CMFB for pseudo-differential source follower

[fightshan]

tp=&arnumber=4145322https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4145322
Generally, in my opinion, there is no need to stablize the common output voltage of source follower(SF), because the type of input transistor and the type of bias transistor are the same, and there is no mismatch in PMOS and NMOS current.

The cirucit give a uncommon (actually,it is uncommon for me ,because I didn`t see it before.) CMFB structure , this circuit doesn`t sense the output voltage, this is really different from the CMFB which using large resistor or switched cap in the usual text books.

So, I wonder what is the advantage of this CMFB structure? Does it has any disadvantage compared to traditional CMFB circuit?

 

[AMS012]

This is not CMFB. It is just a replica bias.

 

[rsashwinkumar]

As mentioned by AMS012, the part marked replica buffer is just used to generate the required bias voltage for the main pseudo differential buffer. Basically you need to set the output common mode voltage to a comfortable level as supported by the ADC which the buffer drives. The high gain opamp sets the bias voltage so that the output DC level is set to Vcm, which might be the supported level by the ADC.

 

[fightshan]

This is not CMFB. It is just a replica bias.

The replica circuit set the Vo of SF equal to Vcm, the effect is the same to usual CMFB circuit, so I call it "CMFB", maybe "common mode voltage stabilization or setting circuit" is more appropriate.

- - - Updated - - -

As mentioned by AMS012, the part marked replica buffer is just used to generate the required bias voltage for the main pseudo differential buffer. Basically you need to set the output common mode voltage to a comfortable level as supported by the ADC which the buffer drives. The high gain opamp sets the bias voltage so that the output DC level is set to Vcm, which might be the supported level by the ADC.

If I want the output voltage of SF to be a specific value, this circuit seems does work. If I utilize this circuit in an op-amp， it seems that there is no need to design a CMFB network.

 

[fightshan]

This is not CMFB. It is just a replica bias.

I find a replica bias circuit in Razavi book. It is not CMFB circuit, the replica bias make the VX/VY=VREF, the function seems like a CMFB, I am right?

 

[AMS012]

CMFBs will have a negative feed back for common mode signals whereas the replica bias doesn't, though both do the job of defining the common mode voltages in the network.

 

[fightshan]

CMFBs will have a negative feed back for common mode signals whereas the replica bias doesn't, though both do the job of defining the common mode voltages in the network.

ok, they do the same job, so who do it better?
I think the replic bias circuit is simplier than CMFB.

 

[rsashwinkumar]

Having a CMFB is no different from this except that CMFBs are used in a fully-differential system, to force the common mode of a pair of node voltages to a fixed value. Here since the circuit is pseudo-differential, setting the DC bias of one node voltage by biasing the top current source, will automatically make the other node voltage equal to the same value.

 

[rsashwinkumar]

I somehow missed to mention this in the previous post. But a CMFB is no different from this replica bias except that CMFB makes the circuit more immune to common mode disturbances at the output node voltages. Here in replica biased scheme, if the output common mode changes due to some reason, there is no way to correct it automatically but a CMFB provides the necessary kick to drive the common mode to the desired value.