Summary: Results from laser vibrometry support the conclusion that scattering from the Swiss Cheese baffle (MCA1) is, along with beam jitter, a dominant source of vibration coupling to DARM. This scattering has been reducing range throughout O2 and may limit any improvements from fixing ITMX. The Q of the worst resonance (12 Hz) is over 100, amplifying ground motions in the tens of nanometers to produce fringe-wrapping shelves reaching above 100 Hz in DARM. Because the Q is so high, it may be possible to significantly mitigate the problem with only minor damping. During the May vent, we could access the baffle for minor improvements through nozzles on the reduction flange that it sits in front of, and avoid a special clean room or removal of HAM doors.
In a recent log, https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=35166 , the input arm, between HAM2 and HAM3 was identified as the main source of noise from trucks, the fire pump, and likely the HVAC. The dominant coupling was at about 12 Hz. While MC1A, the Swiss Cheese baffle (apologies to the Swiss), was suspect, there was also the Eye baffle (MC1B) and the tube itself as possible sources.
In order to narrow down the possibilities, I set up a laser vibrometer at the MC3 camera port (Figure 1) where I could point it at either the ~10 m distant Swiss Cheese baffle or the ~30 cm distant Eye baffle. Figure 2 shows the results of tapping on the input mode cleaner beam tube with the vibrometer on the Swiss Cheese or the Eye baffle. While the beam tube was easily eliminated, the culprit baffle was not immediately obvious because both baffles had strong resonances near 12 Hz: the Swiss Cheese baffle resonance was at 12.095 Hz, and the Eye baffle resonance was at 12.37 Hz.
I discriminated between the two baffles by using a shaker mounted on the beam tube to alternate shaking between the resonant frequencies of each baffle, while measuring the velocity of each baffle with the laser vibrometer. Figure 3 shows that the scattering shelf cutoff was consistent with the motion of the Swiss Cheese baffle, but was not consistent with the motion of the Eye baffle. In more detail, at each of the two frequencies, 12.095 and 12.37 Hz, I adjusted the shaking so that the peak of the largest scattering shelf in DARM would cut off at about 450 Hz. The motion of the Eye baffle for the two different frequencies differed by a factor of about 50, while the motion of the Swiss cheese baffle differed by only roughly 3. Thus the Swiss cheese baffle is much more likely to be the source of the scattering. This consistency check avoids some of the complexities of trying to predict the cutoff frequency of the scattering shelf, such as the potential for a different velocity at the scattering site on the baffle then at the location measured by the vibrometer (e.g. because of different distances from a node).
Nevertheless, a simple prediction of the scattering site displacement is not far off from the displacement of the point measured with the vibrometer on the Swiss Cheese baffle (3 microns predicted vs. 1 micron measured at 12.37 Hz and 1.6 measured at 12.1 Hz). The smaller than predicted motion may result from the placement of the vibrometer beam at the edge of both baffles (because I could only get the edge of the Eye baffle). Also, note that there is a large shelf at 450 Hz, and a smaller shelf at about 650 Hz that may be produced by two different scattering sites on the baffle. The Swiss Cheese baffle has been suspect for a long time because of the high reflectivity seen in beam-spot photos (https://alog.ligo-wa.caltech.edu/aLOG/uploads/9564_20140126161432_Figure4-PR3.pdf) . The ability to measure the baffle velocity at a distance comparable to the laser vibrometer’s stated distance limit is further evidence of a bright retro-reflection. The vibrometer was calibrated by shining it near the blue-trace accelerometer in Figures 2 and 3.
The arguments that the input arm is a dominant vibration coupling site were discussed in the previous log ( https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=35166 ). Here I suggest that this noise has been a daily problem for all of O2. Figure 4 shows a comparison of the inspiral range and the baffle band, 10-30 Hz at the Corner Station, for a few days in April and a few days last December. As an aside, the Gaussian looking peaks at the beginning and end of each Hanford work day are aligned with morning and evening rush hour, when many cars pass the Corner Station on the way to or from the site. Because of the axle spacing and car speed, the traffic produces signals that are strong in the 10-15 Hz band. The peaks are not present during non-working days, were found in samples for multiple years on working days with peaks at the expected times of 5:30 AM and 4:30 PM PT, and they shift appropriately with daylight savings time.
During externally quiet times, our HVAC dominates in the 10-20 Hz band and limits our range (https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=32886 ). Thus any improvements that we get in range from cleaning ITMX will be limited by the scattering noise from the Swiss Cheese baffle.
The width of the baffle’s 12.1 Hz peak in Figure 2 indicates that the Q of this resonance is at least 150, and the ring-down time of the scattering shelf in DARM, shown in Figure 5, suggests that the Q is around 200. This explains why the scattering produces a fringe-wrapping shelf for ground motions that are almost a couple of orders of magnitude smaller than our laser wavelength.
Because the Q of the baffle is so high, it would likely be easy to get a reduction in Q, and a corresponding reduction in velocity, with a minimal damping scheme. If we reduced the Q to 70 from 200, the noise that now reaches about 150 Hz would only reach about 50 Hz. In addition, since the baffle is held to the support ring only at 4 points (Figure 6), it might also be easy to raise the resonant frequency a little, and further reduce the velocity. For example, one could imagine wedging stiff damping material between the support ring and the baffle. Because the baffle is so close to the ports (so as not to block the view) the space between the support ring and the baffle could be reached without pulling doors (Figure 6). We could pull blanking flanges on two nozzles in the reduction flange between HAM2 and the input beam tube during the May vent. If done when the purge air flow was good (e.g. when the doors are being bolted on the other chambers), I am told that we wouldn’t even need a clean room. The laser vibrometer could be used to monitor progress during the damping operation.
