I did the EOM sweep measurement. It looks like we are fine above ~30 MHz, but the junk at low frequencies could explain why we had excess RFAM at, say, 18 MHz. The plot and a diagram of the experimental setup are below.
The source level was the maximum 15 dBm, the frequency range was 1-100 MHz (linear) over 801 pts (maximum) with 30 averages. The RFAM amplitude is now back below about -80 dBm for the standard DC PD voltage of 100 mV we have been using. Note: the strange ~2 MHz oscillation for low incident power remains...
I also reconfigured the AOM double-pass setup, realigning the polarizing optics and extracting the double-first-order beam. I was able to get ~50% efficiency (~125 mW out / 250 mW in), or a single-pass efficiency of ~71%. Tomorrow I will direct this beam into the cavity and maximize transmission, then continue with the loop measurements.
Remember that you still have to check the crystal for resonances. Sweep the EOM and look at the CCW PD with the cavity blocked internally. The mechanical Q of the crystal can be ~1000s, so it needs to be a fairly fine sweep.
If there are resonances close to the 33 MHz, you'll have to tune it off a little bit. And where's the sketch for the Pomona matching circuit?