or did you measure/simulate the PM not correctly?

Normally, the lower limit for the gain margin GM is to be selected according to uncertainties within your circuit. When, let´s say, the loop gain uncertainty is 6 dB on would perhaps require a minimum GM in the order of 9...12 dB in order to be on the safe side. For the time response the gain margin GM is less important as the PM.

In some cases it makes sense to require as a lower limit a kind of combination between both margins: The Vector margin is defined in the Nyquist plane as the minimum distance between the loop gain Nyquist plot and the "critical" point "-1"

I oncced suffered the similar problem when I designed a two-stage amplifier: the PM is sufficient for about 70 degree, but the ring peek is about 20%.

The reason to my case was that the Qfactor of the complex poles in my design was too high and it reached 1.1. I decreased it to 0.7 and the peak is gone. You may check the complex pole in your design to see if the Qfactor is reasonable.

So there isn't some kinda of rule of thumb for Gain Margin? Any reading on this subject?

Also what do you mean LvW by your loop gain uncertainty? Do you mean over Process/mismatch/temperature?

Also BackerShu, I do agree it is related to the Qfactor or peaking of the complex poles. If you push for two high of PM the peaking increases thus reducing the GM and the ringing in the time domain increase alot! That is why I am looking for some type of GM requirements? Everyone always talking about PM but why not GM. It seems to be just as important!

jgk2004 said:

So there isn't some kinda of rule of thumb for Gain Margin? Any reading on this subject?

Also what do you mean LvW by your loop gain uncertainty? Do you mean over Process/mismatch/temperature?
...........
That is why I am looking for some type of GM requirements? Everyone always talking about PM but why not GM. It seems to be just as important!

Click to expand...

Quote: Chapter 8, "Design with opamps...." from Sergio Franco:

"....GM of 10 dB, which is considered a reasonable margin. By contrast, a circuit has...a GM=3 dB, not much of margin: only a modest increase in the gain because of manufacturing process variations or environmental changes may easily lead to instability".

I think, this answers most of your questions (gain uncertainty,...).

And, as I have mentioned already, th e GM by far is not so important as the PM.
For example, there are systems with GM=0 or even negative (!) which nevertheless, are stable !