I hooked up the 4395 to the MC servo board test out (TP2A) and looked at the spectrum using our new SPAG4395.py script. We noticed a huge peak at ~3.8 MHz and correctly guessed that it was due to the beat between the MC modulation frequency 29.5 MHz and 3*f1 (~33 MHz).
So we tuned the Marconi for the main mod. from 11065910 to 11066099 Hz and saw the beat note disappear (to within the 1 Hz tuning precision of our Marconi).
New MC length tuning method! Alert the LA Times!
My conjecture is that this temperature dependent mismatch between the modulation frequency (f1) and the MC length is what leads sometimes to our nasty saturating PC DRIVE signal. TBD.