KAGRA Logbook

Ushiba, K.Tanaka, Hirose

Since the drop in TRANS power could be related to the alignment of ETM_DIFF(DIFFERENTIAL), we planned to raise WFS gain of ETM on PRFPMI. (klog24496)
This time we raised PIT for DIFF and COMM(COMMON) only. As a result, the power spectrum of the error signal was lowered by turning on the control.

Control Plan
Originally, ETM_WFS is a feedback flow of WFS -> "ASC in OPLEV" -> ETM. This means that the transfer function of the control is the following equation...
$G=(S_{WFS}*F_{WFS}+S_{OL}*F_{OL})A$
$G \sim F(F_{WFS}+1)SA$

If the filter of WFS is above gain 1, WFS has priority. And if it is below gain 1, OPLEV has priority.
But this cannot be achieved if the WFS control is not slow compared to the OPLEV control. If the WFS control is faster than the OPLEV control, the following equation holds...
$G \sim (F_{WFS}+F_{OL})SA$
In this case, the control has priority with more gain than the other, depending on the frequency. Since the WFS also detects actuator oscillations, we refer to the MIOPLEV control.

Preparation of the control

We wrote ASC_LOCK guardian status to pass the control through.(PREPARE_TEST_WFS_ASC (1020) in ASC_LOCK guardian)

Turn off WFS of ETM.
HOLD OUTPUT of ISC2OPLEV in TM.

Also measured the actuator balance of EX and EY.(Fig1/Fig2)(/users/Commissioning/data/ASC/2023/0417/SPEC_ACTBALANCE_ETM_MN_{PIT,YAW}.xml)
We compared the OPLEV signal with it set to DOWN to find it.
We put it in OUTPUT.(Fig3: DSOFT,CSOFT->ETMX,ETMY)
The sensing matrices are the same as DHARD.

Set of control

As a test this time, we put the DIFF and COMM controls into DSOFT and CSOFT.(Fig4)
As per the previous control plan, we set the control with reference to the MIOPLEV filter(Fig5).
This time, control was set only for PIT.

The DUMP filter below 1HZ was not included in the preparation because it is adjusted after the control is set.
We set elptic filter as DUMP below 10Hz.
'int' is an integrator below 1Hz.
Compensation filter was added. This is done by setting a filter proportional to the velocity at the resonance peak, so that the output of the filter cancels out the peak there
(In this case, a filter that raises the phase by 90 deg. in the required bandwidth is added.)
Sign: Since the same sensing matrix as DHARD and CHARD is used, the sign in the filter is based on them.

Control ON

The control of DUMPs without Int was put in first. Then, We put gain 1/2 in the filter of WFS and MIOPLEV. This is because, as shown in the above equation, when the WFS filter is added to the OPLEV filter, it is doubled in the band where both MIOPLEV and WFS coexist, and so the 1/2 gain is intended to cancel this out.
Next, the DC control (int) is turned on, and then the MIOPLEV control is turned off.
Did gain adjustment. The control was successfully turned on.

Power spectrum

The power spectrum RMS result of ASC ONOFF shows that UGF is 0.6Hz in DIFF and COMM.(Fig6: /users/Commissioning/data/ASC/2023/0417/SPEC_ETM_WFS_OPLEV_PIT_ASCONOFF.xml )
The RMS has been lowered entirely.
When DIFF was turned on, there was a peak at 0.8Hz. 0.8Hz peak went down when control was turned on for COMM, so it may have been a COMM peak. (Before COMM was added, the 0.8Hz time scale was off by 30 degrees for ETMX and ETMY, so COMM is highly possible.)
There was a peak of -1.75Hz. Notch filters were added to both. This is to avoid returning the peak to the actuator, although it is outside the control band.
The peak below UGF may need to be boosted.
There was a peak at 0.68Hz, but it has not yet been smashed.

Next

Measure the OPENLOOP transfer function.
Try YAW.