Problems with CMFB stability in a fully differential folded cascode amplifier

[ansu_s]

hi,

I try to design fully differential folded cascode amplifier, but have problems with CMFB stability (see attached circuit). There is too much phase shift in CMFB path (260 degrees!), but I not know where it is from!! And magnitude response of CMFB path is 80dB!! I think because of 40dB gain from cmfb amplifier, and then folded cascode output stage looks like additional gain stage to CMFB (another 40dB), is this right?

In CMFB path, I have pole at folded cascode output (two legs of folded cascode combined in parallel for CMFB path). Also, another pole at output of CMFB amplifier. And third pole at mirror node in CMFB amp. But I can't alter folded cascode pole, so can only adjust CMFB amplifier poles. But I must make CMFB output pole << differential GBW to reduce overall phase shift to make CMFB stable. But I need larger CMFB GBW than this, to respond to fast CM changes! What can I do here??

 

[ansu_s]

test cmfb loop stability

hi,

I think more about this and am confused. With simulation of CMFB amp in CMFB amp testbench (with common mode version of differential output stage at output, just combine in parallel all transistors in diff output stage), CMFB is stable - 70 deg phase margin, single LF pole and HF pole too. But when simulate CMFB in real diff loop, get repeated LF pole so phase goes to 180 deg before CMFB gain drops to zero!! Also, in CMFB test bench, DC loop gain is 40dB but in real diff loop, DC loop gain ~ 80dB. Why is CMFB stable in test bench but appears doubled in real diff loop?

 

[ansu_s]

cmfb　stable simulation

So now I think that with two branches in diff amp, there are two capacitances at diff amp output (with same value, if diff amp symmetric), so this could be reason for repeated pole. But then it is never possible to stabilise the CMFB circuit. Any books recommended for this? Razavi and others give example CMFB circuits, but no discussion about CMFB stability...

 

[M.B.]

cmfb phase start from 0

Hi,

I am attaching a document that i think you can help you.
The CMFB circuit is described in a part of this document.

Good work.

 

[M.B.]

common mode feedback stability two stage

This is another interested document.

 

[blam]

cmfb stability simulation bench

what is the source of the 2 chapters?

 

[gunturikishore]

cmfb amplifier

I think if you are using the two stage amplification in the CMFB amplifier, the same problem of two stage Opamp stability problems are appearing in your circuit. May be your test bench could be missing some capacitances required. Connecting the two differential outputs with two identical capacitors to the CMFB node may solve the problem and do some compensation for the CMFB.

 

[ansu_s]

Re: CMFB stability

thank you gunturikishore! That makes sense for me!

Two stage op-amp, I now use compensation caps to the CMFB bias point, with nulling resistor to cancel zero. DIff mode loop has 80 deg phase margin, common mode feedback loop has 60 degrees phase margin - so both loops stable in small signal. But now I have another problem - in transient simulation, loop has common mode oscillation. Any suggestions for where this oscillation comes from?

 

[PoojanWagh]

Re: CMFB stability

This article/post might be useful:
https://www.circuitdesign.info/blog...ack-loop-using-a-pmos-triode-device-for-cmfb/

 

[gunturikishore]

Re: CMFB stability

Generally nulling resistor compensation will vary with output load. May be that is affecting your feedback loop stability. I do only feedback capacitor without nulling resistor compensation for my circuits.