Yesterday was fun...
I started from where I left off, trying to direct the AOM-passed beam to the cavity and get the TEM00 through, but then I noticed that the PBS for the AOM isolation setup was transmitting a fair amount of light, so I had to realign it to minimize this (the setup is that the S light from the CCW/CW separating PBS goes to a second PBS and then into the AOM, hits a QWP twice on the double-pass and then emerges as P and passes through the second PBS again and to the rest of the experiment).
I then had to completely realign the AOM again, which I did and got ~50% double-pass efficiency again, but then I noticed the same ~2 MHz oscillation in the primary REFL PD signal again when i was trying to lock the cavity and sweep the other direction. This time, however, it could not make it go away by increasing the power (using the HWP at the experiment input) as I could before. I thought perhaps it was the PD, so I switched it out for the other PDA255 we had for the CW REFL, but it saw the same effect. I turned off all the RF oscillators and connected the PD supplies to several different power strips to try and eliminate any ground loops (also, the PDs are on insulating posts).
I then noticed that adjusting the drive current and, sometimes, the temperature of the NPRO made the effect go away or at least stop for a short while. I disconnected the slow and fast actuation signals and meddled with it a bit by hand with the knob, but couldn't see anything obvious except that with a low enough drive current (~half the usual value) it stopped oscillating. The shape of the junk also changed depending on the current and temperature.
I enlisted Frank's help, and we systematically went through the experiment, placing diodes at pickoff spots and adjusting optics one-at-a-time until we found the source. We traced it back all the way to the initial high-extinction PBS at the input to the experiment. We noticed that changing the ratio of admitted and rejected light caused the effect everywhere downstream, so we thought it might have something to do with scattered light. In the end, this appears to have been the case, as PBS had some strange ghost beam reflecting inside and clipping on the custom mount. We think that when the power was high enough this would reflect back into the laser and interfere, causing the oscillation.
We made the effect pretty much go away by moving the thing slightly so that the ghost beam wasn't clipping anymore, but Frank thinks that we might want to replace it anyway since it shouldn't produce a ghost beam by design. I am going to inspect it now and then realign everything. One day I'll get back to the loop measurements.