The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> CMFB of OTA
https://designers-guide.org/forum/YaBB.pl?num=1351604397

Message started by urian on Oct 30^th, 2012, 6:39am

Title: CMFB of OTA
Post by urian on Oct 30^th, 2012, 6:39am

Hi,there.
Recently I am designing a OTA used in a front-end S/H for pipeline ADC. The OTA circuit is shown below, and I dont know how exactly the CMFB is working. The VDD is 3V and CM voltage of input and output are both 1.4v, while using SC CMFB. When simulating, I find that the output CM voltage is affected by other transistors parameters especially the M3 and M4 even with CMFB is working! In other words, the output CM voltage is hard to adjust to the desired value 1.4v, and is susceptible by other transistors with CMFB. I dont know why the CM voltage can not be controlled by CMFB. The DC gain of differential is about 60 dB and the first stage has a gain about 2, so I think the CMFB loop gain is large enough to force deep feedback.

Regards
urian

Title: Re: CMFB of OTA
Post by AnalogDE on Oct 30^th, 2012, 5:06pm

What does your CMFB circuit look like?

Title: Re: CMFB of OTA
Post by urian on Oct 30^th, 2012, 9:23pm

AnalogDE wrote on Oct 30^th, 2012, 5:06pm:

What does your CMFB circuit look like?

Hi,AnalogDE, the CMFB is a traditional SC CMFB.

Title: Re: CMFB of OTA
Post by Dan Clement on Oct 31^st, 2012, 4:26am

Shouldn't m3 and m4 be pmos?

Title: Re: CMFB of OTA
Post by urian on Oct 31^st, 2012, 7:34am

Dan Clement wrote on Oct 31^st, 2012, 4:26am:

Shouldn't m3 and m4 be pmos?

Hi,It's NMOS for sure.

Title: Re: CMFB of OTA
Post by Dan Clement on Oct 31^st, 2012, 7:40pm

I've never seen this topology before. From what I see I don't think there is a chance for this circuit to act as an amplifier.

Can you explain how it is supposed to work? It doesn't make sense to me.

Title: Re: CMFB of OTA
Post by urian on Oct 31^st, 2012, 9:11pm

Hi,it's from the JSSC paper " A 10b 20M 35mW Pipeline A/D Converter" by Cho,Mar,1995

Title: Re: CMFB of OTA
Post by Dan Clement on Nov 1^st, 2012, 5:40am

I will have to look into that paper.

After a good night of sleep I can now see it as an amp. The gain of the first stage will be quite low.

The signal swing of the first stage will be severely limited by the cmfb as you can't pull the drains of the input transistors any higher than the cmfb node minusvavthreshold voltage.

You are probably having a startup issue where the input transistors are being clamped by the cmfb source followers or you are having a swing issue and losing gain.

You will need to level shift the cmfb to a high enough voltage to avoid this.

Title: Re: CMFB of OTA
Post by urian on Nov 1^st, 2012, 9:26pm

Yes, the gain of first stage is low, between 1-2, the drain of input transistor is around 800mv with the bias, but the output will change largely with SF transistors parameters.

Title: Re: CMFB of OTA
Post by Lex on Nov 5^th, 2012, 4:25am

Are the problems gone with a higher common mode voltage? Then I guess you know the answer to your problems.

Title: Re: CMFB of OTA
Post by boe on Nov 5^th, 2012, 4:30am

Lex wrote on Nov 5^th, 2012, 4:25am:

Are the problems gone with a higher common mode voltage? Then I guess you know the answer to your problems.

I"d try lower (input) common mode voltage...
- B O E

Title: Re: CMFB of OTA
Post by urian on Nov 5^th, 2012, 7:38am

Lex wrote on Nov 5^th, 2012, 4:25am:

Are the problems gone with a higher common mode voltage? Then I guess you know the answer to your problems.

Do you mean the output CM voltage? If so, I am afraid I cant afford this as the CM voltage is determined by BIAS and it tends to be smaller. Actually, I find that with this cmfb, the DC OP is very important to guarantee that the CMFB loop has enough gain.

Title: Re: CMFB of OTA
Post by boe on Nov 5^th, 2012, 11:08am

urian wrote on Nov 5^th, 2012, 7:38am:

...
Do you mean the output CM voltage? If so, I am afraid I cant afford this as the CM voltage is determined by BIAS and it tends to be smaller.

I believe Lex meant changing it to test the circuit behavior.

Quote:

Actually, I find that with this cmfb, the DC OP is very important ...

As I'd expect given the topology.

Quote:

... to guarantee that the CMFB loop has enough gain.

So what is your question?

- B O E

Title: Re: CMFB of OTA
Post by Lex on Nov 6^th, 2012, 12:03am

boe wrote on Nov 5^th, 2012, 4:30am:

Lex wrote on Nov 5^th, 2012, 4:25am:

Are the problems gone with a higher common mode voltage? Then I guess you know the answer to your problems.

I"d try lower (input) common mode voltage...
- B O E

I think we're on the same page. ;)

Title: Re: CMFB of OTA
Post by urian on Nov 6^th, 2012, 2:19am

Sorry I made a mistake.When I re-check the circuit, I find that it is because the DC OP of output CM voltage will change largely with SF W/L, and this will put great pressure on CMFB.

Title: Re: CMFB of OTA
Post by prcken on Nov 9^th, 2012, 9:25pm

what's your sampling rate? do you use pss and pstb for stability analysis?

Title: Re: CMFB of OTA
Post by urian on Nov 14^th, 2012, 6:10am

prcken wrote on Nov 9^th, 2012, 9:25pm:

what's your sampling rate? do you use pss and pstb for stability analysis?

The sampling rate is 128M. I dont know how to use pss and pac or pstb, so I just checked the transient.