Our BNS range became rattier since yesterday (t=-100000 in the attached left). The range goes up and down slowly, and when it is bad, we can see an increase of feature-less noise in DARM between 70 and 500-ish Hz (attached right, red). Quite often it comes down to 100Mpc, sometimes less.
The only thing that changed is the squeezing. Since the shot noise level for f>1kHz looks OK, it's not like the squeezing level is changing, but it could still be something related to squeezing.
We could disable the squeezer, but our baseline BNS range w/o SQZ is ~100Mpc. The variation we're seeing right now is mostly at the level better than 100Mpc and therefore will be masked once we disable SQZ. It would be difficult to tell anything conclusive by doing that.
The TTFSS would be the first thing to check, if not done already.
Since the current increase earlier (51167), it looks like the squeezer laser power drop has stopped, although the DARM noise is still poor.
The attached screenshot shows DCPD BLRMS (these are the ratio of SUM to NULL BLRMS, intended for monitoring squeezing, but they also see other DARM noise and are pretty well correlated with the DARM BLRMS). Before the squeezer laser current was changed, you could imagine that the worse DARM noise was correlated with the squeezer nonlinear gain dropping (due to the green power drop), but DARM isn't restored to normal once the nonlinear gain is fixed.
The TTFSS looks OK (nothing is railed, it is locked and looks normal), although I can't promise there's nothing wrong with it none of the obvious things have gone wrong.
It still seems possible that the squeezer is to blame for today's troubles, and we will continue to plan on swapping the laser Tuesday, but it also seems possible that the squeezer is working pretty normally and there is a different problem with the IFO, Keita is posting about some DARM coherences that seem unlikely to be caused by the squeezer laser problems.
I ran Bruco here:
https://ldas-jobs.ligo.caltech.edu/~ldvw/bruco/keita.kawabe/H1-CAL-DELTAL_EXTERNAL_DQ_2019-08-09-19.18.04-180/results/
Among other things POP_A_LF and REFL LF and REFL RIN had big-ish coherence of about 0.2+.
In the attached DTT screenshot, green is today (bad time), blue is 2 days ago (reasonable), red is July 29 (before the green camera incident). POP_A_LF noise is larger today, it's not the coupling but the noise itself that increased.
We checked ISS 2nd loop but it was fine.
Daniel checked and CM signals looked OK.
Jitter? Something else?
