The model seems to reproduce quite well the data with a G=1.9E-20. This model predicts that, in quiet conditions, the window might contribute significantly to observed bumps around 50Hz, 100Hz and 132Hz (blue line in Figure 2). These correspond to the most intense peaks in Em_SETODE01 displacement spectrum (Figure 3).
As a check, Figure 4 shows an attempt to fit the same data of DET scroll pump, using the accelerometer on the Cryo-trap. The fit is not as good.
Being curious, I digged a bit in old data: on March 10th we injected a 95Hz seismic line on the DET tower base (seismic injections on EDB and DET tower). Looking carefully
an up-conversion pattern was indeed present. The upconv. model fit to these data is shown in Figure 5. The fit seems to work well (G=4E-20) and the prediction for quiet data leads to a similar conclusions about the 50Hz, 100Hz, 132Hz bumps.
A bump peaked around 100Hz is quite evident in the sensitivity of the latest days, being now much more pronounced than in the good data reference of July 12th. I show this in Figure 6. I admit that the hypothesis that it is due to seismic up-conversion is surely questionable, since it is difficult to believe that the Em_SETODE01 sensor measures real seismic noise at exactly 100Hz. But it cannot be excluded. As Federico suggests, the most likely source of an exact 100Hz vibration is a vibrating transformer.
The correlation between the height of 100Hz line in Em_SETODE01and that of the dark fringe bump has to be investigated further.
