Reports of 27347
VIS (IY)
lucia.trozzo - 17:48 Wednesday 29 May 2024 (29676) Print this report
Comment to Preliminary works to the test the Inertial damping control (29665)

Today I started investigating because I couldn't engage the inertial controls yesterday. I found many issues. First, I noticed a big actuation coupling when I was translating the IP. I decided to decouple the actuation by measuring the decoupling matrix pushing the IP in DC. Here are the old matrix D and the new one D1.

D= [1 0 0.035

0 1 0.206

 0   0    1]

D1= [0.9553 -0.2285 0.0357

   0.2501    0.9282    0.2023       0.1742   -0.1034    1.0004]

I tested the new matrix and it was fine. Then I measured the TFs and saw that there was a sensing coupling into the ACC and GEO (see Pic 1, Pic 2 and Pic 3). I decided to measure the decoupling matrices:

ACC_dec= [1.0052 0.2110 0.0503

          0.0039    1.0008    0.0380             0.1040    0.0218    1.0052]

G_dec=  [ 0.9971 0.1316 0.0901

      -0.0922    1.0307    0.1772           0.1091    0.2552    1.0527]

After implementing these into MEDM, I measured the TFs again and the coupling was redued. I remeasured  all the TFs: GEO/LVDT ACC/LVDT, IS/LVDT to chreck if the phase compensators were still fine and they were the same as yesterday. After that I managed to engage the inertial controls in two configurations:
1) L,T with 50 mHz blending and Y 110 mHz blending
2) L,T 30 mHz blending and Y 110 mHz blending.
The loops were stable. In Pic 4, the H4 spectra, LVDT and IS at IX and IY are shown when both are under inertial controls with the configuration 1. The residual motion of IX and EX at the IP stage is almost the same. Pic 5 shows the H4 spectra, LVDT and IS at IX (50 mHz blending for L and T) and IY when it is under inertial controls with the configuration 2.

Images attached to this comment
DetChar (General)
Chia-Jui Chou - 17:14 Wednesday 29 May 2024 (29675) Print this report
Omicron on K1 O4a Data

This report is based on the presentation in the Detchar meeting on 2024/05/29. Further details can be found here: https://gwdoc.icrr.u-tokyo.ac.jp/cgi-bin/private/DocDB/ShowDocument?docid=15808
We pick the segments in science mode (K1:GRD_SCIENCE_MODE:1) from 1368835218 (2023/05/23 00:00:00 UTC) ~ 1387065618 (2023/12/20 00:00:00 UTC) and apply Omicron with several different parameter settings on the KAGRA strain data (K1:CAL-CS_PROC_DARM_STRAIN_DBL_DQ). Only the segments without IPC glitches and greater than 124 s are chosen.
There are 3,551,222 triggers with SNR >= 5, 38,766 triggers with SNR >= 10, 680 triggers with SNR >= 50. Part of the Q-transform spectrograms of the triggered glitches with SNR >= 50 are shown in the slides. The plots of Q-transform spectrogram can be found here: https://git.ligo.org/agoodmanjerry/k1-o4a-glitches/-/tree/main/O4a-no-ipc/K1?ref_type=heads.
We will apply unsupervised machine learning method on classifying the Q-transform spectrogram of the 38766 triggers with SNR >= 10. Then apply the same methods to the triggers with 6 >= SNR > 10.

Images attached to this report
VIS (OMM)
tatsuki.washimi - 15:41 Wednesday 29 May 2024 (29673) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

I compared the recent results (without additional weight) to the TAMA paper (Rev. Sci. Instrum. 73, 2428–2433 (2002)).

The Q-value is larger (due to harder rubber?) in the recent test and the resonant peaks are sharp.
Also, the number of peaks is in the recent test, but why? (the ancient test should also be a 3-steps stack).

Images attached to this comment
VAC (BS)
takaaki.yokozawa - 14:21 Wednesday 29 May 2024 (29672) Print this report
Status of the vacuum monitor near the BS
I checked the status of vacuum monitor which was recored in
K1:VAC-PRESSURE_CS_BS

Actually, this channel monitored the cc-10 value around PR2 (minus X side as shown in Fig.1.)

When I checked the previous value of this channel, as shown in Fig.2., I noticed the monitor started evening at 16th May and gradually increasing the value, and current value was about 100 Pa.
klog29545 would be the corresponding work.

I checked the status of cc-10 itself using the serial transportation (Both Read and Status commands), no strange setting can be found (I heard Yamat-san also checked it)

And I checked the status of GV near the cc-10, closed between bellows(air) and cc-10 point, opened between cc-10 point and vacuum pump.

So my guess is actually around 100 Pa around cc-10 area near the PR2.
 
Images attached to this report
MIF (General)
takaaki.yokozawa - 12:25 Wednesday 29 May 2024 (29671) Print this report
Now bellows between BS and SR2 was closed.
[Takase, Yokozawa]

As shown in Fig.1., we closed the bellows between BS and SR2 by aluminum foil.
Images attached to this report
VIS (General)
takaaki.yokozawa - 12:06 Wednesday 29 May 2024 (29670) Print this report
Taking photos and oplev path check
[Takase, Yokozawa]

We performed the taking pictures for BS ans SR's, also oplev path check for them.

All pictures were placed in
/kagra/Dropbox/Subsystems/PEM/Picture/20240529/BS
/kagra/Dropbox/Subsystems/PEM/Picture/20240529/SR2
/kagra/Dropbox/Subsystems/PEM/Picture/20240529/SR3
/kagra/Dropbox/Subsystems/PEM/Picture/20240529/SRM
(I forgot to take the photos for PR2 and PR3, should we do it)

Oplev check
BS : Checked the beam path at the viewport, inside the chamber
SR2 : It was quite difficult to access the injection part, so we took pictures from outside of the viewport (Fig.1.)
SR3, SRM : Checked the beam path at the viewport , inside the chamber, and suspended table
Images attached to this report
IOO (IMC)
shinji.miyoki - 10:42 Wednesday 29 May 2024 (29669) Print this report
Comment to IMC blasting resistance check with stomping (29591)

According to the actual tunnel constructor estimation (not the SANKO consulting company), the target speed seems to be 500um/sec in total. Actual blasting by using dynamite consists of 11 layers and will be done in a short time interval. In other words, we will sense 11 times blasting shocks in the timeline. The Max velocity for each blasting seems to be 200 m/sec. 

VIS (OMM)
tatsuki.washimi - 8:09 Wednesday 29 May 2024 (29668) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

[RTakahashi, Washimi]

Yesterday we constructed a vertical shaking setup and tried to measure the transfer functions.
However, the shaking amplitude was very small and SNR was not good, even though we set the signal amplifier gain to be 500 (for lower ACC) and 50  (for upper ACC).

We will try it again in the next week.

Images attached to this comment
VIS (OMM)
tatsuki.washimi - 7:51 Wednesday 29 May 2024 (29667) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

[RTakahashi, Washimi]

Yesterday we investigated some issues for the H->H TF.

We found a strange factor of 2.5 was multiplied for the TF amplitudes of the weight-loaded measurements (due to the GUI application, for plot zooming up).
After canceling it, the DC gain goes to 0 dB. (somehow +63kg data goes below 0 dB)

We also performed additional measurements, with loading +155 kg (without thin plates).

In this result, the strange peaks over 100 Hz are disappeared.

Images attached to this comment
VAC (EYA)
nobuhiro.kimura - 20:17 Tuesday 28 May 2024 (29666) Print this report
Progress report on the EYA vacuum test

[Kimura, M. Takahashi and Sawada (Hokuto)]
 Vacuum pumping began at 10:00 a.m. for the EYA vacuum test.
The following is a record of the valve operation and pressure drop during the vacuum pumping. 

Time Pressure (Pa) Memo.
10:00 9.60E+04 Start pumping. Open the valve about two turns.
10:11 8.80E+04  
10:24 8.50E+04  
10:37 8.10E+04  
10:41 8.10E+04 Open the valve about one turn.
10:48 7.80E+04  
10:54 7.50E+04 Open the valve about one turn.
11:01 7.30E+04  
11:10 7.10E+04  
11:17 6.90E+04 Open the valve about one turn.
11:32 6.60E+04  
12:51 5.80E+04  
13:22 5.10E+04 Open the valve approx. one turn.
14:12 5.50E+04 Pressure drop stops. Failure?

 At about 14:12, we found that the pressure stopped around 0.055 kPa, so we checked the pump and the pressure gauge.
A Bourdon tube pressure gauge was temporarily installed to check the capacity of the vacuum pump alone,
 and it was confirmed that the vacuum pump had failed (Photos 1 and 2). 
 The reason for the failure of the vacuum pump was assumed to be that the shaft seal of the pump was damaged due to the continuous suction of air containing large amounts of water vapor.
 The manual (See attached manual in Japanese) supports opening the ballast valve when suctioning air containing a large amount of water vapor,
 but it is believed that even this ballast valve could not handle the problem.
Therefore, we decided to replace the exhaust pump and resume pumping tomorrow morning.
At that time, we decided to introduce dry nitrogen into the EYA up to atmospheric pressure to remove as much moisture as possible.

Images attached to this report
Non-image files attached to this report
VIS (IY)
lucia.trozzo - 17:06 Tuesday 28 May 2024 (29665) Print this report
Preliminary works to the test the Inertial damping control

Today I started work on implementing inertial damping control at the IP stage.

The inertial sensor consists of a geophone and an accelerometer. IS_{L,T} = ACC_{L,T} * LP + GEO_{L,T} HP. The HP and LP filters have a cut-off frequency of 100 mHz. I measured the TF: LVDT/ACC and LVDT/GEO along L and T to make sure the sensors were calibrated. Pic1, Pic2, Pic3, Pic4 show the results. The amplitude is close to 1 in all the frequency bands of interest and the phase is close to zero from 0.1 Hz to a few Hertz. Below 0.1 Hz we observe a phase shift ranging from a few degrees to almost 25/50 degrees. To solve this, a phase compensator was added to the MEDM as Ph_C and tested. The sensors look calibrated and I was able to build the inertial sensors IS_{L,T}. The phase shift is corrected by the new phase compensator and tested. Then I measured the TF: LVDT/IS{L,T}. It is clear that there is still a phase lag which could cause instability in the loop (see Pic 5 and Pic 6). To reduce the phase lag, an additional phase compensator was implemented on the IS.

As a final step, I tried to engage the inertial control using the 50 mHz blending strategy for L and T and the 110 mHz for Y. When I tried to test the 50 mHz stategy along L and T at the same time, an oscillation at about 100 mHz appeared and made the Y loop unstable.  I don't know where it comes from. I have to investigate further.

Images attached to this report
Comments to this report:
lucia.trozzo - 17:48 Wednesday 29 May 2024 (29676) Print this report

Today I started investigating because I couldn't engage the inertial controls yesterday. I found many issues. First, I noticed a big actuation coupling when I was translating the IP. I decided to decouple the actuation by measuring the decoupling matrix pushing the IP in DC. Here are the old matrix D and the new one D1.

D= [1 0 0.035

0 1 0.206

 0   0    1]

D1= [0.9553 -0.2285 0.0357

   0.2501    0.9282    0.2023       0.1742   -0.1034    1.0004]

I tested the new matrix and it was fine. Then I measured the TFs and saw that there was a sensing coupling into the ACC and GEO (see Pic 1, Pic 2 and Pic 3). I decided to measure the decoupling matrices:

ACC_dec= [1.0052 0.2110 0.0503

          0.0039    1.0008    0.0380             0.1040    0.0218    1.0052]

G_dec=  [ 0.9971 0.1316 0.0901

      -0.0922    1.0307    0.1772           0.1091    0.2552    1.0527]

After implementing these into MEDM, I measured the TFs again and the coupling was redued. I remeasured  all the TFs: GEO/LVDT ACC/LVDT, IS/LVDT to chreck if the phase compensators were still fine and they were the same as yesterday. After that I managed to engage the inertial controls in two configurations:
1) L,T with 50 mHz blending and Y 110 mHz blending
2) L,T 30 mHz blending and Y 110 mHz blending.
The loops were stable. In Pic 4, the H4 spectra, LVDT and IS at IX and IY are shown when both are under inertial controls with the configuration 1. The residual motion of IX and EX at the IP stage is almost the same. Pic 5 shows the H4 spectra, LVDT and IS at IX (50 mHz blending for L and T) and IY when it is under inertial controls with the configuration 2.

Images attached to this comment
VIS (General)
lucia.trozzo - 11:49 Tuesday 28 May 2024 (29664) Print this report
Comment to Cradle effect estimation at IX and EX (29605)

In the last few days, in order to understand if the 22 mHz oscillation was related to a cradle effect or if it was just an accelerometer excess noise at DC, I repeated the test at EX, moving the IP of several mums and looking at the DC variation of the diagonalised accelerometers. Here I will report the results of this test. Figure 1 shows the IP LVDT variations (mum) along L and T, together with the variation of the accelerometer signal (mum/s²). The plot shows that the accelerometer signals are not changing with the IP translations. They show random fluctuations. Indeed, If we were measuring a cradle effect, the accelerometer signal should vary accordingly with the LVDT steps. The data shows this is not happening. The instability at 22 mHz when blending at 30 mHz could be caused by an excess of accelerometer noise at that frequency.

Images attached to this comment
VIS (OMM)
tatsuki.washimi - 9:16 Tuesday 28 May 2024 (29663) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

I guess it is due to the resonant of the top weights, 3 thin metal plates.

Images attached to this comment
VIS (EY)
takafumi.ushiba - 8:36 Tuesday 28 May 2024 (29662) Print this report
Comment to Health check for ETMY (28430)

I made a mistake for fig3.
Attached figure should be fig3 I mentioned in original post.

Images attached to this comment
VIS (OMM)
shinji.miyoki - 7:24 Tuesday 28 May 2024 (29661) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

Do you have any idea on the large peak in the green curve around 120Hz?

VIS (OMM)
tatsuki.washimi - 22:21 Monday 27 May 2024 (29660) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

[RTakahashi, Hirata, Washimi]

We measured the Horizontal -> Horizontal vibration transfer function of the test stack (3 steps) using a shaking stage, by changing the load weight.

The lower accelerometer was TEAC707LFZ (10 mV/(m/s2), amp gain = 100) and the upper accelerometer was TEAC710 (300 mV/(m/s2), amp gain = 10 or 20).
In the plot, the calibration factor and the amplification gain are already canceled.

Changing the resonant frequencies is seen.
The amplitude was larger than 1 below 40 Hz. The shape looks healthy, but is this amplification OK for KAGRA?

Images attached to this comment
VIS (IX)
lucia.trozzo - 21:16 Monday 27 May 2024 (29657) Print this report
Comment to Ineretial damping implementation: preliminary results (29481)

Today I've implemented the inertial damping along L and T using the 50 mHz blending filters. The loops have been tested for several hours without any issues. Fig 1 shows the H4 spectra when L, T  d.o.fs are controlled using the 50 mHz strategy  and the Y Is controlled with the usual 110 mHz strategy (Red curve) while the blue Is the H4 spectra when the IP in controlled whit the DC control.  The 50 mHz strategy reduces the rms by almost a factor of 13 at the microseismic peak. Fig. 2 shows the H4 spectra at IX and EX when both are under inertial controls. The residual motion of IX and EX at the IP stage is almost the same.

Images attached to this comment
VIS (IY)
takafumi.ushiba - 19:50 Monday 27 May 2024 (29659) Print this report
ACC control modification of ITMY

During the work in klog29658, I found ACC might contaminate the photosensor signals.
So, I also checked ITMY photosensor signals and confirmed the same thing happened in ITMY.
Then, I modified ACC local servo in the same manner written in klog29658, and the noise level reduced.

Same modification of ACC local servo should be inplemented in ITMX and ETMX for local sensor noise reduction.

VIS (EY)
takafumi.ushiba - 19:47 Monday 27 May 2024 (29658) Print this report
Comment to Health check for ETMY (28430)

Abstract:

I continued ETMY health check at IM stage.
Still we have several issues to be solved.

Detail:

Figure 1 shows TFs from IM roll excitation.
Though the gain of IM roll to IM roll at low frequency seems lower than before (not so clear due to low coherence), I could not confirm the reason.
Anyway, I hope it is not so problematic because IM stage doesn't use for daming roll motion of the suspension.

Then, I found three strage behaviours:
1. Small signal jump in IMH2 photosensor (fig2), which seems ADC glitches because the jump at raw input is just 1cnt or so. So, it doesn't seem problematic.
2. Large glitcies in IMH3 photosensor (fig3). I have no idea what they are at this moment.
3. Photnsensor noise at high frequency (fig4), which cannot be seen if suspension is in SAFE state (fig4: blue:SAFE, red:ISOLATED). After several investigation, I found the noise has large coherence with ACC local servo feedback (fig5). So, I modified the ACC local servo. Old servo has an UGF of 20 Hz and phse margin is only 30 degrees. Since ACC is used only below 0.1 Hz, I changed UGF at 3 Hz and phse margin is about 70 degrees (fig6). After these modification, photosensor noise at high frequency are disappeared.

Images attached to this comment
DGS (General)
shoichi.oshino - 17:10 Monday 27 May 2024 (29656) Print this report
Comment to Installation of k1nfs2 (29623)
This is the continuous work of installation.
I created a RAID6 volume with 11 HDDs. One HDD is set as a spare.
I then created 50TB volumes to mount on k1nds0 and k1nds1. This capacity can be changed later.
CAL (General)
dan.chen - 17:02 Monday 27 May 2024 (29655) Print this report
Preparation for Tcam data backup

With Ikeda-san, Oshino-san

Summary

Until now, Tcam data had only been stored in data2 of calcenter01 and had not been backed up.
So we decided to make a backup system for this data.
Today, we put two new HDDs in the calcenter01 and mounted them: Tcam_BU_001, Tcam_BU_002.
We plan to use rsync and crontab for backup later.

Details

  1. Checked the current status of calcenter01 about the mounted HDD.
  2. As the new HDDs were not recognized by the calcenter01, we did the followings:
    1. Format the HDDs by another machine.
    2. Cleaned inside of calcenter01 machine. There was a lot of dust in it.
    3. Replaced the CMOS battery.
    4. Then, finally, the calcenter01 recognized the new HDDs.
  3. Mounted the new HDDs on /Tcam_BU_001 and /Tcam_BU_002

Next time, we will test rsync and make crontabs for the backup.

VAC (EYA)
takashi.uchiyama - 16:40 Monday 27 May 2024 (29654) Print this report
Closeing flanges of EYA for the pumping down test
2024/05/27

Kimura, mTakahashi, Sawada, Uchiyama

We closed the flanges of EYA for pumping down.
The flanges are two side flanges, the top chamber, ICF70 at the bottom, and the dia—1000 flange.

Finally, we noticed there was a hole at the center of the dia-1000 flange. So we could not start pumping down yet.
We have already prepared the flanges, gaskets, and bolts to close the hole.
We will close the hole tomorrow morning and then start pumping down.
VAC (EYA)
shinji.miyoki - 15:25 Monday 27 May 2024 (29653) Print this report
Comment to EYA cleaning with cleaning water Day1 (29583)

For memory about water absorption in the space of bellows.

I compared the absorber tube length and the outer/inner height of the bellows as photo1 (outer) and photo2(inner). The length of the absorber tube seems to be enough to reach the bottom of the bellows. The actual absorbing point is around the inner height position for each bellow because we can only access this area from the center hole of the optical bench.

Photo3 and 4 show the absorber tube insert status. Photo3 shows the connecting point of the red plastic tube and silicon tube. It is well inside the bellows. This means that the bottom of the red tube was also around the bottom. The Photo4 is the tube diameter reducer between the larger Nitta plastic tube and the thin silicon tube that was directly connected the red tube.

Images attached to this comment
DGS (General)
takahiro.yamamoto - 15:20 Monday 27 May 2024 (29652) Print this report
renewal of slack bot code
Ushiba-kun asked me how to use "slackpost" function which is used in guardian from the general python scripts such as health check measurements script.
Because of the historical reason, the implementation of "slackpost" is slightly complicated and using it as a python-native function on general python scripts is difficult.
So I implemented the way to use it as a python-native module easily. Now it's available as follows.

import slack
slack.slackpost(channel, message, mention)


-----
Because old slack bot tool is called via subprocess.call as the python executable script in guardian code due to the python version issue.
This issue was already solved by the update of guardian after O3GK. But slack bot code haven't been updated.

Today I refactored the slack bot code in order to call a native-python3 module in guardian code and to use a common library in general python scripts.
A new code is in /kagra/lib/python3/dist-packages/slack and this directory is registered in PYTHONPATH.

For guardian, same code is available in the /opt/rtcds/userapps/release/sys/common/guardian/slack which is symbolic link to /kagra/lib/....
And also a new code is also available via kagralib. So nothing to be done by the end users.
Before loading guardian, a old code is still running. After loading, a new code will run.

We will (I won't have enough time because of other task, so I will ask Ikeda-san) load all guardians in the next Friday.
VIS (OMM)
tatsuki.washimi - 12:51 Monday 27 May 2024 (29651) Print this report
Comment to Stack test @ Mitaka for the OMC (29643)

I calculated the spring constant for the Hs=80 rubbers, from the adding test (on 24) and removing test (on 27).
spring constant [N/mm] = [kg/mm] * 9.8 m/s2 (grav. acc.) / 3 (# of parallel rubbers)

The results were 469 N/mm (adding test) and 572 N/mm (removing test).

Images attached to this comment
Search Help
×

Warning

×