I offloaded the following GAS filters with the FRs.

• ITMX: F1 GAS
• ITMY: F0 and BF GAS
• ETMX: BF GAS
• ETMY: F0, F1, F3, and BF GAS
• BS: F0 GAS
• SR2: F0 GAS
• SR3: F0 GAS

Abstract:

I implemented GAS modal damping for PR3 and checked the performance.
Though the best configuration for low frequency noise reduction and high frequency noise reduction is diferent, M1 modal damping with IMV damping would be good as well as the case of BS.
In addition, sensor correction doesn't seem to work well in the case of using GAS modal damping.

Detail:

I implemented modal damping for PR3 as well as BS, SR2, and SR3 (klog36755, klog36780, and klog36781).
Figure 1 and 2 shows the sensor and actuator matrices for PR3 GAS modal damping.
The measured data are stored at  /users/VISsvn/TypeBp/PR3/{Spectra,TF}/Measurements/2026/0421/.
Figure 3 shows the OLTF of GAS modal damping for the first mode.

After implementing the modal damping, I measured IMV OSEM spectrm with thevarious configuration.
Figure 4 shows the result (see legend for the configuration of each color measurement).
Current LOCK_ACQUISITION state induces large noise at high frequency while it has the best performance at microseismic band.
If only IMV damping is engaged, the noise level at high frequency is better as well as no control case while the peak at microseismic band is larger.
If using the modal damping and IMV damping at the same time, the performance is middle of the above two condition.
Though I'm not so sure which configuration is the best for LOCK_ACQUISITION and OBSERVATION, M1 modal damping with IMV damping would be good because this configuration balances the noise level at low frequency and high frequency.

Anyway, if the suspension motion disturbs the lock acquisition of DRMI and RSE, the modal damping is worth testing to reduce te suspension motion of PR3.

With Misato Onishi, Takahiro Yamamoto (remote), Satoru Ikeda (remote)

We attempted to adjust the alignment of Pcal-Y; however, this could not be completed due to an issue with the pico motor.
After discussing with YamaT-san and Ikeda-san, we suspect a communication problem between the pico motor control script and the pico motor controller.
We plan to proceed with the calibration of Pcal-Y after resolving this issue.

KAGRA Pcal-X updates (2026/04/21)

Workers: Misato Onishi, Dan Chen

We performed monthly Pcal-X calibration on 2026/04/21.

After the calibration, we updated EPICS parameters related to the Pcal-X system. No issues were found.

Conclusion:

It is better to use GAS modal damping only for M1 and the other mode should be damped at IM stage.

Detail:

I checked the current BS GAS damping control at OBSERVATION state and found that only the 1st modal mode is damped at F0 GAS while the other mode is damped at IMV OSEM.
So, I tested the modal damping performance only with M1 controls.

Figure 1 shows the comparison of the spectra with M1 modal damping (red) and current OBSERVATION state (blue).
Owing to turning of M2 and M3 modal damping, high frequency noise reduces a lot above 1 Hz.
Since M2 and M3 mode is damped at IM stage, we can use this configuration during both lock acquisition and observation.
So, it is better to use only GAS modal damping for M1 and the other mode should be damped at IM stage.

Note:

This configuration would be better for the other Type-B suspensions, so it should be checked later.

With Yokozawa-san

Many suspensions were found in a tripped state after yesterday’s earthquake.
We checked each system and recovered them one by one.
Both TWR and PAY channels are being inspected.

• ETMX
  Both TWR and PAY exceeded the thresholds during the earthquake.
  → Reset
• ITMX
  OSEM INMON shows large excursions.
  Now returned to approximately the pre-earthquake level → OK.
  → Reset
• ITMY (IY)
  WD was triggered only in TWR.
  → Reset
• ETMY (EY)
  WD was not triggered, but DK (payload only) was red.
  However, signals appear to exceed the threshold.
  → Reset
• BS
  Checked and reset.
• SR2
  Checked and reset.
• PRM
  WD was triggered without reaching the threshold?
  Left as is for now.

Similar work with klog36755 and klog36780.

Abstract:

I implemented GAS modal damping for BS.
Since the noise performance at high frequency seems worse that the current OBSERVATION state, further tuning is necessary.

Detail:

I decoupled the sensors and actuators with the same manner with SR2 and SR3.
Figure 1 and 2 show the sensor and actuator matrices, respectively.
Figure 3-5 show the suspension plants at modal basis.
Figure 6-8 show the OLTF of modal damping control loops.

Figure 9 shows the spectrum comparison of IMV OSEM with LOCK_ACQUISITION state (blue), OBSERVATION state (green), and modal damping (brown).
The noise above 0.9 Hz becomes worse while the noise beow 0.9 Hz looks slightly better, so further tuning is necessary.

Due to the large earthquake, I gave up to continue the work today.

Similar work with klog36755.

Abstract:

I implemented GAS modal damping for SR2.
They seems to wrok well, so they can be implemented into the guardian but not yet to be done.

Detail:

I decoupled the sensors and actuators with the same manner with SR3.
Figure 1 and 2 show the sensor and actuator matrices, respectively.
Figure 3-5 show the suspension plants at modal basis.
Figure 6-8 show the OLTF of modal damping control loops.

Figure 9 shows the spectrum comparison of IMV OSEM with LOCK_ACQUISITION state (blue), OBSERVATION state (green), and modal damping (brown).
The noise above 0.5 Hz can be reduced by using modal damping, so the modal damping seems to work well.

Due to the large earthquake, I gave up to implement them into the guardian today.

I plotted the absolute frequency together with the results of the main laser from klog:36758.

I checked the IM transfer functions under the TMP operation. The situation was not changed (The DC gain of transfer functions is smaller than the reference, except for L).

Date: 2026/04/20

Member: Dan Chen, Misato Onishi

We performed our usual WSK calibration at UToyama.

The results look no problem.

Results

Comparison with Previous Results

Comparing with previous results, no significant issues were found.
Attached graph is the result summary including the latest measured data.

 I tuned the servo control for the FLDACCs. The UGF was changed from 10Hz to 1Hz to avoid the induced noise in the ACCs (geophons). Plot 1 shows the spectra of each FLDACC and ACC with the 10Hz UGF. Plot 2 shows the spectra with the 1Hz UGF. The noise floor of the ACCs was reduced.

[Tanaka, Saito]

To align the optical path of the sub-laser beam with the IR beam on the POS table, we installed two irises and attempted alignment.
 

• In klog:36752, two irises were installed to match the optical path of the sub-laser beam to the IR beam on the POS table. However, the IR beam on the POS table was weak, and the irises were placed at positions where it was difficult to confirm whether the beam was passing through their centers. Therefore, we repositioned the irises: the first iris was placed in front of the mirror used for measuring the IR beam profile, and the second iris was placed on the reflection side of a nearby BS. We adjusted the mirror and the BS so that the IR beam on the POS table hit the center of the mirror located on the reflection side of the BS. Then, by narrowing the irises, we adjusted their positions so that the beam passed through their centers. In this configuration, we placed a beam dump to block the IR beam on the POS table and used the two mirrors after the mode-matching lenses to align the sub-laser beam so that it passed through both irises. However, almost no change in the optical path was observed. This is likely because the distance between the mirrors used for alignment was too short to produce a significant change in the beam path. A more efficient alignment method needs to be considered.

[Ushiba, YamaT]

Today we faced user code error of OMC_LSC guardian to try locking OMC by guardian.
To be honest, I haven’t identified an exact cause. I have simply fixed a part that seemed suspicious.
If a same problem will be still reproduced, additional investigation and a code modification will be necessary.

Error logs and the relevant code are shown in Fig.1 and Fig.2, respectively. This error was slightly strange. This user code error was occurred in counter==3 of FIND_RESONANCE state. In this state, guardian receives an result of the mode analysis. Mode information is received at L305. Guardian logs shows "USERMSG 0: Locked to a HOM with mode number = 0". This message is provided at L319. After then, user code error occurs due to "a = self.p.communicate()". But L305 after executing L319, ezca.swith at L323 should be executed and some kind of ezca log should be remained in guardian logs before L305 is executed again. So if this python codes are sequentially executed, this bug is never occurred.

I guessed the reason why the code isn't executed sequentially is that p.communicate triggers garbage collection. GC is done in lazy when Python judges relevant objects become unnecessary. For this reason, this lazy GC is conflict after p.stdout.close in L329. I'm not sure my guess is really correct or not. But it's an only plausible scenario for now. So I commented out the relevant code. If a problem will be still reproduced, I'll do additional investigation.

Continued work from klog36759.

Abstract:

I aligned OMMT2 and OSTM to OMC.
I also locked OMC manually and confirmed it can be locked.

Detail:

To check the alignment to OMC, I tweaked OMMT2 and OSTM after performing the same alignment procedure as written in klog36759.
To center the both QPDV1 and QPDV2, I moved pico motors of OMMT2 and OSTM as shown in fig1 (Total value shown in the medm screen is the total steps I moved).

After the rough alignment by using pico motor, I tried to engage OMC QPD centering loop by OMC_ASC guardian and succeeded (fig2).
Then, I tried to lock OMC with IMC output of 8.5W, which corresponds to 1.0W with AR coated SRM.
Though I first tried to lock with the OMC_LSC guardian but gave up due to the user coe error at FIND_RESONANCE state, I manually locked the OMC.
Fiogure 3 shows the OMC_LSC related signals when trying to lock OMC, which seems to lock properly.

I tried switching the evacuation pump from TMP to IP again.
9:46 (SRM)1.3E-4 Pa
9:47 Duct-side-GV close -> 3.5E-5 Pa
9:51 IP(#14) ON   ->2.3E-5Pa
We didn't observe a sudden pressure increase like yesterday.
But unfortunately, the pressure around SRM rose to the 10^-4Pa.
So I turned off the IP and reconnected to the TMP.
12:32 IP(#14) OFF
12:34 Duct-side-GV open
Also, since the IP pressure at SR2 indicated 1.1×10^−4Pa, I closed the GV between SR3 and SRM at14:59 to protect it.
Some outgassing from the IP may have occurred, but it looks like there is a large leak at SRM.
 

I checked the IM transfer functions in vacuum. The DC gain of all transfer functions was decreased by 3~5dB. The position change of the OSEMs was 200um maximum (H3 and V1). The measurement for the V1 failed. 

The log and fig made by Yu-san.