I offloaded the BF GAS with the FR.

Detailed report: link

After recovery of the OMC PZT driver, OMC can be successfuly locked and initial alignment can be done (fig1)

[Hirose, Yokozawa, Ushiba, Fujimoto]

Abstract

We successfully recovered both the DRMI 1f and 3f locks.
For the DRMI 1f lock, we also recovered the ADS, and the lock duration reached approximately 20 minutes.
For the DRMI 3f lock, the lock duration was recovered to approximately 9 minutes, although the ADS has not yet been tested.

However, I observed strange lock transitions during the 3f lock, in which the AS DC signal suddenly increased and then returned to its original level.
The cause of this behavior is still unknown and requires further investigation.

Details: DRMI 1f Lock

Earlier today, the IFO LO amplitude was increased from 10 dB to 13 dB (klog 37678), which resolved the issue of the low POP90 signal during the PRMI lock (klog 37667).
We then proceeded with the recovery of the DRMI 1f lock.

Recovery of the 1f Lock

The following modifications were made to recover the DRMI 1f lock.

1. Restoration of filter settings
   There are several filters in MICH, PRCL, and SRCL that must remain enabled during the DRMI lock but are not automatically turned on or off by the VERTEX Guardian.
   We compared the current filter settings with those before the power outage and restored the necessary filters to their previous states.
    
2. Correction of the ELP filter settings
   We noticed that the MICH ELP100 filter was being enabled by the PREP_FM_FOR_DRMI of the VERTEX Guardian, although ELP300 must have been used for the 1f lock before.
   It is possible that ELP100 was mistakenly implemented in a way that also affected the 1f lock during the last DRMI 3f commissioning before the F2F meeting.
   We therefore modified the Guardian so that ELP300 is enabled in PREP_FM_FOR_DRMI, while ELP100 is enabled only in LOCKING_DRMI_3F.

After these modifications and the additional initial alignment, we successfully recovered the DRMI 1f lock.
The gains and offsets were left unchanged from the values already implemented in the Guardian.
Fig. 1 shows a screenshot of the setup during the DRMI 1f lock.
The maximum lock duration was approximately 20 minutes.

We also noticed that the POP90 and AS34 signals were approximately 1.44 and 1.35, respectively, which were significantly higher than their previous values of approximately 1.
This is likely because the IFO LO amplitude before the power outage was lower than the 13 dB setting used today, although we do not have an exact record of the previous amplitude.

OLTFs

Figs. 2, 3, and 4 show the OLTFs of MICH, PRCL, and SRCL measured during the DRMI 1f lock, respectively.
No significant abnormalities were observed.
A few comments on the comparison with the reference measurements from August 15:

• MICH: The measured OLTF gain was approximately 10 dB higher than the reference. This is because the MICH1 gain was increased during the implementation of the DRMI ADS after August 15.
• PRCL: The measured OLTF gain was approximately 4 dB lower than the reference. It is possible that some gain settings have been changed since August 15.
• SRCL: The measured OLTF was almost identical to the reference.

ADS

We tested the DRMI ADS that was implemented during the last DRMI commissioning before the F2F meeting (klog 37407).
The ADS worked properly for IMMT2, PRM, and BS. However, when the SRM ADS was enabled, the SRM alignment changed rapidly.
Upon investigation, we found that the signal demodulated at the SRM dither frequency contained not only the intended AS34 signal but also the POP DC signal.
The POP DC signal had been used for the SRM initial alignment, but it was not disabled when the ASC_LOCK Guardian went down.
As a result, the POP DC signal was unintentionally included in the DRMI ADS.

We modified the ASC_LOCK Guardian to resolve this issue, after which the SRM ADS also worked properly.
Fig. 5 shows a screenshot of the setup with the ADS enabled.

Details: DRMI 3f Lock

Recovery of the 3f Lock

The following modifications were made to recover the DRMI 3f lock.

1. Adjustment of the REFL HWP angle
   We noticed that the incident power on the REFL DC PD was approximately 30 mW.
   This was caused by the HWP rotary encoder losing its position information due to the power outage (klog 37518).
   We therefore changed the HWP setting from 146 deg to 166 deg to reduce the incident power on the REFL DC PD to 10 mW.
   We also updated the corresponding setting in the VERTEX Guardian.
    
2. Adjustment of the MICH1 gain
   The MICH1 gain had been set to 9 to increase the UGF during the 1f lock.
   However, with this gain setting, applying ELP100 to MICH resulted in insufficient phase margin and caused oscillation.
   I therefore modified the VERTEX Guardian so that the MICH1 gain is reduced to 2 for the 3f lock.
    
3. Adjustment of the MICH1 offset
   The MICH1 offset was originally set to 2 in the Guardian.
   However, I noticed that the AS DC signal exceeded 0.01 during the lock, whereas it should normally remain below 0.01 in the DRMI configuration.
   I therefore adjusted the MICH1 offset to minimize the AS DC signal.
   I found that changing the offset from 2 to 10 minimized the AS DC signal to below 0.01 and allowed the lock to remain stable.

With these adjustments, we successfully recovered the DRMI 3f lock.
Fig. 6 shows a screenshot of the setup during the DRMI 3f lock.
The maximum lock duration was approximately 9 minutes.
The ADS operation has not yet been tested for the 3f lock.
 

*Note: Strange Lock Transition

During the DRMI 3f lock, I repeatedly observed a strange behavior in which the AS DC signal suddenly increased to approximately 0.02–0.03 and returned to its original level a few seconds later.
Fig. 7 shows an example of this behavior.
Since the change occurred almost instantaneously, it appeared as if the interferometer had transitioned to a different lock point.
I also observed that the TEM00 mode image on the AS camera became brighter during this transition.
This behavior was not observed during the 1f lock, and its cause remains unknown.
Further investigation is needed to understand this phenomenon.

[Fujimoto, Yokozawa, Ushiba, Hirose]

Continued from klog37670

Summary

Upon changing the IMC LO from 7.4 dBm to 10 dBm, the IRX successfully locked.(klog37678) In this state, the MIF LO was set to 10 dBm and 13 dBm, and cavity scans were performed for each.

For these measurements, the IMC transmission power was set to 1 W, and the DCPD gain of the TRANS IRX was increased by 40 dB.

The modulation index of SB2, determined from the cavity scan, was 0.0959 when the MIF LO was 10 dBm and 0.1399 when it was 13 dBm. The ratio of the modulation indices was approximately 1.46, which was broadly consistent with the expected value of approximately √2 resulting from a 3 dB increase in RF power.

After the measurement, the DCPD gain on the TRANS IRX was returned to 0 dB.

Details

• The basic preparation procedure followed klog36322. However, the term ‘IMC-SERVO_COMBOOST’ mentioned in Step 6 of the Notes section of klog36322 is a mistake; it should read ‘CARM-SERVO_COMBOOST’. When carrying out the procedure in Note 6, the GRX unlocked. When measuring the UGF with one stage of the COMBOOST filter engaged, the phase margin was located on the low-frequency side of the Phase Bubble’s peak. In this case, the gain would need to be increased by approximately 8 dB to align the UGF with the peak of the Phase Bubble. However, simply increasing K1:LSC-CARM_SERVO_IN2GAIN from 15 dB to 16 dB caused the control to become unstable, resulting in the system losing lock. As the cause remains unknown at this stage, measurements were carried out this time with only one stage of the COMBOOST filter engaged.

• A cavity scan was performed under the following conditions.

  IMC LO: 10 dBm (previously: 7.4 dBm)

  MIF LO: 10 dBm, 13 dBm

  IMC transmitted power: 1 W

  TRANS IRX DCPD gain: 40 dB increase compared to the previous setup

  The measurement results are shown in FIG 1 (10 dBm) and FIG 2 (13 dBm).

  In this study, the modulation index was calculated from the TEM00 resonance peaks of the Upper and Lower SB2. Note that the Upper SB1 was excluded from this evaluation because the resonance peak of the first-order transverse mode was very close to the TEM00 peak, making it impossible to accurately identify the TEM00 peak.

• When the MIF LO was changed from 10 dBm to 13 dBm, the modulation index of SB2 increased by a factor of approximately 1.46. A 3 dB increase in RF power results in an RF voltage approximately √2 times greater. Assuming that the modulation index is proportional to the RF voltage, one would expect the modulation index to increase by a factor of approximately 1.41. The results of these measurements are broadly consistent with this expectation.

[Fujimoto, Hirose, Yokozawa, Ushiba]

Abstract:

We changed IMC and IFO local oscillators' amplitude to 10.4 dBm and 13.0 dBm, which were 7.4 dBm and 10.0 dBm, respectively.
After that, we measured the OLTF of IMC and IRX loops, and confirmed that they are almost consistent with those measured before.

Detail:

As reported in klog37656, IMC UGF is lower than before (klog35394), so we investigated the reason.
Though IMC LO amplitude is nominal (7.4 dBm, which is written on the signal generator), LO amplitude was 10.4 dBm after the planned power outage in March according to the photo in notion page.
So, we set LO amplitude to 10.4 dBm and measured the OLTF (fig1).
Then, the UGF becomes 120 kHz, which is almost same as before.

After that, we increased IFO LO amplitude from 10 dBm to 13 dBm, which should be nominal according to the spec of the harmonic generator (JGW-D1808028-v5).
Though IRX could not be locked with 13 dBm amplitude the day before yesterday, it can be locked without any problems this time.
Figure 2 shows the OLTF of IRX loop with 13 dBm LO amplitude, whose UGF is 11 kHz that is same or slightly high compared with the previous measurement (though we didn't measure IRX OLTF for a long time, it should be -20dB around 100kHz according to klog35455).

After increasing LO amplitude, we locked PRMI and confirmed that POP90 value is slightly higher than before (fig3: Y cursor in the top left pannel shows the POP90 when PRMI was locked with 3F before the blackout).
We are not so sure the reason but it might be due to the LO amplitude is slightly higher than before the unexpected blackout (which was 12.3 dBm after the planned power outage on March 2026).

Anyway, the strange points in IMC, IRX, and PRMI seemed to be resolved by increasing LO amplitude.

Note:

For the future references, I summarized current ISC related signal generator setting in the nortion page.

• The speaker amplifier is OFF.
• The 18 V and 24 V voltages measured at the power strips are within the normal ranges.
• All circuits appear to be operating normally, and no abnormalities were observed in the indicators.
• No abnormalities were observed in the indicators inside the I/O chassis.
• The 3 V, 5 V, and 12 V voltages measured on the boards are all within the normal ranges.
• The Timing Slave indicator indicates that synchronization is in progress.

Whitening Filter Test and Abnormal Output

1. Increase in Noise with Analog Signal In Disconnected

2. Abnormal Output with the D-sub 15-pin Connected

[Hirose, Ushiba]

We turned on OMC PZT driver and confirmed OMC length can be scaned by PZT offsets (fig1).
Since +/- 18 V for the circuit is provided by the external power supply, we set the voltage of +/- 18 V and the current limit of +/- 0.5 A.

Note:

During the recovery work of OMC PZT driver, we found that DC 100V power supply is ON (fig2) though the main power is turned off.
I don't know te detail of the power supply but it is likely that DC 100V was automatically recovered after the unexpected blackout.
If this situation is not good for PZT driver, we need to consider how to avoid it.

[Kimura]

 I am attaching a trend graph.
The partial pressure of H₂O shows a downward trend.

[Fujimoto, Ushiba, Hirose]

Summary

It was confirmed that the POP RF90 signal was reduced at PRMI(klog37667). Currently, as the IRX failed to lock, the LO output is being operated at 10 dBm instead of 13 dBm (klog #37656). It has been suggested that the input-output characteristics of the harmonic generator may be non-linear, and we considered the possibility that the modulation index had decreased in line with the reduction in LO output. We therefore set the LO output to 10 dBm and 13 dBm and compared the sideband modulation index using a cavity scan. As a result, although the measurement result increased from 0.0038 to 0.006, It decreased from the previous reading of 0.01 (klog33306). At present, the cause of this reduction has not been identified. After measurement, the LO output was returned to 10 dBm(Fig1). We plan to continue investigating the cause from tomorrow onwards.

Details

• The measurement results of cavity scan are as follows. (Although we had initially planned to use IRX_NORM for the evaluation, the sideband signals were buried in noise; therefore, we used the p-pol DCPD signal from TMSX.)

  	sideband/TEM00
  LO output 10 dBm	0.18/47=About 0.0038(TEM00:Fig2, sideband:Fig3)
  LO output 13 dBm	0.3/48=About 0.006(TEM00:Fig4, sideband:Fig5)

  ** However, we have not yet been able to identify the peaks at each angular frequency.
  The measurement at 13 dBm is only about 60 per cent of the previous measurement of 0.01, and remains low. At present, we have been unable to identify the cause of the reduction in the modulation index to approximately 60 per cent of previous measurement values, occurring along the signal path from the LO through the various circuits to the EOM crystal. Furthermore, if this were due to variations in the interferometer’s alignment, one would expect to observe changes in the peak intensity of TEM00. However, no significant changes were observed in the TEM00 peak during these measurements. Accordingly, it is considered difficult to explain the current decrease in the modulation index solely by alignment variations. At present, the cause of the decrease in the modulation index compared to previous measurements remains unknown.

• Once the measurements were complete, the LO output was reset to the current operating value of 10 dBm. From tomorrow onwards, we plan to continue investigating the cause of the reduction in the modulation index. In particular, with regard to the TMSX’s DCPD signal, we are considering whether adjusting the gain might reduce the impact of noise, thereby improving the measurement accuracy of the sideband signals.

Note

• We followed the procedure for the cavity scan as described in klog36322.

• If you request ALS_LOCKED whilst the IMC is locked, the low-frequency fluctuations in the MCL are fed back to the IR laser’s frequency. As these frequency fluctuations are also passed to the GR laser via the PLL, the GR laser’s frequency fluctuates. In order to maintain the GRX in a resonant state under these conditions, the feedback to the AOM increases, which exceed the control range of the AOM’s VCO. Consequently, the GRX lock becomes unstable.  To prevent this, the GRX’s RFPD signal must be fed back to the IMC’s MCL (MCE) on the low-frequency side(in the ‘Note’ section of klog36322). In particular, the steps outlined in the ‘Note’ section must be carried out immediately after requesting ‘ALS_LOCKED’ in Step 2 of the ‘Detail’ section of klog36322. Please note that any delay in taking this action will cause the GRX lock to become unstable for the reasons stated above.

• In this cavity scan, measurements were taken under the following conditions:
  IMC output laser power: 10 W or more
  PRM state: MISALIGNED_FOR_LOCKAQ

• The OFFSET application channel and ramp time for the cavity scan were based on klog33306. For this measurement, ‘-1’ was set for FM1 in the filter bank, and the OFFSET was applied by specifying the ramp time in diaggui. The filter bank settings can be checked by clicking on the relevant channel in diaggui. Whilst the sweep time should ideally be determined by taking cavity poles into account, in this instance I referred to the settings in klog33306. As the peak of TEM00 in IRX_NORM remained at a nearly constant value across each scan, I concluded that the sweep time set for this experiment was appropriate.

[Hirose, Yokozawa, Fujimoto]

Abstract

We investigated the GRY locking issue observed yesterday and found that both the phase modulation and demodulation LO outputs for Gr PDHY were turned off.
After turning on both LO outputs, we successfully recovered the PDH error signal and locked GRY.
We also found that the OLTF gain was more than 10 dB lower than previously measured due to an incorrect LO phase.
By optimizing the LO phase, we almos recovered the OLTF gain and achieved a UGF of approximately 5 kHz.

Details

Since we had confirmed this morning that the GRY error signal was missing, we investigated the analog electronics in the mine and resolved the issue as described below.

Turning on the LO outputs

We found that both the phase modulation and demodulation LO outputs for Gr PDHY were turned off (Figs. 1 and 2).
After turning on both outputs, we successfully recovered the Gr PDHY error signal and locked GRY using this signal.

LO phase adjustment

We first set the LO phase to −135 deg, which was the value recorded just before the planned power outage in March 2026 (Notion page).
However, the measured GRY OLTF gain was approximately 10 dB lower than that measured in September 2025 (klog 35124).

We therefore optimized the LO phase and found that −70 deg gave the best result.
As shown in Fig. 3, after performing ADS, the OLTF gain almost recovered to within approximately 1 dB of the previous measurement, with a UGF of approximately 5 kHz.

{Kimura}

 I am attaching a trend graph with the upper limit of the left vertical axis corrected to 1×10⁻⁴ Pa.
The partial pressure of H₂O shows a downward trend.

I tested the automatic Segment production and GauCh analysis as part of the validation of the KAGRA DetChar tools with the latest conda environment.
Overall, both jobs worked well under HTCondor.

The Segment job took 15.5 s on average and up to 25 s to treat 15 minutes data. The memory usage was about 200 MB.

The GauCh job took 116 s on average and up to 136 s to treat one hour data. Its memory usage was about 0.8 GB on average and up to about 1.0 GB.
However, the current test data were taken while the interferometer was not locked, so these resource requirements may be underestimated.
I plan to check the memory usage again after the interferometer is locked and update the requested resources in the HTCondor submit files if necessary.

[Fujimoto, Komori]

Abstract:

We measured the IMC FSR to be 5.6243653(1) MHz, corresponding to an IMC round-trip length of 53.302451(1) m.
To reach the target RF sideband frequency, we will move the MCE by 10.26 mm toward the MCI/MCO.

Details:

We tuned the frequency of the seed RF sideband used for the IFO LSC so as to minimize the peak produced by laser-frequency excitation, using the same method as described in klog:23115.
The optimal frequency was found to be 5.6243653(1) MHz, corresponding to an IMC round-trip length of 53.302451(1) m.
This value differs by 13 µm from the previous measurement performed in 2024 (klog:29385).

The target sideband frequency is 5.6265328 MHz.
Therefore, the IMC round-trip length should be shortened by 20.53 mm.
Since the MCE displacement changes the round-trip length by twice the mirror displacement, we will move the MCE by 10.26 mm toward the MCI/MCO.

[Yokozawa, Ushiba, Fujimoto]

Abstract

We worked on lock recovery and initial alignment for XARM and YARM.
For XARM, we completed the initial alignment, including ADS, after recovering the error signal (klog 37656).
For YARM, I recovered the IR and GR resonances and successfully locked IRY. However, GRY was locked at a point away from resonance, and I could not lock it to the TEM00 mode.

Details

XARM
As described in klog 37656, we successfully recovered the IRX error signal by correcting the LO settings.
Thanks to the XARM alignment performed by Yokozawa-san in the morning, we were able to lock XARM with only minor alignment adjustments and complete the initial alignment procedure, including ADS.
However, as pointed out in klog 37656, some issues still need to be investigated, such as the change in the OLTF gain.

YARM
When I started the initial alignment of YARM, no flashes were observed for either IR or GR.
By adjusting the alignment of SR3, BS, IY, and EY while monitoring the camera images and PD signals, I improved the transmission to approximately 0.8 for both IR and GR and successfully locked IRY.
Fig. 1 shows a screenshot of the setup at that time.

However, when I attempted to lock GRY, it locked at a point away from resonance, as shown in Fig. 2.
As a result, the lock check repeatedly failed, causing the system to attempt to relock GRY.

If anyone has any ideas about what might be causing this issue, I would appreciate your comments at the next morning briefing.