Robert did an HVAC off test. Here is a comparison of GDS CALIB STRAIN NOLINES from earlier on in this lock and during the test. I picked both times off the range plot from a time with no glitches.
Improvement from removal of 120 Hz jitter peak, apparent reduction of 52 Hz peak, and broadband noise reduction at low frequency (scatter noise?).
I have attached a second plot showing the low frequency (1-10 Hz) spectrum of OMC DCPD SUM, showing no appreciable change in the low frequency portion of DARM from this test.
Reminders from the summary pages as to why we got so much BNS range improvement from removing the 52 Hz and 120 Hz features shown in Elenna's ASD comparison. Pulled from https://ldas-jobs.ligo-wa.caltech.edu/~detchar/summary/day/20230817/lock/range/. Range integrand shows ~15 and ~5 MPC/rtHz reduction at the 52 and 120 Hz features. BNS range time series shows brief ~15 MPC improvement at 15:30 UTC during Robert's HVAC OFF tests.
Here is a spectrum of the MICH, PRCL, and SRCL error signals at the time of this test. The most visible change is the reduction of the 120 Hz jitter peak also seen in DARM. There might be some reduction in noisy peaks around 10-40 Hz in the signals, but the effect is small enough it would be useful to repeat this test to see if we can trust that improvement.
Note: the spectra have strange shapes, I think related to some whitening or calibration effect that I haven't bothered to think about to make these plots. I know we have properly calibrated versions of the LSC spectra somewhere, but I am not sure where. For now these serve as a relative comparison.
According to Robert's follow-up / debrief aLOG (LHO:72331) and the time-stamps in the bottom left corner of Elenna's DTT plots, she's is using the time 2023-08-17 15:27 UTC, and that corresponds to the time when Robert had turned off all four the supply fans (SF1, SF2, SF3, and SF4) in the corner station (CS) air handler units (AHU) 1 and 2 that supply the LVEA around 2023-08-17 15:26 UTC.