Ushiba, Komori, Tanaka (original idea by Fujimoto-kun?)

Ushiba-san modified the script of TOF measuremet for schnupp asymmentry in order to be able measure the PRC length. We ran the script and tried to measure the REFL PDA3 RF45 optimal demod. phases for Xarm and PRMI, respectively. This time, we locked PRMI with the 1f signal (VERTEX guardian state is PRMI_1F_LOCKED). The basic procedure is the same as in klog36343. 

Each optimal demod. phase for PRMI and Xarm is written as follows,

$\theta_{\mathrm{PRMI}} = \frac{\omega_{\mathrm{mod}}}{c} (L_{\mathrm{EOM2PRM}} + L_{\mathrm{PRM2PD}})$

$\theta_{\mathrm{Xarm}} = \frac{\omega_{\mathrm{mod}}}{c} (L_{\mathrm{EOM2PRM}} + 2 (L_{\mathrm{PRM2BS}} + L_{\mathrm{BS2ITMX}}) + L_{\mathrm{PRM2PD}})$

where, omega_{mod} is a modulation angular frequency, L_{A2B} is a distance from A to B (A,B: EOM, PD, PRM, BS, ITMX), and c is a speed of light. These phase difference is derived as

$\theta_{\mathrm{Xarm}} - \theta_{\mathrm{PRMI}} = \frac{\omega_{\mathrm{mod}}}{c} \times 2(L_{\mathrm{PRM2BS}}+ L_{\mathrm{BS2ITMX}}) = \frac{\omega_{\mathrm{mod}}}{c} \times 2(L_{\mathrm{PRM2BS}}+ \frac{L_{\mathrm{BS2ITMX}} + L_{\mathrm{BS2ITMY}}}{2} + \frac{L_{\mathrm{BS2ITMX}} - L_{\mathrm{BS2ITMY}}}{2}) = \frac{\omega_{\mathrm{mod}}}{c} \times 2(L_{\mathrm{PRCL}} + \frac{L_{\mathrm{asym}}}{2})$

we can derive

$L_{\mathrm{PRCL}} + \frac{L_{\mathrm{asym}}}{2} = \frac{c}{2\omega_{\mathrm{mod}}} (\theta_{\mathrm{Xarm}} - \theta_{\mathrm{PRMI}})$

## Results

Fig.1 and Fig.2 show the results of measurements for Xarm and PRMI, The optimal demod. phases of REFL PDA3 RF45 for Xarm and PRMI were 230.14 degrees and 117.80 degrees, respectively. (I forgot to note the error value...)

If I assumed 20*360 degrees additional rotation because 45 MHz is 20 times FSR of PRMI, L_PRCL + L_asym/2 = 2.9979e8 * (20*360 + 230.14 -117.80) / (2*360 * 8* 5.624365513 MHz) ~ 67.67 m

It seems to be inconsistent with the design value (L_PRCL + L_asym/2 = 66.591 + (26.6649 - 23.3351)/2 = 68.2559 m (ref. JGWwiki)).

We are not sure of the cause of this discrepancy. We suspect that the RF45 sideband is not exactly at the anti-resonance point of the PRMI. As a result, its reflection phase deviates from 180°. This effect is expected to be more noticeable in the low-finesse PRMI than in the high-finesse X arm cavity.

But we did not confirm whether the SNR is enough or not with PRMI configration. We need more investigation.

[Kawakami, Komori, Tanaka, Saito]

A mixer and a 140 MHz low-pass filter were inserted between the 20 dB RF amplifier and the 45 dB RF amplifier. By setting the mixer's LO frequency to 350 MHz, the beat signal at approximately 450 MHz was successfully down-converted to approximately 100 MHz. PLL operation was successfully achieved using this down-converted signal. In the next experiment, a spectrum analyzer will be installed at the OMC REFL port, and since the mixer supports input frequencies up to 500 MHz, beat signals up to approximately ±500 MHz will be analyzed. The cavity lengths of SRX, SRY, PRX, and PRY will then be measured by applying the same fitting procedure used in klog:37209.
 

• First, the signal after the 20 dB RF amplifier was monitored with a spectrum analyzer, and the sub-laser temperature was adjusted so that the beat frequency became approximately 450 MHz. A mixer and a 140 MHz low-pass filter were then inserted between the 20 dB RF amplifier and the 45 dB RF amplifier. In addition, the 10 dB attenuator located before the PFD was removed. The current control chain from the RFPD to the sub-laser PZT is as follows:

  RFPD → 12 MHz high-pass filter → 20 dB RF amplifier → mixer → 140 MHz low-pass filter → 45 dB RF amplifier → power splitter → PFD → 100 kHz low-pass filter → Moku:Lab (100 Hz integrator, 10 kHz low-pass filter) → SR560 (gain = 200, 1 Hz low-pass filter) → 20 dB high-voltage amplifier → sub-laser PZT
   

• With the mixer's LO frequency set to 350 MHz, the signal taken from the power splitter before the PFD was monitored using the Moku:Lab spectrum analyzer, and the beat signal was observed at approximately 100 MHz. When the mixer's LO frequency was decreased, the beat frequency increased, confirming that the original beat signal was at approximately 450 MHz. Since the PFD locks with an offset when the input frequency is too low, the sub-laser temperature was adjusted so that the down-converted signal became approximately 160 MHz. The mixer's LO frequency was then changed to 340 MHz, resulting in a down-converted beat frequency of approximately 135 MHz. Finally, by setting the PFD LO frequency to 135 MHz, stable PLL lock was successfully achieved.

• In the next experiment, a spectrum analyzer will be installed at the OMC REFL port. Since the mixer can process signals up to 500 MHz, beat signals up to approximately ±500 MHz will be measured. The same fitting procedure as in klog:37209 will then be applied to determine the lengths of the SRX, SRY, PRX, and PRY cavities.

[Kawakami, Tanaka, Hirose]

We performed the initial alignment for Xarm, Yarm, PRMI, and SRY.

• We attempted DRMI initial alignment partway through, but since the BS ADS isn’t working correctly, adjustments are needed to separate the signal degrees of freedom.
• During the SRY initial alignment, the amplitude of the SRCL control feedback signal was saturating, so following Ushiba-san's advice, we turned on FM7 (Elptrick Filter 500 Hz). We also modified the settings so that it turns on every time in the “LOCKING_SRC_3F” state of the VERTEX Guardian.

[Tanaka, Saito]

The sub-laser was injected into the SRY cavity, and the PLL was locked. The LO frequency was then frequency-modulated by ±10 kHz at a modulation frequency of 1 kHz. As a result, a signal appeared at 1 kHz in the power spectrum of the OMC REFL DC PD. By adjusting the LO frequency to minimize this signal, its amplitude was reduced to approximately the noise floor. To further improve the frequency resolution, the frequency modulation amplitude was increased to ±30 kHz. However, the amplitude of the 1 kHz signal changed in response to spontaneous fluctuations in both the frequency and amplitude of the beat signal observed at the OMC REFL RF PD, making it difficult to determine the beat frequency with 1 kHz-level precision. Therefore, it is considered that the beat frequency can be determined with a precision on the order of 10 kHz.
 

• First, the SRY cavity was locked, and the main-laser intensity noise measured by the OMC REFL DC PD was reduced to a level comparable to that reported in klog:37144. The sub-laser was then injected into SRY, and the PLL was locked. A Moku:Lab was used as the LO source. Using its frequency modulation function, the LO frequency was swept by ±2 MHz at 10 mHz, and the LO frequency was adjusted to approximately maximize the beat signal observed at the OMC REFL RF PD.
   
• Next, the LO frequency was modulated by ±10 kHz at 1 kHz. A spectral peak approximately five times higher than the noise floor appeared at 1 kHz in the power spectrum of the OMC REFL DC PD. The LO frequency was then adjusted so as to minimize this peak, and at a certain frequency the peak became comparable to the noise floor. To improve the frequency accuracy further, the modulation amplitude was increased to ±30 kHz. This again produced a peak at 1 kHz, and the LO frequency was adjusted in 1 kHz increments to minimize it. However, the amplitude of the 1 kHz peak varied together with spontaneous changes in the frequency and amplitude of the beat signal observed at the OMC REFL RF PD. As a result, determining the beat frequency with 1 kHz-level precision proved difficult. It is therefore concluded that the beat frequency can be determined with a precision on the order of 10 kHz.
   
• The final power spectra of the OMC REFL DC PD are shown in Figures 1 and 2. In Figure 1, the red trace corresponds to the present measurement. Its higher noise level at high frequencies compared with the other traces is attributed to the additional intensity noise from the sub-laser. In Figure 2, the green trace was measured with the sub-laser injected but without LO frequency modulation; the blue trace was measured after initially adjusting the LO frequency to approximately maximize the OMC REFL beat signal; and the red trace was measured after increasing the modulation amplitude to ±30 kHz and fine-tuning the LO frequency with approximately 1 kHz resolution.

[ Kimura, Nakagaki ]

We tested the device for acquiring the open/closed status of the PRM-PR3 gate valve and confirmed that it operates correctly.

Abstract:

I measured the schnupp asymmetry in the same manner as written in klog36343.
Since REFL PDA3 is the best sensor to use according to the previous measurement, I only used REFL PDA3 in this time.
Obtained result is  3.3443(16) m, which is really consistent with the values calculated from the results of the individual length measurement of PRX/PRY and SRX/SRY reported in klog37209 and klog37223.

Detail:

To verify the PRCL/SRCL measurement by checking the consistency of the Schnupp asymmetry between the values calculated from the length of PRX/PRY and SRX/SRY reported in  klog37209 and klog37223 and from the time of flight measurement, I repeated the Schnupp asymmetry measurement reported in klog36343 with greater precision.

What I changed from the previous measurement are followings:
1. Increase the laser power from 1.2 W to 11.5 W.
2. Increase the averaging time for each measurement data from 10 second to 60 second.
3. Increase the amplitude of frequency modulation (excitation to MCE suspension) from 100 cnts to 200 cnts.

Thanks to the above modification, errors of the estimation of the optimal phase difference between X arm and Y arm were improved, resulting in the accurate result.
Figure 1 to 4 shows the measured data and fitting results.
Measurement data and fitting results are stored in //users/Commissioning/scripts/schnupp/260718/.

I summarized the optimal phase and its errors in the following table.

During the measurement, seed frequency for the f2 sideband was 5.624365513 MHz.

By using above results, schnupp asymmetry can be estimated as follows:
dL = 3.3443(16) m from PDA3 REFL 45 I with XARM and PDA3 REFL 45 I with YARM.
dL = 3.3443(14) m from PDA3 REFL 45 Q with XARM and PDA3 REFL 45 Q with YARM.

Schnupp asymmetry calculated from the PRX/PRY and SRX/SRY length in klog37209 and klog37223 are 3.3405(48), both results are very consistent.

Note:

For the measurement, I deleted the sensing matrix values putting at XARM and YARM, which is implemented in the work reported in klog37228.
Figure 5 shows the values before deleting the input matrix.

We confirmed that INITIAL_ALIGNMENT seems to work well. We made a medm screen for DRMI alignment. You can align PRM, SRM, BS, IMMT1 and IMMT2 by following this procedure.

## What we did

• We implemented the states, "PREP_FOR_DRMI_ALIGN", "LOCKING_DRMI", "ENGAGE_ADS_DRMI", "ALIGNING_DRMI", and "RECORD_GOOD_VALUES_DRMI".
• We made a medm screen in INITIAL_ALIGNMENT2407.adl (fig.1) to operate the DRMI initial alignment
• We tested the state one by one from this medm screen. First, The guardian reaches the RECORD_GOOD_VALUES_DRMI state smoothly. The good alignment values of PRM, SRM, BS, and IMMT2 are recorded in each DRFPMI_GOOD_OPLEV_{PIT, YAW} channel (fig.2). However, offloading was not worked at that time. We found that an offload script, reset_ASC_feedback.py was not worked if the values were recorded in DRFPMI_GOOD_OPLEV channels. Yamamoto-san modified the script.
• After the modification, offloading works well.

[Min Chen, Yuchen Liu, Chia-Jui Chou, Takaaki Yokozawa]

We investigated the coherence of 24 auxiliary channels with the online strain channel during the shaker injection test conducted in February 2026. The summary of the shaker injection test can be found here:

https://gwdoc.icrr.u-tokyo.ac.jp/cgi-bin/private/DocDB/ShowDocument?docid=17238

We calculated the coherence between 24 auxiliary channels and online strain channel and ranked the channels according to the coherence values. Then we pick the channels with top 4 ranks with coherence value larger than 0.2 into the important channel list between 280 Hz to 420 Hz. Here are the important channels:

– K1:IMC-REFL QPDA1 DC PIT OUT DQ
– K1:IMC-REFL QPDA1 DC YAW OUT DQ
– K1:IMC-REFL QPDA1 RF14 I PIT OUT DQ
– K1:IMC-REFL QPDA1 RF14 I YAW OUT DQ
– K1:IMC-REFL QPDA1 RF14 Q PIT OUT DQ
– K1:IMC-REFL QPDA2 DC PIT OUT DQ
– K1:IMC-REFL QPDA2 DC YAW OUT DQ
– K1:IMC-REFL QPDA2 RF14 I PIT OUT DQ
– K1:IMC-REFL QPDA2 RF14 I YAW OUT DQ
– K1:IMC-REFL QPDA2 RF14 Q PIT OUT DQ
– K1:IMC-REFL QPDA2 RF14 Q YAW OUT DQ
– K1:PSL-IP QPD1 DC PIT OUT DQ
– K1:PSL-IP QPD2 DC PIT OUT DQ
– K1:PSL-IP QPD2 DC YAW OUT DQ

The detailed report can be found here:

https://gwdoc.icrr.u-tokyo.ac.jp/cgi-bin/private/DocDB/ShowDocument?docid=17451

The coherence plots can be found here:

https://ldas-jobs.ligo.caltech.edu/~chiajui.chou/K1-o4c-coherence/

I offloaded the F0 GAS with the FR a little. The FR closes a movable limit.

I had tested the Pcal GRD code.

It works as we expect.

Aritomi, Tanaka

We performed some trials, mainly transition from 1f to 3f. But the trial was not work. We need more investigation.

### DRMI lock modification

This morning, DRMI lock got unstable for some reasons. One reason is that MICH loop is oscillated at SRM ADS frequency: ~6 Hz. At that moment, we were not sure of the cause of the oscillation. At once, we turned off only SRM ADS. Also, we changed the setpoints. Then, DRMI lock became stable a little. Another reason is that sometimes PRCL began the oscillation at ~100 Hz. After DRMI lock becoming stable a little, we measued the OLTFs. PRCL UGF increased to 100 Hz though the phase margin was less than 10 deg. So we lowered the gain. Similarly, SRCL OLTF seems to get better. We adjusted the gain. Current SRCL UGF is ~60 Hz (fig.1), PRCL UGF is ~40 Hz (fig.2).

Finally, the gains of the loop summerized the FM4 filter.

### Trial of the transition from 1f to 3f

To transit to 3f signal, we performed the phasing of 3f signals. we followed the procedure in klog32604.  As for PRCL, we performed the phasing for REFL135 PD to maximize the I signal when PRCL was excited(fig.3). As for SRCL, we performed the phasing for REFL51 PD to maximize the I signal when SRCL was excited (fig.4). After that, we checked the RF51 signal when MICH was excited. Although the demod. phase is optimized for the SRCL signal, I signal was larger than Q signal when MICH was excited because BS motion is mixture with common and differential length change (fig.5). So we may need to decouple each motion. However, this time, we proceeded our work with this state.

We measued each relative gain and phase between 1f and 3f signals. the figures in the right lower 2 panes of fig. 3, 4, and 5 show each gain and phase of PRCL, SRCL, and MICH. As you can see, the relative phase of SRCL and PRCL seems to be 180 deg. But, the phase of MICH is 40 deg since there are some couplings. Anyway, we input these values in matrix (fig.6) and tried the transition. Then we succeeded PRCL and SRCL transition but failed MICH transition. So we moved the decoupling.

To decouple the DoFs, we excited the one DoFs, PRCL (left panel in fig.1) or SRCL (left panel in fig.2) and measued the relative gain and phase of between excited DoF error signal and others' error signal. From the results, I chosed CARM_IN1 as decoupled sensor for MICH 3F for the test, XARM_IN1 as decoupled SRCL 3F, and YARM_IN1 as decoupled PRCL 3F. After the decoupling, I measured the coupling but the ratio seems not to changed. 

### DRMI ADS modification

This morning, we also found that DRMI began the oscillation at ~6 Hz just after engaging ADS. 6.3 Hz is the dither frequency of SRM YAW. The resG for SRM ADS in the MICH loop caused the oscillation. After decoupling the 1f error signals, MICH OLTF seems to got better than before. According to the MICH loop, current MICH UGF is around 10 Hz (fig.6). So I turned off the resG filter in engaging ADS. Then, ADSs seem to become stable.

### DRMI alignment Implementation to the INITIAL_ALIGNMENT guardian

After the modification of DRMI ADS, I implemented the DRMI alignment state with DRMI_1F_LOCKED into the INITIAL_ALIGNMENT guardian. I have not checked INITIAL_ALIGNMENT guardian behavior due to the earthquake.

With Dan Chen

We continued the work from the previous day. (klog_37206)

By adjusting the two mirrors newly installed inside the Pcal-Y Tx module (Fig. 1), we successfully aligned the new laser beam and propagated it to the RxPD.

On the RxPD, the beam position of the new laser appeared to be nearly identical to that of the current laser. The beam positions of the current laser and the new laser are shown in Fig. 2 and Fig. 3.

We also closed the OFS loop, and it appeared to operate normally.

Due to some remaining issues with the previous Pcal Guardian system, we have now updated the Pcal Guardian code to address these issues. The specific details are as follows:

Issue 1: Swap the order of HIGH_POWER and HIGH_POWER_RX_MON.

Before: INCREASE_OFFSET → HIGH_POWER_RX_MON → HIGH_POWER ⇄ WAITING_IFO    HIGH_POWER_RX_MON →  DECREASE_OFFSET   HIGH_POWER →  DECREASE_OFFSET

Update: INCREASE_OFFSET → HIGH_POWER ⇄ HIGH_POWER_RX_MON ⇄ WAITING_IFO   HIGH_POWER →  DECREASE_OFFSET   HIGH_POWER_RX_MON →  DECREASE_OFFSET

2.Issue: Turn on the injection switches in HIGH_POWER, and turn them off automatically when leaving the state.

Update: Added code to automatically turn on the injection switches in the HIGH_POWER state and automatically turn them off when leaving that state (entering HIGH_POWER_RX_MON or DECREASE_OFFSET).

3.Issue: Send a Slack notification if Guardian remains in HIGH_POWER for an extended period.

Update: Added the code to send a Slack notification.

4.Issue: Allow the real-time model to be stopped in the SAFE state.

Update: changed @lpd_check to a comment in the SAFE state.

The fitting results for the individual SRX resonance peaks presented in klog:37223 are summarized below. The measurement data are stored in Dropbox → All files/Dropbox KAGRA/Measurements/IFO/SRCL/SRCL_OMC_REFL_beat_signal_LO_sweep/2026_07_14_SRX.
 

• Figure 1: Around 190 MHz, without a linear background. Using data from 194.7 MHz to 195.5 MHz, the fitted peak frequency is 195.1834 ± 0.0034 MHz.
  Figure 2: Around 190 MHz, with a linear background. Using data from 194.7 MHz to 195.5 MHz, the fitted peak frequency is 195.293 ± 0.018 MHz.
   

• Figure 3: Around 160 MHz, without a linear background. Using data from 161.8 MHz to 162.6 MHz, the fitted peak frequency is 162.3010 ± 0.0040 MHz.
  Figure 4: Around 160 MHz, with a linear background. Using data from 161.8 MHz to 162.6 MHz, the fitted peak frequency is 162.415 ± 0.019 MHz.

• Figure 5: Around 140 MHz, without a linear background. Using data from 126.9 MHz to 127.48 MHz, the fitted peak frequency is 127.1931 ± 0.0026 MHz.
  Figure 6: Around 140 MHz, with a linear background. Using data from 126.9 MHz to 127.48 MHz, the fitted peak frequency is 127.2057 ± 0.0026 MHz.
   

• Figure 7: Around −140 MHz, without a linear background. Using data from 125.0 MHz to 125.8 MHz, the fitted peak frequency is 125.3625 ± 0.0061 MHz.
  Figure 8: Around −140 MHz, with a linear background. Using data from 125.0 MHz to 125.8 MHz, the fitted peak frequency is 125.235 ± 0.031 MHz.
   

• Figure 9: Around −160 MHz, without a linear background. Using data from 157.9 MHz to 158.7 MHz, the fitted peak frequency is 158.2421 ± 0.0048 MHz.
  Figure 10: Around −160 MHz, with a linear background. Using data from 157.9 MHz to 158.7 MHz, the fitted peak frequency is 158.1853 ± 0.0091 MHz.
   

• Figure 11: Around −190 MHz, without a linear background. Using data from 193.0 MHz to 193.8 MHz, the fitted peak frequency is 193.3150 ± 0.0033 MHz.
  Figure 12: Around −190 MHz, with a linear background. Using data from 193.0 MHz to 193.8 MHz, the fitted peak frequency is 193.2798 ± 0.0058 MHz.

The fitting results for the individual PRX resonance peaks presented in klog:37223 are summarized below. The measurement data are stored in Dropbox → All files/Dropbox KAGRA/Measurements/IFO/PRCL/PRCL_OMC_REFL_beat_signal_LO_sweep/2026_07_14_PRX.
 

• Figure 1: Around 190 MHz, without a linear background. Using data from 188.4 MHz to 189.0 MHz, the fitted peak frequency is 188.6436 ± 0.0024 MHz.
  Figure 2: Around 190 MHz, with a linear background. Using data from 188.4 MHz to 189.0 MHz, the fitted peak frequency is 188.6809 ± 0.0042 MHz.
   
• Figure 3: Around 160 MHz, without a linear background. Using data from 161.8 MHz to 162.7 MHz, the fitted peak frequency is 162.2408 ± 0.0040 MHz.
  Figure 4: Around 160 MHz, with a linear background. Using data from 161.8 MHz to 162.7 MHz, the fitted peak frequency is 162.3162 ± 0.0068 MHz.
   
• Figure 5: Around 140 MHz, without a linear background. Using data from 128.9 MHz to 129.8 MHz, the fitted peak frequency is 129.3171 ± 0.0033 MHz.
  Figure 6: Around 140 MHz, with a linear background. Using data from 128.9 MHz to 129.8 MHz, the fitted peak frequency is 129.3037 ± 0.0070 MHz.
   
• Figure 7: Around −140 MHz, without a linear background. Using data from 125.1 MHz to 125.7 MHz, the fitted peak frequency is 125.3904 ± 0.0041 MHz.
  Figure 8: Around −140 MHz, with a linear background. Using data from 125.1 MHz to 125.7 MHz, the fitted peak frequency is 125.346 ± 0.011 MHz.
   
• Figure 9: Around −160 MHz, without a linear background. Using data from 157.9 MHz to 158.7 MHz, the fitted peak frequency is 158.3322 ± 0.0035 MHz.
  Figure 10: Around −160 MHz, with a linear background. Using data from 157.9 MHz to 158.7 MHz, the fitted peak frequency is 158.2937 ± 0.0089 MHz.
   
• Figure 11: Around −190 MHz, without a linear background. Using data from 186.4 MHz to 187.2 MHz, the fitted peak frequency is 186.8475 ± 0.0060 MHz.
  Figure 12: Around −190 MHz, with a linear background. Using data from 186.4 MHz to 187.2 MHz, the fitted peak frequency is 186.7790 ± 0.0078 MHz.

The PRX and SRX data measured in klog:37209 were analyzed. The cavity lengths were determined by fitting the measured resonance frequencies. The differences between the measured and design cavity lengths were 1.25 ± 0.46 cm for PRX and 2.56 ± 0.45 cm for SRX. Since these differences exceed the estimated uncertainties, the actual PRX and SRX cavity lengths may differ from their design values.
 

• The cavity lengths of PRX and SRX were determined using the following data.

  PRX
  Minimum (MHz)    Maximum (MHz)
  188.6412    188.6851
  162.2368    162.3230
  129.2967    129.3204
  -125.3945    -125.3351
  -158.3357    -158.2847
  -186.8535    -186.7711
  SRX
  Minimum (MHz)    Maximum (MHz)
  195.1800    195.3102
  162.2970    162.4341
  127.1905    127.2084
  -125.3685    -125.2040
  -158.2469    -158.1762
  -193.3183    -193.2740
   

• From the measurement results, the midpoint between the minimum and maximum frequencies was calculated for each resonance. These midpoint frequencies were divided by the FSR calculated from the design cavity lengths (68.2563 m for PRX and 68.2562 m for SRX). The resulting values were rounded to the nearest integers, and the measured frequencies were fitted with the linear function AN+B, where A and B are fitting parameters and N is the rounded integer. The fitting results were as follows.

  PRX (Fig. 1)
  A: 2.19568 ± 0.00015 MHz
  B: −0.2170 ± 0.0099 MHz
  SRX (Fig. 2)
  A: 2.19526 ± 0.00014 MHz
  B: −0.1252 ± 0.0092 MHz
   

• Since A corresponds to the FSR, the cavity lengths were calculated from the fitted values of A.

  PRX
  Measured cavity length: 68.2688 ± 0.0046 m
  Design value: 68.2563 m
  Difference (measured − design): 1.25 ± 0.46 cm
  SRX
  Measured cavity length: 68.2818 ± 0.0045 m
  Design value: 68.2562 m
  Difference (measured − design): 2.56 ± 0.45 cm

  Therefore, the differences between the measured and design values for both PRX and SRX exceed the estimated uncertainties, suggesting that the actual cavity lengths may differ from their design values.

The SRM GRD says "GAS is close to saturation (F0)".

The K1:VIS-SRM_F0_COILOUTF_GAS_OUTPUT is ~25100 counts now.

I updated the Pcal guardian code based on the preparation report from Jinshui Tian and Yuli Liang.

We incorporated the minor improvements discussed so far.
A detailed report will follow.

Tanaka, Aritomi

We succeeded in locking DRMI with 1F signals and also in engaging ADS for DRMI (IMMT2, PRM, BS, and SRM). We measured Open loop TFs of DRMI DoFs (PRCL, SRCL, MICH). Current Lock duration is at most 20 mins. However, the lock can be restored automatically soon. According to the results, there seem to be some couplings between the DoFs. Anyway, It is enough duration to characterize DRMI more finely.

### DRMI 1F locked

We checked the current DRMI lock with 1f and 3f mixtur configuration (PRCL:3f, MICH, SRCL:1f, (ref. Takano-san's summary slide)). DRMI could be locked with this configuration but some offsets are necessary as setpoints of errro signals. And lock duration seemed to be short, less than 10 sec. We performed try and error to improve the duration. Eventually, we tried to change PRCL error signal from 3f (REFL135I) to 1f (POP45I) and succeeded in DRMI lock with 1f signal. This time, we changed the sign (3f:+ -> 1f:-) and increased the gain (3f:1 -> 1f:30). Also, we implemented the roll-off filters in loops. Moreover, There seems to be some offsets in error signals

After that, we measured OLTFs of their loops.

Fig.1 shows the PRCL OLTF, red shows the one in locking DRMI with the 1f signal and brue shows the one in locking PRMI with the 1f signal. According to KAGRA schnupp asymmetry, RF45 at AS port should be dark. So the PRCL response should not be changed if we use RF45 for PRCL, that is, the PRCL OLTF in locking DRMI should not be changed from the PRMI one. In this sense, the shape of red gain TF seems to be almost same as the brue one. However, as for the overall gain, we did not tuned finely. So the change of the overall gain is fine. On the other hands, phase TF seems to be changed from the PRMI one. This indicates there are some coupling between PRCL and other DoFs for some reason (MICH detune? upper/lower sideband asymmetry?)

Fig.2 and 3 show the MICH and SRCL OLTFs, respectively. Reds are the ones in locking DRMI, and blues are the ones in locking SRMI. OLTF could not be measued well maybe due to large coupling between DoFs. 

### ADS implementation for DRMI

After that, we tried to implement ADS for DRMI. I expected that the response of the RF45 build up power, POP90 shoud not be changed by the SRM exsitance. In this case, I assumed that only SRM ADS should be tuned and tuning for others is not necessary. I chosed AS34 as SRM ADS signal in locking DRMI and performed the phasing of the signals. Then we obtained the demod. phases of SRM ADS PIT:-174 deg(Fig.4), and YAW:170 deg(Fig.5). Thanks to the phasing, Q signals seems to become 0 after the phasing (Fig.6). They seem to be fine.

I engaged SRM ADS at the same time as BS, PRM, and IMMT2 ADSs after DRMI was locked. Fig. 7 shows the time series of ADS error/feedback signals and build up power in PRC (POP90) and SRC(AS34). After calming down the feedback signals, especially SRM YAW, POP90 and AS34 seems to be improved.

Now, the lock duratiion of DRMI with ADS is at most 20 mins (fig.8). For now, we do not investigate the cause of lock loss. Tomorrow, we will perform the DRMI characterization.

The fitting results for the individual PRY resonance peaks presented in klog:37209 are summarized below. The measurement data are stored in Dropbox → All files/Dropbox KAGRA/Measurements/IFO/PRCL/PRCL_OMC_REFL_beat_signal_LO_sweep/2026_07_14_PRY.

• Figure 1: Around 190 MHz, without a linear background. Using data from 186.5 MHz to 187.4 MHz, the fitted peak frequency is 186.9075 ± 0.0027 MHz.
  Figure 2: Around 190 MHz, with a linear background. Using data from 186.5 MHz to 187.4 MHz, the fitted peak frequency is 186.9512 ± 0.0049 MHz.
   
• Figure 3: Around 160 MHz, without a linear background. Using data from 163.4 MHz to 164.4 MHz, the fitted peak frequency is 163.8465 ± 0.0039 MHz.
  Figure 4: Around 160 MHz, with a linear background. Using data from 163.4 MHz to 164.4 MHz, the fitted peak frequency is 163.8930 ± 0.0044 MHz.
   
• Figure 5: Around 140 MHz, without a linear background. Using data from 131.05 MHz to 132.05 MHz, the fitted peak frequency is 131.4789 ± 0.0028 MHz.
  Figure 6: Around 140 MHz, with a linear background. Using data from 131.05 MHz to 132.05 MHz, the fitted peak frequency is 131.506 ± 0.010 MHz.
   
• Figure 7: Around −140 MHz, without a linear background. Using data from 115 MHz to 116 MHz, the fitted peak frequency is 115.5063 ± 0.0042 MHz.
  Figure 8: Around −140 MHz, with a linear background. Using data from 115 MHz to 116 MHz, the fitted peak frequency is 115.4978 ± 0.0068 MHz.
   
• Figure 9: Around −160 MHz, without a linear background. Using data from 156.6 MHz to 157.6 MHz, the fitted peak frequency is 157.0631 ± 0.0042 MHz.
  Figure 10: Around −160 MHz, with a linear background. Using data from 156.6 MHz to 157.6 MHz, the fitted peak frequency is 157.0710 ± 0.0098 MHz.
   
• Figure 11: Around −190 MHz, without a linear background. Using data from 189.1 MHz to 189.8 MHz, the fitted peak frequency is 189.3984 ± 0.0065 MHz.
  Figure 12: Around −190 MHz, with a linear background. Using data from 189.1 MHz to 189.8 MHz, the fitted peak frequency is 189.363 ± 0.018 MHz.