How to determine CMFB open loop bandwidth and gain

cmfb bandwidth

Hi all
Since CMFB is a closed loop, how to determine CMFB loop gain and bandwidth?
Does it have any relationship with input signal or ....... Thanks.

cmfb gain

Assuming we are talking about a continuous time CMFB ckt, if the CMFB shares the same path as that of the differential gain path of the main amp (as is the case in many CMFB circuits), cut open the loop at the input of the CMFB circuit, inject an AC signal, and measure the return signal (just as you would in a regular amp). Thus, you can measure the CMFB gain (which should be similar to the differential open loop gain of the main amp altered by the CMFB input stage which typically should have a gain of 1)
do not inject any signals into the input of the amplifier.

open loop ckt

use stb analysis in Spectre

gain bandwidth of cmfb loop

Sorry, I should state my question more clear.
My question is when I design CMFB, what the spec. of loop gain and bandwidth should be set.
1. Does the spec. relative to input signal frequecny?
the same with input signal frequency??
or 1/10 of input signal frequecny
or 2* input signal frequecny
2. Or as low as possible or as high as possible
3. Or it does'nt matter, just take care of loop stability
4. low loop gain or high loop gain, which is better

Thanks !

how to determine cmfb gain and bandwidth

Huang Cheng Chich said:

Sorry, I should state my question more clear.
My question is when I design CMFB, what the spec. of loop gain and bandwidth should be set.
1. Does the spec. relative to input signal frequecny?
the same with input signal frequency??
or 1/10 of input signal frequecny
or 2* input signal frequecny
2. Or as low as possible or as high as possible
3. Or it does'nt matter, just take care of loop stability
4. low loop gain or high loop gain, which is better

Thanks !

Click to expand...

1.Yes. It is a possible objective to make the unity-gain BW of differential-mode and CMFB equal, but it's not a easy job. U can try it
2.theoritically, the higher the BW of CMFB, better. because it can reject any signifcant CM signal in the CMFB bandwidth. But the same, it's not a easy job
3.Exactly, stability is the point. First stable then consider the BW and Loop-gain of CMFB
4.I think you have the answer

we all know that the BW of Differential-mode gain is set by the bandwidth of the DM signal. But as for CM, you should tell yourself what about the CM signal bandwidth you are interested. Thate is , one must guarantee the loop-gain of CMFB is bigger than 1 at frequencies where a significant ac CM output voltage would be generated without CMFB. You may think that the saftiest way is to design a CMFB with equal or even larger BW as DM loop, but in practice, the CM loop usually includes more high frequency poles, so very difficult, sometimes impossible to achieve a stable CMFB loop unless decreasing the gain or/and BW.

Best regards

open-loop bandwidth

the gain of the CMFB is determined by the error between the CM output voltage and its defined vulue. For example, if the gain is 60dB, then, the error is 1mv.

the bandwidth of the CMFB is determined by the applications. for example, if at some frequency, a disturbance or noise will be injected into the amplifier, then the bandwidth of the CMFB should be larger than that frequency.
by the way, there is tradeoff between the bandwidth and the power dissipation.

cmfb bw

so, the bandwidth of cmfb is usually smaller than one of differential path.

bandwidth of cmfb

My experience: yes, BW of CMFB is usually smaller.

Let's look at the problem this way,
first off, we are mostly intersted in the DM as that
is the main applicatin, so we'd like to use most of
our power in that path. For CM, we'd like to use
as little current as possible, so usually, we reuse
some of the DM amplication path to save the power
and get the desired speed/gain. But fundamentally,
we're limited by the available power we're allowed
to use, so CMFB will be slower for given gain/static
error requirement.