Again following un on Kiwamu's report, I took a look at the non stationarity of the noise during the low noise lock.

In brief, the largest contribution to noise in the 50-200 Hz region comes from MICH/SRCL, and it'slargely modulated by angular motions, mainly ETMs and (maybe) PRM.

Some more details follow. The first plot shows a spectrogram, with a quite clear almost periodic fluctuation of the noise. As usual, I computed the BLRMS in the 100-200 Hz region, to obtain the second plot. In the same plot I show the reconstructed BLRMS using local witness sensors of all main mirrors. The reconstruction is resonably good, and the channel ranking (figure 3) seems to indicate that PRM and ETMY are the most important contributors to the noise non stationarity.

A much better reconstruction of the BLRMS time variation can be obtained using the WFS and QPD angular signals, as shown in figure 4 and 5. The BLRMS is almost perfectly reconstructed. The channel ranking shows that AS_B_RF45_Q is the largest contribution, which seems to confirm that DHARD is important. The strange thing is that I can't see any signal here to indicate that PRM is important.

The last three figures show a "coherogram" (not sure if this word really exists) of DARM with auxiliary d.o.f.s output. Those plots show the coherence as a function of both frequency and time, similarly to what a spectrogram does. Clearly, the fluctuating noise is due mainly to SRCL noise, with a coupling coefficient strongly modulated by the angular fluctuations.