Designer’s Guide Community

Neha Bhargava

Guest

cmfb stabilization
Dec 19^th, 2003, 3:09am

Hi ...

I've designed a fully differential folded cacode opamp with continuous time cmfb.The cmfb compares the output common mode with reference and applies correction to the bias of tail current source of input differential pair.

I am not able to stabilize the cmfb loop.It gives high frequency oscillations 100MHz.

Can anyone tell what are the compensation tech. used?

Thankyou so much
Neha

Back to top

IP Logged

Paul

Community Fellow

Offline

Posts: 351
Switzerland

Re: cmfb stabilization
Reply #1 - Dec 21^st, 2003, 10:44am

Hi Neha,

as for the differential path, you should check the phase margin of your common-mode amplifier through AC simulations. What margin do you see?

I would also suggest to apply the feed-back not to the tail current, but to the current sources in the output branches. It is preferrable not to move around the source of the differential pair.

Paul

Back to top

IP Logged

b_neha

New Member

Offline

Posts: 3
India

Re: cmfb stabilization
Reply #2 - Dec 21^st, 2003, 9:41pm

Hi Paul,

If Acms is gain in sense circuit and Acmc is gain in common mode control, the forward gain in CMFB loop will be Acms*Acmc.

The lf gain for this is 85dB, and phase=180 degrees.
The phase decreases from 180 degrees to negative values at high freq.
The freq at which phase is 0 deg. while gain is still >1 is 2MHz.
But I'm getting oscillations at 84 MHz.

Thanks
Neha

Back to top

IP Logged

Paul

Community Fellow

Offline

Posts: 351
Switzerland

Re: cmfb stabilization
Reply #3 - Dec 25^th, 2003, 1:51pm

Neha,

having a gain>1 at 0 deg phase shift is a bad sign for stability. But it is not yet clear why your oscillations occur at much higher frequencies. Can you give more informations on the input signals of your amplifier in this transient simulation (differential input signal, common-mode input level, output load, feed-back type), please?

Paul

Back to top

IP Logged

rajath

New Member

Offline

Posts: 9

Re: cmfb stabilization
Reply #4 - Feb 2^nd, 2004, 4:57pm

Neha,
From your post it's obvious that the Phase margin is
less than zero at 0dB gain, which is an obvious sign of instability. A phase margin of at least 45 degrees is needed for a stable transient response. If your load
and process variations are tightly controlled you can get away with a PM of around 20-30 degrees.
I'm still not sure about one thing though: What frequency will a circuit having a negative phase margin oscillate at? Is it the unity gain frequency?

rajath

Back to top

IP Logged