Glacier,

  It would be best to include the CMFB network in the simulation.
To include the CMFB network, you might try two approaches:

1) Use Spectre's analysis dependent switches to force the dc
   operating point with common-mode feedback active and
   the ac analysis with the common-mode feedback disabled.

2) Use SpectreRF's sampled pac analysis. Sample the frequency
   response when the amplifier is in the normal mode of
   operation.

                                                                  Best Regards,

                                                                     Sheldon

Glaicer,

I am not familiar SpectreRF. But I have a simple method. Make a CMFB macro model using R, C, and controlled voltage source, although it is not very accurate.

.SUBCKT CMFB_G7 P1 P1D P1DB P2 P2D P2DB CBIAS CMC CMO VDD VON VOP VSS
R1 CMC CMO1  200E3
R0 CMO2 CMC  200E3
V1 CMO 0  1.65
E1 CMO1 0 VOL='V(cbias)-(V(cmo)-V(VOP))' max=3.3 min=0
E2 CMO2 0 VOL='V(cbias)-(V(cmo)-V(VON))' max=3.3 min=0
C2 VON CMC  125E-15
C0 VOP CMC  125E-15  
 
 
* END OF SUBCIRCUIT DEFINITION.
.ENDS CMFB_G7

where VOP and VON are the differential output of an opamp, CMO is the bias for common mode voltage, CBIAS is the system bias for tail current, CMC is tied to the tail transistor in the opamp.


Best regards,
Yawei