One preliminary investigation indicates that this noise has an environmental component. A coherent signal is seen in seismic sensors (Figure 3, and Figure 4). The seismic signal is present also now (Figure 5). It is visible in the episensor seismometers on TCS benches and the guralp seismometer on CB floor. Elsewere it is either too small or below sensors intrinsic noise.
Slow cooling fans might be suspected as source (most of racks cooling fans have 45-49Hz frequency). At the moment I can just exclude the fans of TCS driver electronics located next to TCS benches, which yesterday I checked out with a portable accelerometer. TCS chillers is another good suspect which shall be verified when possible.
On the other hand, it is interesting to understand the noise path. A first check shows no evidence of coherent noise in *TCS* signals.
--plotting the critical ratio as a function of frequency in the band 22.0 Hz up to 23.0 Hz it can be seen that the worsening factor is of the order of 16% and a huge peak with two horns is clearly visible. It is also well visible plotting the energy. While in the persistency plot the disturbance is not visible, due to the varying frequency and the fact it is not always present
-- another point is that in the first 1.5 months of VSR2 the Vela frequency was exact in between the two
horns, which was "good" . Then the frequency of the disturbance, which was slowly increasing, started to coincide exactly with the VELA frequency (22.38 Hz).
- Thus our observation is that the frequency of this effect was increasing, and
looking at the graph attached here I have got the impression this is still true: the disturbance now is around 22.4...thus not at the Vela frequency but near enough to
create problems..
Pia