Attached is a recent analysis of the "noise spillover" from the primary loop to the secondary (rotation sensing) loop. The plot shows the frequency noise as measured in 1) the primary error signal and 2) the secondary actuation signal. The idea here is that whatever noise is not removed by the primary loop (i.e. the error signal) should be attacked by the secondary loop. In the limit of an ideal secondary loop, the correctly calibrated signals should be exactly equal (that is, the rest of the noise is exactly cancelled by the secondary actuator).

The calibrations are:

• AOM actuation: 6.103 x 10^-4 V/ct * 1/42 (INMON gain) * 500,000 kHz/V (VCO gain)
• CCW error signal: 6.103 x 10^-4 V/ct * 1/50 (SR560 preamp gain) * (1 / 3 x 10-7 V/Hz) (CCW optical response)

It appears as though the gyro noise is in fact dominated by residual common-mode noise above about 30 Hz. Below this, some other source contributes more strongly. The noise here is roughly 300x higher than the expected fundamental gyro noise from displacement, so this suggests that some other non-ideal condition is at fault.