When we can talk about phase margin ?

[achilles09]

in some papers ,they concerns the phase margin when gain is 0 dB, but when when the loopgain is not come to 0, the phase margin is not enough, is this not important?

 

[keith1200rs]

about phase margin

If the phase margin is OK at 0dB then it will be better when gain is more than 0dB. When gain is less than 0dB then the phase margin is not important - it cannot oscillate with loop gain less than unity.

Keith

 

[LvW]

Re: about phase margin

in some papers ,they concerns the phase margin when gain is 0 dB, but when when the loopgain is not come to 0, the phase margin is not enough, is this not important?

I don't understand the question. Each loop gain drops with rising frequencies to zero dB and less!
Comment to Keith' reply: If the loop gain never is higher than 0 dB the whole feedback makes no sense. Or did I misunderstand something?
LvW

 

[keith1200rs]

Re: about phase margin

Comment to Keith' reply: If the loop gain never is higher than 0 dB the whole feedback makes no sense. Or did I misunderstand something?
LvW

I am assuming from the question that the phase margin is OK at 0dB. When loop gain is >0dB therefore, the phase margin will be higher (unless it has a very peculiar phase response). So the only time where phase margin could be worse is at higher frequencies than the 0dB loop gain. That is what I was trying to say

Keith.

 

[LvW]

Re: about phase margin

.............
When loop gain is >0dB therefore, the phase margin will be higher (unless it has a very peculiar phase response). So the only time where phase margin could be worse is at higher frequencies than the 0dB loop gain. That is what I was trying to say
Keith.

OK, but - for my opinion - this sounds a bit confusing since the phase margin is defined ONLY for the 0 dB point. For loop gains smaller or larger than 0 dB there is no phase "margin".
LvW

 

[keith1200rs]

about phase margin

What would you call the phase "margin" at 20dB then?

The original post was "but when when the loopgain is not come to 0, the phase margin is not enough" so I was replying to that question.

Keith.

 

[LvW]

Re: about phase margin

What would you call the phase "margin" at 20dB then?

The original post was "but when when the loopgain is not come to 0, the phase margin is not enough" so I was replying to that question.

Keith.

Hey Keith,
The term "phase margin" is a defined parameter. It is defined ONLY at a loop gain of 0 dB (like the gain margin is defined at zero deg). Thus, you are not allowed to use the term "phase margin" for gains other than 0 dB.
But you can say, of course, the loop gain phase is 15 deg. away from 0 deg. at a gain value of 20 dB.
But it's not a "margin" - for example, because at 0 dB gain the margin could be even larger (e.g. 30 deg). That's quite normal for lead compensated networks.

Asa far as the original post is concerned, I have mentioned already that I did not understand ("margin is not enough"?)

 

[checkmate]

Re: about phase margin

I normally regard the phase margin as the minimum phase difference from -180 degrees while gain is above 0dB. This will take care of cases whereby the magnitude response crosses the 0dB line multiple times, and when "bumps" are found in the phase response. Such scenarios are possible when LHP zeroes are used for compensation.

 

[LvW]

Re: about phase margin

I normally regard the phase margin as the minimum phase difference from -180 degrees while gain is above 0dB. This will take care of cases whereby the magnitude response crosses the 0dB line multiple times, and when "bumps" are found in the phase response. Such scenarios are possible when LHP zeroes are used for compensation.

Of course, you can do that (for your personal purpose). But I think - in order to avoid misunderstandings - one should not use this term "phase margin" when it is referred to a loop gain value which is not 0 dB (because of its definition!).

The reason is as follows: If the magnitude response crosses the 0 dB line multiple times (as in your example) the simplified Nyquist criterion is not applicable anymore.
And: The definition and the meaning of the stability margin is connected to the simplified Nyquist criterion (as, for example, also the stability criterion derived from the gain slope).