Common mode loop stability (cmdmprobe vs iprobe vs R/C)

Hi,

I'm trying to simulate the DC gain, bandwidth and phase margin of the common mode feedback loop in a fully differential amplifier. I have tried three methods, and though they give almost the same bandwidth and phase margin, they give different DC gains. I'm not sure which is correct. The three methods I'm using are shown in the attached figure.

Using testbench (b), breaking the loop with a large timeconstant R/C network (1MOhm, 1F), I get Av0=27.2dB. Using testbench (d), using iprobe and stb analysis on the common mode loop only, I get Av0=27.3dB. Using testbench (e), with cmdmprobe breaking the loop at the output of the diff amp (so both the common mode and differential loops are broken), I get Av0=67.7dB.

The differential loop testbenches (a),(c) and (f) give almost identical results for gain/bandwidth/phase-margin of the differential loop. I'm not sure why the common mode loop shows the difference for DC gain with the last testbench, so I'm not sure which is correct. Can anyone help?

Ansu

(b) and (d) is the same.
(f) is not correct.

Added after 4 hours 38 minutes:

because the cmdmprobe in (f) is in two loops, the diff loop and the cm loop.if you want to find the cm PM,you must put iprobe only in cm loop as (d).