* A recombined interferometer is a good enough configuration to see the scattered light on SWEB (and presumably SNEB)
* Controlling the bench Z motion directly at the bench level can work
The bench control in Z loop is designed just to attenuate the micro-seismic motion (than then causes glitches above 10Hz through upconversion).
It subtracts the ground motion using geophones at the base of the tower, but these give a good signal above 0.2Hz. At 0.1Hz there is 40 deg phase difference, and is high-passed at 0.05Hz (information from Joris). So the loop is AC coupled, and has a resonant gain at 0.4Hz
Figure 1 shows the analysis of the last two injections at 0.3Hz, which are in the band where the loop has some gain.
The first half of the figure is with the loop off, and the second with the loop on. Clearly the predicted arches match the glitches in B8_DC during the whole time, and enabling the Z control loop reduces the motion, and the glitches go up to only 15Hz instead of 80Hz.
Figure 2 shows what is going on with the bench motion. In the first half the bench doesn't have the additional control on, and in the second half clearly the motion is reduced, while during all the time the injected motion at the top stage of the suspension stays about the same.
The next step would be to try the loop on a day with high micro-seismic, so the injection is done with the ground and not using the suspension. This would check if using the geophone to remove the ground motion works well enough, or if the mirror local controls signals have to be use instead. Looking at the predicted arches, the latter is slightly better (and used in Figure 1).
