Operators name: Hirose, Aoumi

Shift time: 9-17 JST

Check Items:
VAC: No issues were found.
CRY cooler: No issues were found.
Compressor: No issues were found.
Temp check: (10:00)There is a temperature change in PSL in general, FIELD_IYA, COOLER_BS.
VIS GAS filter output check: No issues were found.

IFO state (JST):
09:00 Start of shift, STANDBY remains
17:00 End of shift, STANDBY remains

PSL:Uchiyama, Tanaka, remote: Ushiba, Miyakawa

## Abstract

We took over the investigation in klog35365. We perfomed some investigaion (polarization, power, temperature) but we failed to recover the 3rd fiber amp. laser today. 

## What we did

### Monitor seed laser output

This morinig I heard from Miyakawa-san that the fiber amp. fall down during increasing the output power and then the error massage "Master fault, Seed Power is too low" was poped up in the controller screen. According to p.25-26 in the manual (JGWdoc, need usual password), when the "Seed" value gets less than 1.1 V, the fiber amp. this results low seed power fault. So we checked whether the seed output was stable or not, at first. 

As a preparation, we connected a T connecter at the PD output which is used for the interlock and the one port of T connecter was connected to the interlock and the other port was conneceted to DGS to be able to monitor the PD output with DGS. In this time, we used the channel, which was used to monitor a RefCAV trans. power, K1:PSL-REFCAV_TRANS, as the PD output monitor. Also, K1:PSL-REFCAV_TRANS_OUT seems to be calibrated from DGS cnts to V. 

Then, we turned on just the seed laser. According to Uchiyama-san, in this early morining, the laser fall down with 7 mins (fig.1). So we monitored the output more than 7 mins. Fig.2 shows the trend of the PD output, K1:PSL-REFCAV_TRANS_OUT. This value at the vertical axis is calibrated to V. In the case of turing on the seed laser, the output is about 2.58V. The amount of the output seems to be fluctuating in +/- several percents. However, the power don't seems to drop down to 0 or the interlock threshold value, 1.1 V within more than 10 mins. Therefore, the seed power at PD seems to be stable enough.

Similarly, we checked whether the actual input power to the fiber amplifier was stable or not. We turned off the seed laser and disconnected the fiber from the input port on the amplifier. We noitced that there seemed to be a small black dot on the edge of the fiber (fig.3). So we cleaned the edge of the fiber with a fiber cleaner (OPTIPOP R2) and then we succeeded in removing the black dot (fig.4). We checked the power before and after cleaning but the power was not changed by the cleaning. 

After cleaning the fiber, we connected the fiber to the PD and monitored the actual input power to the amplifier. Fig.5 shows the trend of the input power. The PD output of the input power is ~0.25 V. This is because we used a 90:10 fiber splliter (Thorlabs, PN1064R2A1) and we used the 10% side for the input power. Therefore, the PD output with the input power, 0.25 V is consistent with 1/9 of the PD output power, 2.6/9 =0.289 V in the order. From the result, the input power seems to be also stable within more than 10 mins. So the input power seems to be stable enough.

(Note: we cleaned the other edge of the input fiber. But the situation was not changed.)

### trial of 1 W emission

Secondly, we tried to emit the laser from the fiber amplifier. In this time, any current for the amp. was not applied, that is, we wanted to check the fiber amp. could emit 1W stablely. We turned on the fiber amplifier and emitted the laser from the fiber output. Fig. 6 shows the trend of the PD output for the interlock, corresponding to the input power of the fiber amplifier (K1:PSL-REFCAV_TRANS_OUTPUT) and the fiber output power (K1:LAS-POW_FIB_OUT_DQ). We noticed the "Seed" value in the controller screen seems to be fluctuated largely and the fiber output power seems to decrease gradually even though the fiber input power seems not to be changed. And then, when the Seed value got less than 1.1, the laser fall down and the error massage "Master fault, Seed Power is too low" was popped up in the screen. At that time, we thought there was something wrong for this Seed value. We tried the laser emission again to check that the phenomenon was reproducable or not. In the second trial, the fiber output power decreased like the first trial but the Seed value was not changed. Then, the laser fall down and the only error message "Master fault" was popped up. The emission duration is typically less than 1 mins.

Unfortunately, there is no detail explanation about "Master fault" in the manual. Maybe, if any fault happens in the fiber amplifier, the "Master fault" message is popped up according to little information in the manual. 

### polarization check

According to the datasheet and the manual, a slow axis polarization should be used as the input beam polarization. So we checked the polarization of the actual input beam by using PBS (Thorlabs, PBSW-1064). We illminated PBS with the beam from the fiber. The beam was relfected from PBS when the fiber key is the top side of the fiber.  Since Thorlabs PBS reflected s-polarization beam mainly, the input beam polarization is s-pol, that is, slow-axis. So the input polarization is not wrong.

### power check

From the PD output with the input power, 0.25V and the PD efficeincy. Ushiba-san estimated the input power on PD is 1 mW or less. This value is not consistent with the previous measurement in klog35344, 4.5 mW. And also, 1 mW is close to lower limt to operate the fiber amplifer. The value of the power meter in PSL(Newport, 843-R), which is used previously seems to depend on the beam position on the power meter. Therefore, we brought the other power meter (Thorlabs, S121C) from outside. And we measured powers at some points. The results are summarized in following table.

The output power from Thorlabs's power meter seems not to depend on the position on the power meter. In this sense, the power with Thorlabs power meter may be more reliable than the Newport one. On the other hand, the input power, 2 mW seems to be still inconsistent with Ushiba-san's estimation from the PD output. However, the input power should be lower than our previous expectation. Therefore, we increased the input power by increasing the current of the seed laser from 1.1 A to 1.3A (The upper limit is 1.4 A). When the current applied to the seed laser is 1.3 A, the input power increased to 3.7 mW by the Thorlabs power meter. Even If the true power is half of the measured one with Thorlabs power meter, the input power maybe large enough to operate the fiber amplifier. So we tried to emitt laser. Unfortunately, the situation was not changed (fig.7).

### Switich seed laser

We switched the seed laser to the spare one because the current applied the seed laser could be increased to 2 A if my memory corrected. However, my memory was wrong. the spare one could be increased to 1.7 A. Furthermore, when the applied current to the spare seed laser is 1.7 A, the input power is only 3 mW. Anyway, we tried to emitt but the situation was not changed.

### Change the temperature of the fiber amplifier 

According to the datasheet, the nominal temperature for the operation is 23 degrees. But the current temperature was. 21 degrees. This tempereture is close to the minimum value for the operation. So we increased the chiller setting from 19 degrees to 21 degrees. We waited for increaing the temperature. After the fiber amplifier temperature in the controller screen got 23 degrees,  we tried the emisstion. the emission duration becomes longer, 1 mins 45 seconds but the power seems to be unstable (fig.8). We tried it again but the laser fall down soon (fig.9).

---

We gave up to recover the laser today. We turned off the whole laser system at last.

## Next

• switch the seed laser for NeoLASE
• try to recovery the 1st or 2nd fiber amp.

Currently (2025/10/17) there is missing data causing "gaps" in some of the KAGRA low-latency gwf files. These gaps will cause I/O error when reading strain data in the science mode from these files. I used lalfr-dump to find all the gaps in KAGRA O4a low-latency strain channel and masked the gaps from the current science segments. By reading strain channel in the masked science segments, we can avoid I/O error.

The gaps will be filled after the regeneration of KAGRA O4a data is done.

The gaps, masked science segments and the details of the generation of the masked science segments can be found here: JGW-G2516947. The scripts used and the segment file can be found on Main Data Server (kmst2-01), please refer to the slides.

(Uchiyama, Miyakawa)

We investigated the cause of the laser down in the PSL room this morning.

The first error was "Seed Signal Too Low. Master Fault.". At this point, the maser was ON. First, we turned the amplifier off, unplugged the power cable and USB cable, waited several minutes, and engaged the amplifier again, but the same error message "low seed power "showed up.

Then we increased the master current from 1.1A to 1.2A, and engaged the amplifier again, and it worked. We increased the power up to ~30W, and locked the PMC. However,  during the lock of IMC, the amplifier went down, and the error message " Master Fault" showed up. Meanwhile of this process, we measured the master power at the DC PD for interlock. The value of this PD yesterday was 2.4V vs 1.1A, but today it was 2.7V vs 1.2A, and 2.1V vs 1.1A. It became lower, obviously.

Then we repeated the same thing again with 1.2A, and the amplifier went down during increasing the power to ~30W (actually went down when close to 30W). We did not have enough time this morning, so we stopped at this point, and we turned off both the amplifier and master. We measured the PD voltage for the master laser again at after shutdown, the result was 3.1V vs 1.2A, and 2.4V vs 1.1A, so the power was back.

I performed the lockloss investigation for the lockloss which happened between 2025/10/16. The previous lockloss investigation was posted in klog35276.

• 1444666741.4375 (duration 17915 s)
  • As reported in klog, K1:ASC-PRC2_P_OUT_DQ shows the drfit from +0.5 to -1.5. (fig1)
  • K1:ASC-DSOFT_P_IN1_DQ shows the oscillation just before the lockloss with 0.039 Hz. (fig2)
  • The ASC controls for arms were stable.  
• 1444670416.3125 (duration 928 s)
  • This lock was not enough long. So the drfit of PRC2 PIT was not related to the lockloss.
  • The IR trans gradually droped over 12 s.
  • no earthquake
  • Figure 3 shows the large amplitude on K1:LSC-DARM_OUT_DQ at -30s. K1:ASC-DSOFT_P_OUT_DQ shows the coincident behavior (Figure 4). The similar behaviour can be seen in PRCL (Figure 5).

I took the TCam photos of ITMX and ITMY for commissioning at 08:39 ~ 08:47 this morning. The previous work is klog35302.
Because the laser was off and PMC and PLL were unlocked (klog35362), I skipped taking photo for ETMX and ETMY. For ITMX and ITMY, we don't need to update the reference positions.

I found the main FIB laser got down around 2:30am this morning.

(There was not significant earthquake.)

A CAL Tcam session was performed to obtain beam position information necessary for Pcal. The parameters have already been updated, and SDF has already been accepted.
During this work, we adjusted Pcal-X beam position by using PCAL_EX2 pico. The related SDF has been accepted.
The RxPD output is backed to usual values form the changed value reported on klog35342.

A CAL Tcam session was performed to obtain beam position information necessary for Pcal. The parameters have already been updated, and SDF is expected to be accepted.

Operator: Dan Chen

Update Time: 2025/10/17 08:05:30

Update Time: 2025/10/17 08:06:16

We accepted the following SDF changes caused by klog35360.

CALEX, CALEY

K1:CAL-PCAL_{EX,EY}_TCAM_{PATH1,PATH2}_{X,Y}

SDFMANAGE

K1:PICO-PCAL_EX2_{1,2,3}_TARGET_POSITION

Since PR3 ASC seems unstable, I keep IFO at PRFPMI_RF_LOCKED state to see the stability of ASC over night today.

Since the DGS output for DC centering of REFL WFS2 PIT was saturated, I moved pico motor for offloading.
I moved REF_WFS2:PIT pico motor fom 400 to 0 for the work.

I adjusted the REFL HWP angle to optimize CAMR loop UGF after increasing the laser power.
Thanks to the increase of fiber laser output (klog35349), IMC output can be increased upto 14.8W, so I adjusted the REFL HWP angle with that condition.

Figure 1 shows the OLTF of CARM after HWP angle ajustment.
UGF is 57kHz and the phase marging is 34 degrees, which is almost same as before.
For achieving above condition, I changed REFL HWP angle from 155 to 157 as shown in fig2.

I found that GRX transmission sometimes varies a lot when PNCX was engaged as shown in fig1.
Figure 2 shows the same signals when handover was succeeded.
Though some fluctuation of GRX transmission power can be seen even when the handover was succeeded, fluctuation amplitude is sometimes very large as shown in fig1 and causes the lockloss.
I'm not so sure the reason but is it related to the initial GRX alignment conditions?

I measured OLTF of CARM when PRFPMI was locked with RF signals (fig1).
Thin line shows the OLTF with original IMC CMS IN1 gain and thick line represents that with IMC CMS IN1 gain reduced by 2dB.
Since CARM OLTF has a large bump at 200kHz with the original IMC CMS IN1 gain due to gain peaking, it shuld be reduced slightly.

So, I modified the IMC guardian so that the nominal gain becomes 2dB smaller (fig2).
To keep the relative gain between IN1 and IN2, I also reduced IN2 gain by 2dB: final IN2 gain is 0dB.
Since offset compensation at 0dB was not implemented yet, I measured it and updated the values between 0dB and 7dB (fig3).
Since Moku:Lab is not connected, I didn't measure the OLTF of IMC loop with new gain but the UGF and phase margig should be 140kHz and 35degrees, respectively, according to the previous measurement (klog35340).

I also measured the MCL loop crosspver frequency and no significant change was observed, so I kept the current gain for MCL loop.

Operators: Aso, Washimi
Shift time: 9:00-17:00 JST

Check Items:
VAC: No issues were found.
CRY cooler: No issues were found.
Compressor: No issues were found.
Temp check: (at 10 AM) PSL; -0.4, IYA; +0.4 (at 4 PM) No issues were found.
GAS filter output: No issues were found.

IFO state (JST):
9:00 STANDBY
13:56 M 4.8 Earthquake at West Chile Rise
14:48 M 6.5 Earthquake in Indonesia
17:00 STANDBY

After increasing the laser power (klog35349), we attempted to resolve the mode-hopping phenomenon reported in klog35344 by adjusting the temperature of the main IR crystal.  This time, we unlocked the IMC before making the temperature adjustments.

Figure 1 shows the original IR crystal temperature before the adjustment: 46.5°C.  We first decreased the temperature by 0.5°C from the original value, but this made the mode-hopping phenomenon even worse.  We then tried increasing the temperature, but could not go higher because the dial had already reached its upper limit.  Therefore, we further decreased the temperature by more than 1°C from the original setting.  Figure 2 shows the IR crystal temperature after adjustment: approximately 45°C. Figure 3 shows the PMC transmission power.  Before the adjustment (at the right cursor), large fluctuations were observed.  After the adjustment (left cursor), the transmission power appears to have stabilized.

Following this, PLLX and PLLY locked automatically via their respective guardians.  However, the TEMP BIAS offsets for both systems increased from 1.0 to ~1.5.  In this state, the GRX temperature was 30.1°C and the GRY temperature was 27.2°C, as shown in Figure 4.  We set both TEMP BIAS offsets back to 0 and manually adjusted each temperature to the desired value.

Note: We modified the PLL guardian so that the scan now starts from -1 instead of –2, since we know the good offset is typically around 0.

Abstract:

There are two lockloss due to the drift of PR3 pitch, which implies that PRC2 PIT ASC becomes unstable again.

Detail:

We succeeded to reach PRFPMI_RF_LOCKED with new fiber laser.
At this moment, we achieved RF lock two times but the lockloss for both of them seemed triggered due to PR3 pitch alignment drft.
Figure 1 and 2 show the alignment of XARM when the two locklosses happened (2025/10/16 04:11:05 UTC and 2025/10/16 05:47:21 UTC).
Left and right T cursors show the time when starting PR3 pitch drift and happening locklosses.
In both cases, PR3 pitch drift seemed to start well in advance, which implies that the instability of PR3 ASC.

>The 30 kHz LPF provided better roll-off and a sufficient phase margin, so we adopted this configuration.

A passive 1-pole (RC) 30kHz LPF is enough, low-noise and easy to make. It is better to ask Shimode-san to make it housed in a box with BNC connectors. For future usage, HV type capacitor should be used. I have it in Shimode-san's room. 

Since nominal IMC CMS IN1 gain was changed due to the work reported in original post, I changed IMC CMS IN2 gain and LSC-MCL servo.
I changed IMC CMS IN2 gain from 7dB to 2dB because IN1 gain was changed frm 23dB to 18dB.
For MCL, I reduced the gain by -1.8dB so that crossover frequency of MCL keeps 100Hz.
Figure 1 shows the crossover frequency measurement result after gain adjustment, which seems fine.

Miyakawa, Tanaka

We increased fiber output power to ~30 W by increasing the current applied fiber LD to 30 A in order to compensate the optical gain. The current fiber output is ~31 W.

I have fixed some troubles in the Road41 monitoring page. It failed to update due to the layout change of the weather forecast page.

To treat similar problems in the future, I have made it to show the problem as a text in the plot.