Shindo at KAGRA was -0.5

[Kimura and M. Takahashi]
 While organizing the regular inspection report every Friday, weI found an anomaly with EYV's ion pump. 
 Attached is a picture of the Error displayed on the LCD panel of the ion pump power supply.
(Photo 1)
The Error 41 displayed is not mentioned in the attached instruction manual and details are not available.
 We are inquiring with the manufacturer about the details of the Error.
Kohguchi staff reported that the area around the ion pump power supply was wet.
We cannot rule out the possibility that the ion pump power supply failed due to dripping water from the ceiling.
 To prevent a secondary disaster, the circuit breaker named "IP-EYV" supplying AC power to the ion pump power supply was disconnected.
(Photo 2)
 The recovery of the EYV ion pump is planned for the afternoon of April 21.
Since there is no spare power supply and high voltage cable for replacement, the power supply and high voltage cable will be diverted from the out-of-service ion pump at X-end.

The measurement records of the Q-mass connected to the EYC were examined backward for 7 days.
The results showed no significant pressure changes due to the ion pump shutdown.

[Ushiba, Komori]

In order to reduce the noise around 100 Hz, we attempted to tweak the alignment of mirrors that had not yet been adjusted in klog:33324—specifically, IMMT1 and PR2.
These mirrors are not controlled by any WFS signals and are currently aligned only by their local control loops.

We varied the IMMT1 pitch by ±50 µrad, but observed no significant improvement or degradation in the sensitivity.
Our next target will be PR2 pitch adjustment.
However, we suspended the work due to the large earthquake.

I tweaked CARM offset (K1:LSC-CARM_SERVO_OFS_COM_CALI_OFFSET) from -40 to +40 with a step of 10 to find the good offset.
Figure 1 shows the spectrum with the various CARM offset (offset value can be seen in the legend of the graph).
The best sensitivity curve was also plotted for the reference.
There seems no significant change by tweaking CARM offset, so CARM offset adjustment doesn't seem effective.

[Kimura, M. Tajkahashi, H. Sawada and Yamaguchi]
 We removed two flexible tubes for Xer cryo-cooler at X-end.
The working time for the flexible  tube removal was from 1:30 p.m. to 2:30 p.m on April 18.
 

It is easy to engage band-stop filter around 59Hz, I impleented it (6-th order eliptic band-stop filter from 47Hz to 61Hz).
As we expected, 59Hz peak was disappeared as shown in fig1.

As reported in klog#33422 and klog#33423, long term data is lost when restarting daqd is executed at the same time as writing down gwf files.
In the minutes-trend case, a duration of gwf file is 3600s and conflict a timing of restarting and writing makes to lose 3600s long data even if a taking time for daqd restart is only a few minutes.
So I inserted a waiting time to complete a file dump process when restarting daqd is requested around a timing of writing gwf files.
It should remove a data lost issue such as klog#33422.

----
Minute trends is written every hour around 59:00 ~ 00:15. And also, it takes ~2-3 minutes to resume daqd process of the frame writer after restarting daqd. In order to avoid execution of daqd restart at the same time of writing minute trend files, I added a proper waiting time when restarting daqd is requested between 56:00 and 0:25 in every hours. In the case of requesting at another time, SYS_DAQ guardian works in the same manner as before.

A same discussion can be done for second trend. But a duration of second trend files is 600s. So in order to avoid losing second trend data, a restricted time of restarting daqd is required as 5 minutes in every 10 minutes. (For minute trend, 5 minutes in every hour is enough as a restricted time.) This restriction seems to be so inconvenient and I passed to implement for second trend in this time.

Analysis for

PEM injection using Saker #4 IFI shaker
2025/04/18 7:10:00 - 07:20:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_12_EXC
100 - 150 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt
Fig.4. showed the results.

Analysis for

PEM injection using Saker #3 IFI shaker
2025/04/18 6:59:00 - 07:09:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_11_EXC
100 - 150 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt
Fig.3. showed the results.

Analysis for

PEM injection using Saker #2 IFI shaker
2025/04/18 6:48:00 - 06:58:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_10_EXC
100 - 150 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt

Date: 2025/04/18

Member: Misato Onishi, Nagisa Sembo, Dan Chen

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.

Analysis for

PEM injection using Saker #1 IFI shaker
2025/04/18 6:37:00 - 06:47:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_9_EXC
100 - 150 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt
Fig.1. showed the results.

The IPC glitch times are removed in the projection.

> I noticed the recording guardian comments "excitation is detected" even the EXC checker was all green.
It's due to klog#33454. The modified code worked as expected in terms of enhancing flexibility during commissioning and reducing mistakes during the observation. But anyway, I forgot to do an after care of recovering MCF0. Sorry for inconvenience.

Request from Ushiba-san

[k1ALS0]
model: k1alspll

[K1IOO]
model: k1imc

Changes
 COMMON_SERVO_MODE/COMMON_MODE_SERVO_ALL_LATCHED_BIO
  Support for turning off the IN{1,2}GAIN path when IN{1,2}ENSW is turned off

Affected locations
K1:LSC-CARM_SERVO_IN{1,2}GAIN related
K1:IMC-SERVO_IN{1,2}GAIN related
 

Summary

This report documents the update of EPICS parameters for the ETMX photon calibrator (Pcal-X) on 2025/4/18. The update reflects the results of a calibration conducted on 2025/4/16, including measurements of photodetector background offsets, optical efficiency, power splitting ratio, and Watt-per-Volt (WpV) conversion factors. The update also includes a correction to the way background values are handled in the update script.

Background on BG Parameter Update

During this update, a bug was identified in the online parameter update script. Previously, the script directly wrote the measured background (BG) values to the EPICS parameters. However, the measured BG values are recorded after the existing BG correction has already been subtracted from the sensor signal. Writing them directly would therefore result in incorrect total correction.

To address this issue, the script was modified to compute the updated BG values as the sum of the currently set BG value and the newly measured residual. The parameter update was carried out using the corrected script.

Updated Parameters

Comparison with Previous Results

Compared to previous calibrations, there were no significant changes observed in the calibration factors of the RxPD and TxPDs.

I noticed the mistake about the fig.8

fig. -> fig.8 (attached figure in this post) shows the results. red is the spectrum when the excitation frequency is 108 Hz, and blue is the one when the exc. freq. is 117 Hz. the bump appear around the related frequency.

Fig.8 -> Fig. 9 shows the whitened spectrograms of CARM in/out -loop sensors (K1:LSC-REFL_PDA1_RF45_Q_OUT_DQ, K1:LSC-REFL_PDA3_RF45_I_ERR_DQ), accelarometer (K1:PEM-PORTABLE_PR_BOOTH_BNC4_OUT_DQ), and the DARM error signal (K1:LSC-DARM_IN1_DQ).

Ushiba, Tanaka

We measured the TF from CARM in out-loop sensor (K1:LSC-REFL_PDA3_RF45_I_ERR) to DARM (fig.1). In this time, due to increasing IN1GAIN from ~-20dB to +10B since the previous measurement, we increased the excitation amplitude to ~30dB. The coherence below 200 Hz is low but the excitation amplitude could not be increased due to the DAC limit.

From the TF, we projected the out-loop CARM to DARM. Fig.2 shows the result. Current residual CARM seems not to limit the DARM sensitivity. As for the region below 200 Hz, it shows a kind of the upper limit but does not limit the sensitivity.

On the other hand, the bump around 520 Hz in DARM seems to be caused  by CARM

• Whatever generated at the input side can be rejected at dark port when the DARM offset can be smaller, though non-linearity effects due to reducnig the DARM offset might happen. 
• Anyway, phase mod component due to the unwanted "cavity" might be thought like frequency-noisy input, then CARM loop may suppress it. Amp mod component might be suppressed by ISS in the same manner...
• If changing the beam spot position on IMMT1 did not show any effects on the fringes, that would indicate the unwanted "cavity" may be formed by some optics within IFI; there might be two or more (unwanted) cavities may be formed, though.
• If the "cavity" made by IFI and PRM, improving the REFL contrast may erase the carrier at REFL and such fringe in RF demod might be gone, though it might show no effects for DARM improvement.
• By the way, if the fringe at REFL can be also found in DC readout, then using the DC fringe vs RF fringe, one can make a phase space trajectory to see, say, what is the unwanted "FP" cavity fluctuation frequency. It is like the attocube sensor.

[Tanaka, Ushiba]

Abstract:

We performed shaker injection from the shaker on the beam duct between IFI and IMM chambers.
Below 113Hz and between 120 and 124Hz, DARM is sensitive to the vibration around the IMM-IFI beam duct but projection is lower than the current sensitivity.
On the other hand, between 114 and 120 Hz, DARM (and CARM sensing noise) is not so sensitive to the vibration around the beam duct.
The bumps in DARM around 108 Hz and 117 Hz come from the other sources, so we need carefully check the behaviour of these bumps separately.
In addition, to investigate the vibration around IFI-IMM area contaminates the current sensitivity, it would be nice to perform the similar injection test with the various excitation point.

Detail:

According to the sweep injection from the shakers on IFI-IMM beam duct, large CARM sensing noise excess can be seen around 108 Hz (klog33458).
There is also an excess around 117Hz though it is smaller than the one at 108Hz.
In addition, we could seem the similar excess in DARM error signals, so we investigated more detail around these frequencies.

Figure 1 shows the DARM spectrum (top left), TF from REFL PDA3 RF45 signals to DARM (top right), coherence between REFL PDA3 RF45 signals and DARM (bottom left), coherence between ACC signals on the beam duct and DARM (bottom left) during the injection.
Red, blue, green, brown, and pin line shows the signals measured with the latest measurement (same as pink) 10-cnt 108-Hz injection, without injection, with 10-cnt 117-Hz injection, and with white noise injection around 117Hz, respectively.

In the data with line injection at 108Hz, strong coherence can be seen between ACC signals and DARM at injected frequency, so there is a linear coupling from vibration at the beam duct to DARM.
On the other hand, there are no significantly large coherence around the injection frequency though large bump can be seen in DARM.
This result implies that there are non-linear coupling from vibration to DARM and also REFL PDA3 signals to DARM around 108Hz.

In the data with line injection at 117Hz, however, strong coherence cannot be seen at injection frequency.
In addition, large broad coherence can be seen between REFL PDA3 RF45 signals to DARM, which implies that vibration makes the broad sidelobe around 117Hz to both DARM and CARM with some mechanism.
In this sense, noise coupling path around 108Hz and 117Hz seems completely different.

In the data with white noise injection around 117Hz, Large coherence can be seen between ACC signals and DARM around  90-115Hz.
So, I calculated the TF from ACC signals to DARM and projected the ACC signals into DARM as shown in fig2, which is well below the current sensitivity.
Of course, one possibility is vibration around IFI-IMM area doen't affect to the current DARM sensitivity but another possibility is current excitation position is just far from the vibration sensitive position.
So, it would be nice if we can perform the similar measurement with various excitation point to confirm if the vibration around IFI-IMM chamber contaminates the current sensitivity or not.

Aso, Ushiba, Tanaka

We found some issues between IFI and PRM as below.

### fringe-like signal in REFL PDs even though all Type-As misaligned

At first step, we confirmed whether the misalignment amount of TypeAs in the MISALIGNED state is enough or not. This morning, Ushiba-san pointed out that the yesterday's misalignment amount of (at least) ITMY got smaller than usual because of lower actuator efficiency by inserting the resistor in TM stages (klog33426), that is, the fringes in REFL PDAs were caused by this unintentional PRY cavity. So we increased the misalignment amounts of both ITMs by rotating BF stages in Yaw direction to ~3  mrads. However, the fringes still remained in this state. Then, we moved each ITM to the L direction one by one to check whether ITM misalignment mount. The fringe-like signals were not changed. This indicated that both ITMs were not cause of the fringe. Similarly, we moved BS and the fringe was not changed. However, we moved PRM or IMMT1 (fig.1, 2) and the fringe responded in either case. Therefore, there seems to be a kind of fabry perot cavity somewhere between IFI and PRM.

We suspected the cavity was consisted of one of IFI component and PRM. In this case, we expected that the fringe got smaller if we tweaked the alignment by changing the beam position on IMMT1. We tried it by sweeping the setpoint of IMMT1 TRANS QPDA1 from -0.5 cnt to 0.5 cnt in each P and Y direction. According to QPDA2, which has almost same gouy phase as QPDA1, the calibration factor from cnts to mm on QPD is 1.6, so we tweaked the beam position from +/- 0.8 mm in both direction. Whenever tweaking the beam position on IMMT1, we aligned IMMT1 and IMMT2 to keep the beam position on PRM to keep the input alignment for PRM. Unfortunately, the fringe seems not to be changed by changing the beam postion (fig.3).

For now, we are not sure that hom much this fringe-like signal affect the current DARM sensitivity.  But, we must remove it untill the future observation.

### Issue of bumps in REFL PDs with PRM single-bounced beam

During above investigation, we found that the 117 Hz bumps in REFL_PDA1_RF45_Q_OUT_DQ with PRM single-bounced beam appered just by misaligning PRM. Fig.4 shows the spectra of REFL_PDA1_RF45_Q_OUT_DQ, when the PRM alignment is nominal (black) and when PRM misaligned to 70 urad in Yaw (brown). This time, we increased the power on REFL_PDA1 from ~10 mW to ~ 30 mW. As you can see, the bump around 117 Hz grew up. Furthermore, other bumps appearded around 140 Hz and around 160 Hz. This bump frequency and shape seem to be consistent with the ones in the DARM sensitivity around these frequencies in this morning (fig.5). On the other hand, the fringe in PDAs got smaller at that time. Therefore, the fringe may not be related to these bumps. This indicates that the scattered light somewhere between IFI and PRM couples to DARM for some reason and the coupling ratio depends on the alignment of the input axis to PRM.

Next, in order to check the effect of the scattered light, we performed the shaker injection with the shaker on the duct between IFI and IMM chambers (K1:PEM-EXCITATION_MCF0_RACK_12_EXC), according to klog33360. We tried to excite the duct at some frequencies, then REFL_PDA signals responded. So we sweeped the excitation frequency from 100 Hz to 500 Hz for 30 mins. This time, the excitation amplitude was 10 cnts. Fig.6 shows the whitened spectrograms of the CARM in-loop sensor (K1:LSC-REFL_PDA1_RF45_Q_OUT_DQ) and the accelarometer next to the shaker (K1:PEM-PORTABLE_PR_BOOTH_BNC4_OUT_DQ) by Pastavi. As for REFL_PDA1, there are some responces just for first 3 mins. Fig.7 is zoomed up one for the first 3 min. As you can see, there seems to be the strongest responce when the excitation frequency is around 108 Hz, and there seems to be the second strongest responce when the exc. frequency is around 117 Hz. So, we excited at these frequencies, one by one and checked the spectrum of REFL_PDA. fig. shows the results. red is the spectrum when the excitation frequency is 108 Hz, and blue is the one when the exc. freq. is 117 Hz. the bump appear around the related frequency.

After that, we locked PRFPMI and then performed the injection with the same procudure, The sweep range of frequency was from 100 Hz to 150 Hz for 10 min. Fig.8 shows the whitened spectrograms of CARM in/out -loop sensors (K1:LSC-REFL_PDA1_RF45_Q_OUT_DQ, K1:LSC-REFL_PDA3_RF45_I_ERR_DQ), accelarometer (K1:PEM-PORTABLE_PR_BOOTH_BNC4_OUT_DQ), and the DARM error signal (K1:LSC-DARM_IN1_DQ). There are resoponces around 108 Hz and 117Hz in PDA3, at the same time, there is the frequency broad responce around 108 Hz and 117 Hz in DARM_IN1. This means that the vibration around the duct in these frequency region may have some effect in DARM.

On the other hands, other frequency region, for example 140 Hz or 160 Hz seems to be not responded. These frequencies may be unable to excite from the shaker on the duct. So we wanna use the shakers at the other points next time.

Analysis for

PEM injection using Saker #4 IMM shaker
2025/04/17 9:23:00 - 09:33:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_12_EXC
100 - 150 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt
Fig.5. showed the result

The projected noise spectrum shape around 116 Hz is similar to the current sensitivity, but its amplitude is smaller.

Analysis for

PEM injection using Saker #3 IMM shaker
2025/04/17 8:42:00 - 09:22:00
Excitation : K1:PEM-EXCITATION_MCF0_RACK_11_EXC
100 - 300 Hz , 1Hz resolution, 10 s in each measurement, 100 cnt
Fig.4. showed the result, there are several linear coupling 200 - 300 Hz

The projected noise is about 2-orders below.