I took the Earthquake downtime to align the PMC using the picomotors. The IMC was set to 30 W with the second loop ISS engaged and DC coupled. A 90 Hz line with amplitude 0.2 V was injected into PMC_TF_IN_EXC. The picomotors were moved to minimize the 90 Hz line in the IMC WFS signals. The attached plot shows that the rms seen by the IMC WFS was reduced by about 7. However, the overall coherence between PMC_HV_MON and the IMC WFS did not change significantly; see plot 1. The diffracted power increased from 3.8% to 5.05%; see plot 2. At the same time the sum of reflected and transmitted power of the PMC decreased from 76.5 W to 74.6 W, a 2.5% change; see plot 3.
Currently, the picomotor controller uses channel 1 and 2 for mirrors 1 and 2, respectively. It probably would be easier to hook up the pitch degrees-of-freedom to channel 1 and the yaw ones to channel 2. This way it would be possible to move both mirrors simultaneously in pitch or yaw.
Some numbers to explain the observed 90 Hz jitter by the IMC:
The PMC drive corresponded to ~65 pm/√Hz, and produced a frequency noise of ~3 Hz/√Hz.
The mode cleaner sees a frequency noise amplitude that changed from 0.35 Hz/√Hz to 0.26 Hz/√Hz. This is much less than the expected frequency noise suppression provided by the FSS.
When we started, the angular jitter at 90 Hz had within 20% the same amplitude as the 260 Hz periscope peak which has an amplitude of ~10-4/√Hz in units of beam size and divergence angle.
Making a (very) rough estimate how a misalignment into the PMC together with a length dither and a frequency stabilization system generates angular jitter: A 2.5% power loss in the PMC requires a ~√0.025 ~ 0.15 misalignment (in units of divergence angle). The PMC attenuates HOMs by 1.6%, giving us a TEM01 mode of amplitude ~2.5 x 10-3. The modulation index of a ~3.5 Hz frequency line at 90 Hz is ~0.04. Taking both together we get a jitter of order 1 x 10-4. Hmm, somehow seems too good to be true.