How imp is CMFB for fully diff OTA

Hi all,
I want to know if its possible to make a fully differential folded cascode to work without common mode feedback ckt, I mean, I just cannot get to DC bias the OTA without the CMFB, it either goes to rail or gnd at the output without it, is it supposed to do so or is it s'thing wrong with my bias? as I am reading someones paper, and he designed the full OTA then added the CMFB circuit to the simulation. please tell me your views,
thanks,

Hi,
I think u should be able to have your amp bias point set without the CMFB.
One of the purpose of the CMFB is to counter the mismatch effect in real circuit and keep the bias point constant. In simulation, you do not encounter this effect. If you set the size of transistors and DC voltage in the 2 branches to be the same you will have the same DC point.

- Check all the devices to find the one that are off.
- What kind of bias circuit do you use?

Good luck

Really!! have you tried it and it worked without CMFB? I have wasted a lot of time on it, so wanna make sure before I give it another shot
I have a high swing bias(with drain tapped to the gate of the bottom nmos), but I tried everything, even running a multisweep to get the values of xtors to bias them in sat, but if the size or gate V changed just a little bit they went into linear, so I presumed its not possible to make them work without CMFB,and it does work with it but my CMFB reduces the BW. I also read in a book that the impedance at the output node is high and its hard to keep voltage fixed at the output node for a little mismatch will throw the output to Vdd or Gnd, thats why I assumed i def. need to use CMFB, but I would like to simulate the ckt without the CMFB first and then with CMFB, but I just cant get it to work, please throw some more light no whether its possible or not, thankx

Well to be honnest i never check without the CMFB but the bias point of the CMFB and normal bias are very close, so I assume it might be ok.
Anyhow, what I do when designing is to first use an ideal CMFB (verilog model) and size the amp to get the desired noise, BW and gain. After i size the the CMFB and check I still got the same perf. I usually don't have trouble, unless the loop become unstable. At the case I add compensation cap.
So maybe loop stability is your problem. Have you checked it? Put a pulse on the ref voltage of the CMFB and check the amp output cm voltage. It should settle without ringing but not be too slow.