Very strange during simulating loop-gain

[mike_bihan]

I've come up with a strange issue.

When simulating a OTA with feedback, I simulated the loop gain by cutting the loop at OTA input.

In almost all corners, the phase margin will be close to 70degree. However, in fast corner and hot temperature, the loop DC gain is positive. However, transient simulation shows that there is no oscillation.

Any clue and suggestion?

 

[Vamsi Mocherla]

Dear Mike,

When you said that your loop Phase Margin is about 70 degrees, you should not be expecting a peak in your transient response. Can you rephrase the question so that I get the clear point?

 

[mike_bihan]

Transient is fine.
But I do see in some corners, the DC loop gain is positive.

 

[Vamsi Mocherla]

Well, loop gain is Aβ. As long as A is high, you can get a positive loop gain. It is when the loop gain crosses the zero line, we measure the phase margin.

 

[mike_bihan]

Well, loop gain is Aβ. As long as A is high, you can get a positive loop gain. It is when the loop gain crosses the zero line, we measure the phase margin.

No. If the loop gain is positive as DC, then any disturb in the loop will result in amplfied loop response. Start to oscillate.

 

[mikechiang]

would you post your .ac simulation figure?

 

[paladinzlp]

I think your feedback is negative.
If that is positive , the circuit must be oscillate!
Can you show your schematic figure?

 

[mohsenruz]

High loop gain by itself is not the condition of oscilation, dont forget that the overal system transfer function is equal to A/(1+AB), so when AB>>1 we can say that overall system transfer function is 1/B.
the condition of oscilation is magnitude equal 1 and phase equal 360 (or 180 if you dont consider the sign of feedback).
means that any oscilation regenerate itself.