As it turns out, our version of the common mode board does not have high input impedence. I think this is what is messing with the lowpass.
I added photos of the PCB to our 40m DCC page about this board: D1500308, wherein you can see that we have Revision B.
On the aLIGO wiki's CommonModeServo page, one finds that high input impedence was added in Revision E. At LIGO-D040180, one finds this was implemented via an additional dual AD829 instrumentation amplifier stage before the input amplification stage that exists on our board.
Also, I find that the boosts installed are the default 40:4k, 1k:20k, 1k:20k, 500:10k pole zero pairs. Given our 30-40kHz UGF for CARM thus far, maybe we would like to lower some of these boost corner frequencies, to actually be able to use them; so far we only use the first two.