Dropbox Paper

VIS Meeting 2020/7/17

Agenda/Minutes of the VIS Meeting

2020/7/17 14:00 - 15:20

Participants: Takahashi, Aso, Terrence, Hirata, Miyo, Shoda, Tanaka, Washimi, Fabian, Lucia

Progress report

Terrence:

We have diagonalized (at DC) the IP actuator and TM actuators for all suspensions. Kontrol.visutils.Vis.actuator_diag() does the job automatically.

We have checked the diagonalization of the optical lever (before TM actuator diagonalization) and then we rescaled the calibration to match that of the IP. This might cause trouble to the payload control but so far there seems to be no problem.

We have enabled sensor correction path for all suspensions. Lucia checked the reduction of seismic noise at the microseism region, in general there’s at least a 2 factor improvement. We haven’t fine tune the sensor correction gain to match inter-calibration between LVDT and seismometer. This should be done when time is allowed.

We have reshaped the IP damping filter so enable active isolation at microseism region. In general, the UGF of the loops are around 1 Hz.

In the middle of this, we found that the suspension models installed in the filter banks don’t match the real model at times. We had to measure transfer function and did sloppy fit before shaping the damping filters. At some point we have to revisit all these suspension models and do better fitting (I should cooperate with Shoda-san to fit the simulated models to the measured data).

For the BS, Lucia measured the noise of the LVDT and geophone and I have done some fitting to the noises. With the fitted sensor nosie, I have optimized blending filters accordingly. They are installed in the system and seems to work without problem. For this optimization, I assumed no residual seismic noise presents in the LVDT signal because of the sensor correction. As a result, the optimal filter blends at around 200 mHz which is not particularly good for active isolation around microseism. We will revisit this later when we have time.

There was a glitch-like feature and a ramp in the F0 LVDT readout. (Miyo-kun and Fabian.)

At the beginning we were looking at a slow channel and it only appeared when the control was on. However, Shoda-san pointed out it appeared in the fast channel with control off.

We finally tracked the problem to the sensor correction path which is not shown in the medm screen. The sensor correction block didn’t have an input but the gain was set to one, injecting noise.

We temporarily replaced the LVDT driver by adjusting the values of resistances and phases reported in the klog. However, the results were not satisfactory. Whereas some offsets were relatively good, some others were clearly bad. This method seems to have worked in the past for SR3.

Before placing back the original LVDT driver we measured the resistances and phases again and we found differences with the values reported in the klog. This is definitively bad.

We still have to prove the replacement method works by testing the procedure with an LVDT which we can calibrate after replacing the card.

Towards O4

KAGRA new state "Upgrade phase for O4 (bKAGRA phase 4 ??)" started from 11th July.