Sharan Banagiri, Anamaria Effler, Philippe Nguyen, Kara Merfeld, Robert Schofield
We made about 110 acoustic, magnetic, shaker, and HEPI injections last week (DTT files: https://lhocds.ligo-wa.caltech.edu/exports/pem/19aMarPEMinjections/LHO/ ). We used 27 hours of low-noise interferometer time. If we count recovery from losses between the injection periods, we used 36 hours of time (compare to 48 hours for 6 days of 8 hour shifts). We asked for a total of 36 hours of actual injection time, and we think we will need the remaining 9 hours, and possibly more, to complete the injections the week of April 14.
Coupling functions for most channels will eventually appear at http://pem.ligo.org/couplingfunctions
No coupling found at ISCT1
We mounted a shaker on ISCT1 again and found very little coupling to DARM, again clearing in-air POP.
Worst coupling found so far is a factor of a few below DARM
The worst environmental coupling locations that we have found so far have been in the HAM5/6 area, including the squeezer (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=47551), and at EY. The broad-band ambient background at these and other tested locations are expected to produce noise in DARM at a level of a few or less below the current noise floor. Figure 1 shows preliminary estimated vibration contributions to DARM at EY and some photos.
Though we have not completed injections at all locations, we have not noticed any location where the broad-band ambient environmental coupling (excluding peaks from motors, fans etc.) would produce displacement noise at greater than a factor of a few below the noise floor.
Reduction in 59 Hz DARM peak
The largest two peaks in DARM near 59 Hz were identified as coming from motors in the PSL and diode room chillers. We considered two coupling paths: through the floor, and through pressure fluctuations in the water, driving the PSL table and beam jitter. We placed the chiller on rubber machine feet (Figure 2), which reduced the size of the peaks by a factor of about ten in floor accelerometers. However, the reduction of motion at the PSL table was only about a factor of two. Thus, the table motion at this frequency is likely now dominated by pressure fluctuations in the water. The peak in DARM appeared to be reduced by more than a factor of 2, so, coupling through the floor path likely dominated, possibly at HAM5/6.
48 Hz peak source likely not found
We found 3 or 4 locations with resonances at roughly 48 Hz, but none appeared to couple strongly enough to account for the peak in DARM. The frequency modulation of the DARM peak, on time scales consistent with ASC but not temperature, is also not consistent with simple coupling of mechanical resonances.
Intermittent coupling behavior and inconsistent results from placing IO chassis fans on separate supplies
We attempted to remove the IO chassis fan peaks from DARM by placing the chassis fans on separate supplies. This seems to have worked for the CS ISC but not for the EX ISC IO chassis fans (Figure 3). As we have previously noted, the fan coupling varies a lot for some chassis -, note the transient large DARM peaks in Figure 3 (black) from a chassis near the CS SUS rack magnetometer. The SUS IO chassis fans at EX also come and go, though the ISC fans seem always there. More study is needed, but we might go ahead and suggest that LLO put their ISC fans on separate power.
Apparent magnetic coupling resonances revealed by broad band big-coil injections.
Figure 4 shows that a broad-band injection made using the large coil produced a resonance feature in DARM at about 112 Hz and at least four more resonance features between 600 and 900 Hz. A 100-500 Hz injection did not produce the 500-900 Hz features, so the coupling appears to be direct (as opposed to upconversion). The magnetic field measured at the vertex magnetometer (the closest one to the coil) at the 112 Hz resonance frequency is at about 1e-9 Tesla/sqrt(Hz), much larger than the background at that frequency, but about the size of the 120 Hz magnetic field and much smaller than the 60 Hz field. The resonance features are narrow and could have been missed by comb injections, supporting the need for these large coils.
We haven’t yet found the magnetic coupling site that produces these features in DARM; the 112 Hz coupling site does not appear to be the squeezer, because we shut off the squeezer and the peak remained. More study is needed.
More HEPI plots
We made HEPI injections at all test masses and the BS (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=47805). We did not find evidence of scattering from the ACBs, or BS clipping. Plots for all BS HEPI injections are shown in Figure 5.
To do, week of April 14:
1) Acoustic injections in multiple standard locations: CS ebay, two output arm locations, input arm, and Y-man.
2) Shaker injections for scatter tests of BSC2, reduction flange by ITM optical levers, gate valve seat by ITMs.
3) Shaking of P-Cal nozzles at EY as potential scattering site, and possibly other locations
4) HVAC shutdown to check for self-inflicted noise that could be reduced by reducing the air flow.
5) Site activities tests (car, control room etc.)
6) Magnetic injections repeats for saturation in CS ebay and any other repeats found necessary during analysis
7) Mid station injections to check coupling level there (was significant in O2)
8) Mobile accelerometer check for 70 Hz peak candidates
