CMFB biasing current in fully differential Opamp with adaptive current source

[Junus2012]

Hello,

I am trying to apply an adaptive biasing technique to source the differential pair of the fully differential amplifier, the adaptive biasing source a current of let say Ib under small signal condition. Under slewing or large signal behavoiral this current is increased proportionally to the input difference amplitude.

My question, I alredy design the CMFB circuit according to small signal currents in the circuit, where the tail current of the differential pair is already defined and fixed, but now the tail current is adaptive, do I need to biase the CMFB current to be adaptive as well ?

Note I am using continous time CMFB circuit

Thanks

 

[erikl]

... do I need to bias the CMFB current to be adaptive as well ?

I'd think so, if you want your diff amp react to large signal/step input fast enough.

 

[Junus2012]

I'd think so, if you want your diff amp react to large signal/step input fast enough.

Dear Erikl,

I spent a lot of time to see some people if they adaptively supplying the CMFB ampliifer and indeed I never find. It looks for me that CMFB is to control the circuit under small condition only

 

[erikl]

It looks for me that CMFB is to control the circuit under small condition only

As you probably know, a differential output stage absolutely needs a CMFB, because otherwise the output levels would be undefined.

So CMFB's mission is to keep the voltage levels of the differential output stage in the mid-range of the operating voltage, in order to allow for max. Output Common Mode Range (OCMR), irrespective of the inputs' position anywhere in their ICMR.

Now under large signal / fast step conditions, I think it is particularly important to keep the outputs' levels in an optimal position for largest possible OCMR - and this fast enough for such fast slewing conditions, because the full output range is needed for both outputs. IMHO the CMFB circuit must react even faster than the output stage, in order to fulfill its task well & fast enough, which means even higher pole(s) resp. f_T frequency than the actual diff amp. You know what this means for the CMFB's operating current.

I remember I've read a paper about this context, but unfortunately can't find it any more.

 

[sutapanaki]

There are different opinions on how fast the CMFB circuit has to be. I've seen anything ranging from same speed as the main amplifier down to about 30% of the speed of the main amplifier. Especially in the pulsed circuits as for example the S/H amplifiers if the CMFB circuit is slower, say at 30% then for the time we have for the main amplifier to settle to 10bit accuracy for example, the CMFb circuit will settle to less. But then does it really matter if the common-mode voltage settled to within 1-2mV of its final value? The signal is in the difference, not the common-mode.

 

[Dominik Przyborowski]

It's all true. CMFB does not matter in many points, but can kill performance of IP3 or response on the kickback noise from SAR ADC loading opamp. Also, if output swing is rail to rail, then loussy CMFB killing all benefits from fully differential signal processing.

The main point is "what is an application of opamp?"

 

[erikl]

But then does it really matter if the common-mode voltage settled to within 1-2mV of its final value? The signal is in the difference, not the common-mode.

Very true. But what if the common-mode voltage is just - say - on its half way to its optimum level, before a real large signal arrives at the output stage?

Also, if output swing is rail to rail, then loussy CMFB killing all benefits from fully differential signal processing.

Yeah, that's what I meant.

 

[sutapanaki]

Yes, if the output common-mode voltage is half way to its settling value it will be probably a disaster. That's why I mentioned in my previous post that if it settles to within few mV of its final value by the time the differential voltage had already settled to its full precision, it is most of the times OK.
Yes, CM voltage can kill distortion performance, especially if for some reason one side of the diff amplifier slews differently than the other or if CM is such that it puts the devices at the output in bad operation. These are all things that have to be watched and carefully considered.