Jenne, Anamaria

Continuing the measurements from alog 65895, we did a PRCL sweep in the 60W state. The UGF is 35.5 Hz, which is fine, see first plot. There's some coherence with MICH and SRCL, which is not great. In general, PRCL should couple way less to DARM, so in the DRMI diagonalization it's important to keep it out of polluting MICH or SRCL. Viceversa is not really an issue.

Then the power had to be reduced to 50W for violins so the rest of the measurements were done at 50W.
To phase POP I injected a line in PRM M3 at 91 Hz, having engaged the Notch88 filters (band 83-93) in the LSC MICH2, PRCL2 and SRCL2 banks. POP9 looks good, better than some factor of 25 I to Q so it remains unchanged. POP45 however wasn't great, so I changed the phase from -15 deg to 8 deg. See second plot, where references are with initial -15 phase, and current traces are with 8 deg phase. The signal disappears from MICH. (Would be nice to redo the PRCL sweep and see if the coherences are lower with the others)
SRCL subtraction also looks good, but I increased the line amplitude and was able to further reduce it by a factor of 3 by changing the SRCL POP 9I element from 0.048 to 0.049. 
One issue is that PRCL and MICH are not perfectly diagonal in real life, so optimizing for PRCL out of POP45Q might mean a bit less gain for MICH (worth it). I swept MICH and found the UGF to be slightly lower than a couple weeks ago at 50W (8.0 Hz versus some 9.25 Hz). See third plot.
Dan had measured  at 60W the MICH gain with nominal -15 deg phase was 7.5 Hz so this could mean we lose MICH gain as we go up in power, and this further small reduction might make it unstable the higher we go. We should increase the MICH gain by some 2dB beyond 50W to keep it from becoming marginal at higher powers. (regardless if we keep this new phase or not)

Does this matter at all?
Looking at the calibrated spectra of CTRL and ERR in CAL-CS (more on that calibration and noise in a later alog), it seems that some noise goes away from SRCL and especially MICH, see fourth plot. Caveat that I didn't do changes back and forth to ensure this difference is fully true. What I find interesting is that the 60 Hz harmonics are just gone, so they're likely coming from PRCL. Is it to be expected to have such high 60 Hz harmonics in POP9 but not POP45?
The coherence of SRCL and MICH goes down significantly, and also some of PRCL to MICH. NB some coherence could be expected because we (both sites) drive BS for MICH, which optically also makes some PRCL and SRCL. After that, it depends on the noise level.

I have not saved any of these numbers in any DOWN script or config files, I leave that to locals to decide.

These changes will likely modify the FF to DARM for MICH and SRCL. Hopefully making it more reliable since the DOFs should be less muddled, and less polluted by PRCL, but I expect we might have more coherence with DARM without any retuning of the FF filters.