In October I found a 13 Hz peak in the CER microphone channel whose frequency was similar to a comb of broad 13 Hz harmonics in DARM (broadened due to intermodulation). We found that we could eliminate the peaks in DARM by shutting down AC1 (73430 ). The acoustic peak from AC1 was not very large, so I looked into the high coupling this trip. I first injected globally with the big speaker, and then narrowed the coupling location down with more localized shaker injections, until I could produce it with small shaker injections at the input mode cleaner tube. AC-1 increased the vibration level seen by the MC-tube accelerometer by about a factor of 3 and shaking the tube by a similar factor produced a similar comb in DARM.
The 6 CER ACs all produce peaks near 13 Hz with similar amplitudes, but only AC-1 matched the frequency of the MC-tube coupling resonance, explaining why it alone produced the peaks in DARM. I think these peaks drift some so another AC may couple in the future and AC-1 may not be a problem if we turn it back on.
The coupling site may be the MC baffles, possibly MCA2
There is not a lot in the MC-tube other than the MC baffles. The last time I measured the lowest resonance of MCA2 (the eye baffle) it was 12.37 Hz (35735), but we have damped it since then. MCA1 (swiss cheese) moved from 12.1 to 12.4 Hz when it was damped (36979 ) , so I would expect the resonance of the eye baffle to be close to 13 Hz. The MC tube walls may have resonances in this band, but no peak was evident at the resonance frequency in ambient signals from the accelerometer mounted on these walls. However, I can not yet eliminate the possibility that the resonance may be something in HAM2 or 3, though the frequency is a little low for most other things. So I will test the coupling site identfication when I return, using the beating shaker technique.
The main HVAC may also couple at this site
A recent HVAC turbine shutdown at the corner station increased the range by about 4Mpc (74108). The coupling mechanism is uncertain and produces noise in the 20-200 Hz band. The HVAC turbines only increase the vibration background at the MC tube by factors of about 1.5 between 7 and 27 Hz, but I shook the tube to increase the level by a factor of about 2 and found that the injection produced strong features in DARM (see Figure 1). Thus the HVAC produces at least part, if not all, of the noise it produces in DARM at this site. Note that this also likely means that other ambient sources are producing noise in DARM at this site even when the HVAC is off.
MCA2 could probably be improved in-situ with viton washers and new panels with larger bolt holes
The eye baffle (MCA2) has been damped with Viton corks and pieces ( 38757 ). I noted at the time that there were fairly high-Q panel resonances in the tens of Hz. The next step would be to put viton washers between the panels and all bolts. I think this could be done in-situ. Also, I noted that bulges in the panels defeated the angling of the baffle and produced reflections in beam-spot photos (see figure photos in 38757 ). These bulges could be eliminated with new panels with larger bolt holes to allow for the slight mismatches of the holes.
The SR tube MC baffles also seem to couple, but not as strongly
I also injected at the SR tube, where there are two more MC baffles. I could produce noise in DARM when shaking at 10 and 14.1 Hz, similar to the 13 and 15.2 Hz resonances for the input MC tube. But the velocity from ambient noise does not appear to be high enough to make noise at harmonics in DARM.
Why is it so hard to find low-level coupling from scattering noise?
For finding linear coupling, we can inject louder vibration signals and produce a larger, more obvious peak in DARM. But for scattering noise in the regime where the motion is comparable to the laser wavelength, increasing the vibration injection amplitude tends to produce smaller, not larger peaks. The increased vibration produces more peaks, at harmonics of the fundamental frequency, and so the scattering “shelf” extends to higher frequencies, but, if the total scattered light does not increase, energy conservation requires that the more numerous peaks be lower in amplitude. If DARM is more sensitive in the higher frequency band that the shelf of harmonics reaches during the injection, the scattering noise can become more obvious even though the peaks are smaller. But when the scattering noise peaks are already small in the most sensitive regions of the detector, increasing the vibration amplitude can actually make them harder to find.
To find the subtle noise in DARM without being able to increase its amplitude, I used spectrograms with very small normalized color scale ranges (I start with 1.3 to 1.5 while the typical default covers several orders of magnitude) to observe small changes in DARM as the “shelf” cutoff moved.
