Virgo Logbook

Injection system (EIB-SAS)

chiummo, Gosselin, Melo, Spinicelli, Bulten, Bersanetti - 19:19 Tuesday 27 September 2022 (57211)

INJ recovery

This morning we found EIB ringing since around 4:20 local time. The oscillation was such to unlock the IMC, but not to drive the beam out of the BPC quadrants, so the PMC stayed locked. The usual operation of opening the PID of the bench and then close it again after a few tens of seconds was able to stop the ringing, but it left the bench actuators with a high force (see https://logbook.virgo-gw.eu/virgo/?r=57209). The origin of the the ringing of the bench is still unclear.

Then we proceeded to recover the lock of the IMC cavity, but it took a very long time, only partly because of the tests on the beam crane ongoing in the MC building. Actually we found that the cavity was extremely misaligned and it required to change a lot the TZ tilt of SIB1. We suspected either a sudden misreading of the IB local controls or something malfunctioning on the EIB setpoints In this configuration, anyway, the flashes on the B1p camera occured on the very top of the sensor, and there were no flashes visible at the trasnmission of the arms. Furthermore, looking at the ground-connected out-of-loop BPC quadrant (INJ_IBJM), it was clear that the beam was exiting from the EIB with a very different pointing with respect to the night before, even though the BPC setpoints were unchanged.

After discussions with Henk Jan, we decided to modify the TX setpoint for the EIB control in its configuration file. We aimed at having the SIB1 TZ position back at its original value and at restoring the transmission from the arms.

We changed slowly the EIB_TX from -600urad to -525urad and let the IMC autoalignment follow the beam. This brought back the Sc_IB_MAR_TZ value to around -150, the B1p spot on the camera to its usual position and the flashes in the arms. To be noticed that the new values of IBJM_DC_V and H were not completely recovered to their original values, but we trusted more the suspended bench and the flashes in the arms over the IBJM values, we will anyway keep monitoring the situation.

At the end of the day INJ was completely recovered up to FMODERR_tuned and it was possible to relock the arms.

Images attached to this report

Comments to this report:

bulten - 23:47 Tuesday 27 September 2022 (57214)

The EIB controls keep the position loop closed, but since the ringing of this morning around 2am, there is a noise sholder between 1 and 10 Hz.
It appeared during the early morning. I have to speculate here, but either it is a loud noise source in the lab of some equipment that turned on
(unlikely) or maybe some thing shifted and now something is rubbing against the bench. This could be a cable, something with the water supply hose
of the faraday insulator, or maybe the perspex guide around the laserbeam between laserbench and EIB bench. Maybe that was moved during some
activity. However, the noise source appeared early in the morning, when I think no-one was in the lab, so all these possibilities are very speculative.

The noise in Tx,Tz is below 1 murad RMS so I think that it is save to not block the laserbeam.

I will call tomorrow morning to discuss with the on-site experts to see what we can do.

Images attached to this comment

bertolini, bulten, chiummo, spinicelli, INJ team - 22:46 Wednesday 28 September 2022 (57229)

Today we continued the investigation about the excess noise observable (between 1-10Hz) in Tx and Tz degrees of freedom after the system recovery from the event (loop oscillation for a few hours) of September 27th. This morning the INJ team inspected once again the EIB bench searching for mechanical shorts (rubbing etc); nothing was found. Then we blocked the laser beam and performed further checks on the control signals integrity. Correct operation of the coil drivers was confirmed when looking at the current monitor signals in closed and open loop. Instead a problem was discovered in the three vertical LVDT channels on which a large excess electronic noise is visible above 1Hz. In plot 1, the EIB error signals and the LVDT raw signals are compared in closed-loop before and after the accident of two nights ago. Plot 2 shows how the excess noise persists in the three vertical LVDTs also with all loops open.
The noise seems to be responsible for the contamination of Tx and Tz channels; due to the high blending frequency (s=6) the roll-off of the LVDT blending filter is not sufficient to absorb it. The cause of this malfunctioning is unclear and it will require an intervention. INJ team also attempted to temporarily block the bench to check whether the noise was affected but no effect was observed.
In order to mitigate the problem, additional low pass filtering has been introduced in the standard 5-th order LVDT blending filter; the new filter can be further optimised but some reduction of the excess noise has been achieved (see Plot 3). The laser beam was unblocked around 17:30 UTC.

Images attached to this comment

bertolini, bulten, chiummo, gherardini - 18:36 Thursday 29 September 2022 (57247)

This afternoon we continued investigating the cause of the excess noise on the EIB vertical LVDTs. After Antonino has blocked the beam, Fabio disconnected and reconnected all 3 LVDT/voice coil pairs from the signal conditioning board in the EE room; surprisingly, it turned out that the excess noise has disappeared right before
starting the test (that therefore was not adding any new information). The change in the noise level clearly happened when an oscillation of the control loop was triggered, likely, by the work of Antonino in the laser lab at 14:33:05 UTC. The oscillation stopped when opening the position loop at 14:33:35 UTC (see Plot1).
Plot 2 shows the difference in the LVDTs raw noise level before and after the oscillation. Plot3 shows the error signal of the EIB control after the lucky intervention, with normal operating conditions recovered. The behaviour of the EIB is clearly not reliable and further investigation is needed in the coming days.

Images attached to this comment

gosselin, chiummo, mantovani, gherardini, ciani - 19:25 Thursday 29 September 2022 (57249)

After having revover a normal behavior for EIB, we checked the signals of the injection. It appeared that the beam was quite low on the IBJM quadrant, also low on B1p. This makes sense since we had to lower it to recover the alignment.
We brought the EIB Tx set point back to -600 urad as it was before the noise could appear (it had been set to -525 urad on tuesday).

We slightly adjusted the setpoint from -600 urad to -596 urad to recover the same position on the IBJM_v quadrant. By taking a closer look now it seems that we can still improve it (figure 1). We may want to do it in the coming days.

The IMC followed and went back to the position it had before the noise could appear.
The excess of noise that could be seen in the last two days disappeared. See RFC signals before the noise on EIB and now (figure 2)

Images attached to this comment