I just had a look at the paper mentioned above and had a go at the test bench provided in Figure 6. When measuring the differential mode gain using an AC analysis I noticed that the roll-off of the gain occurs slightly earlier when compared to an ordinary test bench set up for diff.-mode gain, meaning that the UGF is lower. Am I doing something wrong? I am used to taking the dB20((vout+ - vout-)/(vin+ - vin-)).

Thanks.