The move of the piezo mirror from the top of the periscope to the table top (Link) left large features in DARM at least in part because a resonance of the mirror mount that replaced the piezo mirror mount, overlapped with a higher frequency periscope peak, just as the lower frequency resonance of the piezo mounting had overlapped with a periscope peak at lower frequencies. Figure 1 shows, in red, periscope spectra before tuning, and, in blue, after tuning. The idea was to move the optic resonances, indicated by the high DARM coherence in red at 340 and 390 Hz, into the red valley centered at about 310 Hz by adding weight.  The blue trace shows that the peaks were moved to 310 and 340 and the coherence with DARM was reduced.

Figure 2 is an “after” photo showing the weight that was clamped to the mirror mount to make this change. In addition, figure 2 shows the safety cover, which covers the vertical path of the beam, that is responsible for the peak just below 300 Hz. Unlike the rest of the periscope, this safety cover is not damped and definitely needs to be - the safety cover resonance is showing up in DARM.

I also added a little weight to the top of the periscope to reduce coherence with DARM at 190 Hz by splitting overlapping resonances. This also helped.

I think that the next step is to minimize jitter coupling from the PSL table to DARM, possibly by injecting a peak using the PSL piezo and modifying alignment to minimize the height of this peak in the IM4 pitch and yaw signals and eventually by minimizing the injected peak height in DARM.