Basic Op Amp Design in Cadence - Feedback Phase Margin

[min2209]

Hi all,

I've been trying to design a CMOS op amp using Cadence to achieve 45 degrees phase margin and a gain of 1500 Vout / Vin at 100KHz. The circuit is as follows:



This circuit achieves the appropriate amplification, but to find out whether or not it satisfies the phase margin requirement, I need to find the loop gain of the feedback path. I've been trying to use the iprobe element in the analoglib, but it hasn't been giving me useful results (it says the phase margin and gain both start at <0).

Does anyone have any ideas?

Thanks!!!
Min

 

[nitishn5]

I think you would have to put some feedback to do a loop gain analysis. You amp is now in open loop. Give a unity gain feedback and try.

 

[min2209]

I think you would have to put some feedback to do a loop gain analysis. You amp is now in open loop. Give a unity gain feedback and try.

Thanks!

So what you mean is that feedback in the second stage is not considered to be the actual feedback path, right? In other words the circuit we have right now is only the forward gain circuit?

We were actually a little unsure about what it means to achieve 45 degrees phase margin. We were unsure whether it refers to the feedback in the second stage or the entire thing.

Thanks again!

 

[mtwieg]

Obviously to measure feedback you have to define a feedback path. Normally for op amps it's measured when configured as a unity gain buffer, though not always.

So to measure phase margin, connect the op amp for unity gain, then just measure the frequency response from input to output. That will give you your PM, GBW, etc.

 

[crutschow]

If you want to simulate the typical open loop gain/phase plot as shown on op amp data sheets you can configure the op amp as a follower with a low resistance feedback from output to the minus input (say 1kΩ) to establish the DC operating point without saturating the op amp. Then add a very large capacitor (say 1000F) from the minus input to ground, which will give the open loop response to well below 1Hz.

Edit: From that you can observe the phase versus frequency. The phase margin is measured when the gain goes to 0db. It's the difference between 180 degrees and the value measured at that point.