Lance, Genevieve, Robert
Recently, we shut down specific components of the HVAC system in order to further understand the loss of about 10 Mpc to the HVAC system (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=72308 ). We noted that shutdown of the EX water pump had shown that the 52 Hz DARM peak is produced by the chilled water pump at EX. Based on coupling studies during commissioning time yesterday, the coupling of the water pump can be predicted from shaking injections in the area around the EX cryo-baffle, supporting the hypothesis that the water pump couples at the undamped cryo-baffle (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=72769 ). Here we report on other results of the shutdown tests that we have been able to do so far.
CS Fans SF1, 2, 3, 4, 5, and 6 cost roughly 6 Mpc – coupling via input jitter noise and unknown coupling.
Figure 1 shows that the range increased by about 6Mpc when only the CS turbines were shut down; no chillers or chilled water pumps were shut down. Figure 2, a comparison of DARM spectra before, during, and after the fan-only shutdown, shows that there were two major differences. First, a decrease in peaks associated with input jitter noise, particularly the 120 Hz peak. Second, a broad band reduction in noise between about 20 and 80 Hz. This is not consistent with input jitter noise and represents an unknown noise source that we haven’t found yet.
There is a third difference that could be coincidence. The 9.8 Hz ITM bounce modes are higher in the before and after of Figure 2. I was tempted to wonder if the broad band noise was upconversion from the 9.8 Hz peak. We also have harmonics of roughly 10 Hz in the spectrum every so often. I compared BLRMS of 8.5-10 Hz to BLRMS of 39-50 Hz but didn’t see any obvious correlation. But Im not sure this eliminates the possibility.
120 Hz peak in DARM due to periscope resonance matching new 120 Hz peak from HVAC, possibly due to a new leak in LVEA ducts.
Figure 3 shows that the 120 Hz peak in DARM went away when only SF1, 2, 3 and 4 were shut down. It also shows that the HVAC produces a broad peak between 115 and 120 Hz. I looked back and the 120 Hz vibration peak from the HVAC appears to have started during HVAC work at the end of May, beginning of June. There was a period when flows were increased to a high level for a short time that might have pushed apart a duct connection that is now whistleing at 120 Hz. I think it would be worth checking for a leak in the ducts associated with SF1,2,3 and 4.
In addition to fixing a potential duct leak, we could mitigate the peak in DARM by moving the PSL periscope peak so that it doesn’t overlap with the HVAC peak. In the past I have moved PSL periscope resonances for similar reasons by attaching small weights.
EY HVAC does not contribute significantly to DARM noise
Figure 4 shows that on/off/on/off/on/off/on series of EY fan, chiller and water pump shutdowns does not seem to correlate with range.
Screen shot of the logs attached.
H1:SQZ-SHG_SERVO_IN1GAIN was changed to hopefully add more stability to the SQZr
SDF Diff accepted.
In the Process of taking SQZ_MANAGER to NO_SQUEEZE to Stay in Observing for longer than an hour a Lockloss happened.
Vicki gave me the following instructions to get the IFO into Observing with no SQZ.
Overall:
bring SQZ_MANAGER --> DOWN --> NO SQUEEZING
(top of guardian code, Line 36) set SQZ_MANAGER:
nominal = 'NO_SQUEEZING' (instead of 'FREQ_DEP_SQZ')
REVERT following SDF diffs, such that we run with:
(SHG SERVO IN1GAIN = -9
FIBR_SERVO_COMGAIN=15)
anthony.sanchez@cdsws13: caget H1:SQZ-SHG_SERVO_IN1GAIN
H1:SQZ-SHG_SERVO_IN1GAIN -9
anthony.sanchez@cdsws13: caget H1:SQZ-FIBR_SERVO_COMGAIN
H1:SQZ-FIBR_SERVO_COMGAIN 15