The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Common mode BW and Differential mode BW
https://designers-guide.org/forum/YaBB.pl?num=1185775161

Message started by Shano on Jul 29^th, 2007, 10:59pm

Title: Common mode BW and Differential mode BW
Post by Shano on Jul 29^th, 2007, 10:59pm

Hello all smart guys:

In a fully differential OPamp design,
should the common-mode-BW(CMFB) be as large as differential-mode-BW?

In my case, I am designing an OP for a S/H.
My target spec of the clock is 40 MHz.
So I set the differential-mode-BW to be about 150MHz around.
My common-mode-BW now is about half of that, and it is 70MHz around.
I am not sure that if I burn too much current in the CMFB or not!

Does anyone enlighten me on this issue kindly?

Thanks a lot!!

Shano

Title: Re: Common mode BW and Differential mode BW
Post by monte78 on Jul 30^th, 2007, 1:12am

I would be interested too in some references or papers about common-mode feedback bandwidth and implementations....anyone can suggest something?

Thanks, monte78

Shano wrote on Jul 29^th, 2007, 10:59pm:

Title: Re: Common mode BW and Differential mode BW
Post by Berti on Jul 30^th, 2007, 10:36pm

Hi Shano and Monte78,

the bandwidth of the CMFB should be as large as the differential-mode BW.
But often this would result in large power consumption for the CMFB (or
big capacitors and reduced diff.BW in case of a SC-CMFB).
Therefore, in most designs the BW of the CMFB is propably smaller than
the diff. BW. However, how large your CMFB BW should be depends on
the application (and how important good common-mode regulation is).
But to design the CMFB BW to be at least 1/2 of the diff. BW is reasonable
trade-off (in my opinion).

Regards

Title: Re: Common mode BW and Differential mode BW
Post by Monkeybad on Aug 1^st, 2007, 9:05pm

In some papers it mentioned the CMFB BW should be at least 50% of the diff closed-loop feedback BW for stability consideration.
I know it uasully set 50% but I don't know why and the detail. Maybe someone can give more explanation.

If you are designing a S/H circuit, maybe you can try to use dynamic SC-CMFB. It uses two caps to sense the CM voltage and feedback to the current source MOS gate to control the CM voltage. It doesn't need other circuit to implement.
So it can save more power. :)

Best Regards

Title: Re: Common mode BW and Differential mode BW
Post by Berti on Aug 3^rd, 2007, 12:22am

Dear Monkeybad,

may you give a reference to those papers you are referring to (CMFB stability
considerations)? I am interested, because to me the relationship
is not obvious.

You are right that SC-CMFB are simple circuits (from the number of devices).
But it is wrong that a SC-CMFB doesn't need power. The size of the CMFB capacitors
determines the BW. Increasing the CMFB capacitors increases the BW of the CMFB
but also increase the loading of the amplifier. Therefore, you sacrifice BW of your amplifier
(you would need to increase the power of your amplifier to keep the same BW).

Regards

Title: Re: Common mode BW and Differential mode BW
Post by Monkeybad on Aug 5^th, 2007, 7:17pm

Hi Berti
I read this consideration from an article. And it refer to
Mike Shuo-Wei Chen and Stanley Bo-Ting Wang
"Design of a Low Power, Low Voltage, CMOS Fully Differential Operational Transconductance Amplifier"
Electrical Engineering and Computer Sciences in the University of California, Berkley.
But sorry I can't find this reference.
In my point of view, CMFB stability needs another analysis. We can break the CMFB loop and do the .AC analysis to find the frequency response. Check the PM to find out whether the CMFB is stable or not.

Yes you're right the SC-CMFB will degrade the DM BW, because the out load become larger.
But compare to the continuous CMFB, SC-CMFB may consume less power. And it has some advantages such like
it doesn't compress the out voltage swing.

The Berti's opinion is very good.
The CMFB BW should be as fast as DM BW. Even faster can be better.
But it is difficult to achieve this goal in practice because the CMFB loop often incudes more transistors and has more non-dominant poles than the DM loop. And it would consume large power.

The DM loop BW is set by the DM input signal BW.
So the CMFB BW is set by the CM input signal BW. If the variation of the CM input signal is small, the CMFB BW
is not important.
But there are other source can be the variation of the CM signal, not only the input signal source, such like the VDD noise, and the imbalance of the circuit that causes DM to CM. So even if the input signal is purely differential , there are still need a CMFB to suppress the other source of the CM variation.

Title: Re: Common mode BW and Differential mode BW
Post by gsheng on Aug 7^th, 2007, 7:09pm

Hi
It's an interesting topic. I have found some discussions in Paul, Gray's "analysis and design of analog ic" . It is almost same as what monkeybad said. You can read page 856.