[Takahashi, Sato]

We investigated the cause of the change in the output of MNV3. From the measurement of the resistance between each pin from the feedthrough MNV3 (Port6-4), it was confirmed that there was a short circuit between Pin3 and Pin7.

The short circuit had occurred inside the Dsub cover on the IP. The bare conductor of the Pin3 wire was longer than the others and was close enough to touch Pin7, so Pin3 was reattached to fix the problem.

With Hirata-san.

See pictures in the album BS Remedying work after O3.

Summary: We aimed to calibrate the F1 LVDT with the optical displacement sensor, however, we realized the holder requires adjustment for an additional angular degree of freedom in order to achieve alignment.

The holder allows adjustment of the vertical and one angular degree of freedom. We successfully put the sensor approximately 20 mm above the surface of the BF and, by adjusting those two DoF, we nearly achieved the alignment. Sometimes the sensor came online during a few moments, but it never remained aligned.

Currently, we are thinking of ways to modify the holder.

Abstract

I started the measurements of the Q facters using the excited signal to compare the attenuation condition with / without the damping filter.
I have measured only ITMX Y and L, so there are a lot left, but the script was completed and I just measure in the same way for other suspension (each DOF, resonance frequency), so it'll be over soon ... maybe.

Detail

To know how much the damping filters (Ushiba-san and I made) work, I started to measure the Q factors.
* These are just the first criteria, so I should consider the modal damping (using DMD) in the future.

To get the Q facer, I input the excited signal and cut it. Then, from the attenuation speed, I calculated the Q (fitting with exponential function).
* As for {L and T} (and maybe also {R and P}), their first resonance frequency is very near, so attenuation doesn't go well. So, when I measured L, I turned on the damping filter of T.

1. make the band pass filter for each resonance frequency
2. in DAMP filter bank, turn off the DAMP filter and turn on the BP filter
3. in SUMOUT fiter bank, turn off SW1
4. in DAMP filter bank, input 1.0 in the gain
5. exite the signal (SUMOUT_(DOFs)_EXC) using awggui, then turn off this excitation
6. using the script, calculate the Q factor

However, the real signal is very bumpy (like Fig1), so in my script, the difference between uper peak (Blue point) and lower peak (Purple Point) is used to fit. 
* The script I made (qmeasurement.py) is in /users/tamaki/Qmeasurement.

The fitting resut is shown in Fig2, and the Q is calculated automatically (for example, Q=279.78 for 0.294 Hz resonance of ITMX Y without damping filter).

Now, I have measured ITMX Y (all resonance) and L (ony 1st resonance) without damping fiter.

I installed an NFS server and tried to export an NFS drive to Piconet, but it failed.
It seems that only physical drives can be added to exports.

$ ssh k1script1
# su
# apt -y install nfs-kernel-server
# emacs /etc/exports
/opt/rtcds 10.68.160.0/24(sync,rw,no_root_squash,no_all_squash,no_subtree_check)
# systemctl restart nfs-server
Currently, nfs-server is assumed to be stopped.
# systemctl stop nfs-server

After discussing with Yamamoto-san, we decided to consider another way to handle the issue.
We plan to discuss it again next time next week.

1.k1boot(site) and k1nfs0(analysis building), which are NFS servers, should be physically connected to Piconet's VLAN by LAN cable so that they can refer to NFS.
2. Create /export directory on k1script1 and configure it to be NFS mountable.
 

Currently, the elapsed time of k1pr2, k1pr0, k1bs, k1ex1, k1ey0, and k1test0 is abnormal time.

To recover, I rebooted k1ex1.
$ sudo reboot -r now
The elapsed time returned to normal after the reboot.

The other RTPCs will be performed on another day.
 

Deleted the non-mechanical parts of SR2 and SR3.
However, the dummy data of DAQ channel was added to TOWER_MASTER/TYPEB_TOWER_MASTER because the timing error occurred when deleting it.
To separate Type-B and Bp, Bp side was also moved from TYPEBP_BFMASTER to TYPEBP_TOWER_MASTER.

BS and SRM will be implemented in the next opportunity.

Change in data volume :
K1VISSR{2,3}T Total: 634,624 bytes -> 602,432 bytes (487,744bytes if we can remove the temporary measures)
K1VISSR{2,3}P Total: 401,792 bytes no change

Git: release/0.1.56
GpsTime:1326774018
 

Routine check for DGS maintenance day (1/21)

Stepper motor
 [Checked]
  Check if the power supply can be controlled by BIO.
  Check if the script can be started.
  Confirmed that the limit switch installed is in Green.
 [Check result]
  No problem.

 The following settings were made.
  Change the value of software limit of ETMX and ITMX GAS (E2213706-v3)

 The HW limit switches on the ETMX IP were all red because the cable had been unplugged during the leak test, but later checks showed that they were all green, so no problem.
  

Release Sensor
 [Checked]
 Confirmed that the Release Sensor is 0. 1: Seated
 [Check result]
 ETMX, ETMY, ITMX, ITMY : No problem

Picomotor
 [Checked]
  Ping Test is performed.
 [Check result]
  Known only

 NG: 
  MCI,STM1,IMMT2
  AS_WFS,POP_POM,POS_POM,REFL_WFS,POM1
  MCF(MCO,MCI)
  　=> Confirmation only, no recovery work was done.

 OK:
  ETMX,ETMY,ITMX,ITMY
  BS(IM,BF),SRM(IM,BF),SR2(IM,BF,STM),SR3(IM,BF)
  PRM(IM,BF),PR2(IM,BF),PR3(IM,BF)
  MCE,IMMT1,OMMT1,OMMT2,OSTM

  PCAL(EX1,EX2,EY1,EY2):Not confirmed because it is ON only when used.

Temperature controller
 [Checked]
 Check if the temperature acquisition is working properly.

 [Check result]
  No problem

Precision air processer
 [Checked]
 Is the temperature acquisition function working properly?

 [Check result]
  No problem

[Takahashi, Sato]

We attached the ballast weights onto the proof masses of the accelerometers.

Ikeda, Kenta

In this morning, Ikeda-san restart the k1psliss model and DAQ.

And I made the related medm screens to see the channels of new monitors and add the link for their screens in IMC LSC medm screen (red circle in fig. 1)

So, if you want to see the signals from the monitors, you should click the link.

---

Note

Now, I give the temporal channel name of new power monitor(K1:LAS-POW_PSL2_xx) because there are similar names in some other models and the channels are already used in some guardians. So, we must confirm their complex relationships and sort them...

Abstract:
In-vac health check of ITMX BF stage was done.
The suspension mechanical transfer function at vacuum condition is not significantly changed from that at atmospheric pressure.

Detail:
Figure 1 to 6 shows the TF of BFL, BFT, BFV, BFR, BFP, BFY, respectively.
We can observe the small sprious coupling at high frequencies in TF of BFT but it is probably fine (in fact, it was there even before vacuum evacuation).
Also, the other DOF, including coupling, is not changed due to vacuum evacuation significantly.

Then, I made new suspension model for BF stage.
Figure 7 to 12 shows the model of BFL, BFT, BFV, BFR, BFP, and BFY, respectively (model: red, measured blue).
These models are stored at FM10 (SUSMod_220114) of each filter bank at BF_DAMP.
The model was based on the TFs measured on 14 of January, which are stored at /kagra/Dropbox/Measurements/VIS/PLANT/ITMX/2022/01/PLANT_ITMX_STANDBY_BF_TEST_{L,T,V,R,P,Y}_202201141937.xml.

Ushiba, Nishino, Tamaki, Terrence, Aso

In a nutshell

Now the X-arm can be locked with guardian.

What was wrong ?

The normalization factor for the X-arm transmission in K1:TMS-X_IR_PDA1 was adjusted so that the normalized power becomes 1 or a bit above 1 when locked.

Since the threshold to judge the arm lock is 0.4, this normalization enables the guardian to properly detect the lock acquisition or lock loss.

With the above modification and the modification to the @IO_lock_check decorator (explained below), we were able to lock the X-arm with the guardian.

Temporary modification to the guardian code

We modified the class IO_lock_check to do nothing.

This decorator is supposed to check the status of the MC. It wants the MC to be in the PROVIDING_StABLE_LIGHT state.
The XARM guardian uses this decorator extensively.
However, due to the recent changes in the MC guardian, we are not yet sure if we want to move to this state. So currently, MC is staying in the IMC_LSC_LOCKED state.
This makes the IO_lock_check decorator to fail always.
Therefore, as a temporary measure, we bypassed this decorator.
This change must be reverted when the MC guardian is back to normal.

[Terrence, Ushiba]

Terrence worked for the damping controls of ETMY and found that the transfer function from MNP excitation to MNP motion measured by photosensors seems strange (fig1).
So, I checked the history of the transfer functions and at least TF seems healthy in the last year (fig2).
Then, Ifound that photosnsor V3 has very small signals (about -300 cnts), which should be about -3000 cnts.
Accordingto the photosensor signals, MNV3 photosensor output changes on 12 of January around 15:00 JST (fig3).
On that day, some works around EY was done (klog19421 and klog19425).
So, tower part, especialy at IP stage, needs to be checked.

During working on data quality evaluation, I found that several microphones are missing on summary page. So, I added them on summary page.

And also I found the microphon at EYA booth has a strange spectrum, which looks like just ADC noise.

As I checked by Pastavi, it's happening from sometime in the period between 2021 April 1 and  2021 May 1.

With Hirata-san.

See album SR2 Remedying work after O3.

Summary: The connection is healthy, the resistance is slightly higher than expected because an extension cable was used.

The problem described in the commented entry was solved on the 29th of September 2021 according to notebook notes, but I forgot to write a klog report.  The instability of the resistance value was solved that day, but the value measured at the outside of the chamber remained higher than in other OSEMs, so we investigated this issue today. The measured value is 22.3 Ohm, whereas typical values are ~19.5 Ohm.

As reported before, this line has three segments. The last segment is an additional extension that is not used for connecting other devices. Therefore, we decided to measure the resistance introduced by this extension:

• Resistance along the line connected to pin 2: 1.4 Ohm
• Resistance along the line connected to pin 7: 1.3 Ohm

We checked two spare similar extension cables  and confirmed these values are typical.

At the top of the 2nd segment, the resistance of the coil is 19.7 Ohm, which is the value we were expecting. The addition of the three values is 22.4 Ohm, which is consistent with the value measured from the outside.

There is nothing wrong with the cables.

Today, we can lock the X-arm with IR for more than 2 hours, which is probably long enough to characterize several parameters of IFO.
Since we would ike to use IX suspension, we turned off the control at around 14:30.

Abstract:
In-vac suspension health check and modeling work was started.
As a reault, ITMX IP stage seems healthy after evacuation.
I updated the IP models baed on the measurent on 14 of January.

Detail:
I checked the suspension mechanical transfer function measured on 14 of January (klog19475) and compared the TF measured efore vacuum evacuation.
Figure 1, 2, and 3 show the TFs of IPL, IPT, IPY, respectively.
Even though resonant frequencies are slightly changed from the measurement before evacuation, there seems no significant changes except for that; IP seems healthy.

So, I modified the suspension models of IP, which is made at FM10 of IDAMP filter banks (SUSMod_220114).
Fgure 4, 5, and 6 shows the comparison between models (red) and measured one (blue) of IPL, IPT, and IPY, respectivaly.
The original diaggui files are /kagra/Dropbox/Measurements/VIS/PLANT/ITMX/2022/01/PLANT_ITMX_STANDBY_IP_TEST_{L,T,Y}_202201141937.xml.

Note:
Since the two resonat frequencies in IPY TF are very close after evacuation and difficult to distinguish them, I just set one resonance around 0.43 Hz.

Date: 2022/1/19

Members: Dan Chen, Takayuki Tomaru

We made jigs to fix LED in Tx/Rx module.

And we fixed LEDs in Tx/Rx module of XPcal.

What we checked after the LED installation:

• remote control of the LEDs
• Pcal beam alignment in Tx and Rx module
• relationship between Pcal beam height and LED position 

These look good.
The attached pics are inside of Tx and Rx module

Ushiba, Tamaki, Nishino, Yasui, Yano, Aso

In short, X-arm can be locked with IR by feeding back to the laser frequency.

Today's work

After removing the target plate from PR2, we scanned IMMT2 and PR2 to find a good alignment. We found a reasonably good alignment giving a few thousand counts in transmission.

Then we constructed a temporary beam path for REFL by putting a PYD-20 on the temporary breadboard on the MCF chamber.

Initially, we were puzzled by the weakness of the PDH signal from REFL PDA1. It turned out that the half-wave plate needed to be rotated to increase the light power reaching the PD.
Since the polarizer mount cannot be controlled remotely at this moment, we moved it by hand.

After this adjustment, we were able to lock the X-arm by following the steps in the XARM guardian code manually.

We should be able to also run the guardian to lock the X-arm. However, it fails at this moment because the threshold to judge the lock of the arm is wrong. We plan to fix this tomorrow.

[Takahashi, Sato]

We installed the accelerometers in IY.

1. Opened the top chamber. Fixed the IP and the F0.
2. Removed the ballast weights from the proof masses.
3. Put the accelerometers on the IP.
4. Attached the ballast weights again and implemented the protection covers.
5. Set each accelerometer (ACC #4, #5, #6) by each geophone at the position of H1, H2, and H3.
6. Adjusted the tilt of accelerometers and fixed the ACC base.
7. Connected the cables for the accelerometers.
8. Checked the signals from/to LVDT-ATC in the accelerometers. The in-vacuum cable for ACC #2 was repaired.
9. Weights of three accelerometers were 3.5kg x3 = 10.5kg.
10. Replaced the ballast ring of 17.2kg (t=38mm) to 4.5kg (t=10mm) + 2.7kg (t=6mm). Total 10.0kg was removed.
11. Released the IP and the F0. Adjusted the tilt of IP with the jacks.
12. Since the resonant frequency of IP was too low, the ballast ring of 2.7kg was removed.
13. Closed the top chamber.
14. Measured the spectrum of the accelerometers. The resonant frequency of IP was 60-70mHz.

I implemented the controls for the Tower and made Guardian work for coarse alignment (reproducing the green alignment in 19439 and 19465 but with better controls)

The design of the filter follows a similar workflow in 19429, 19144, and 17957.
OLTFs and the filters are shown in figures.
For IP, GAS and BF yaw control, they are DC controls and for BF L and T they are just damping,
They all come with 2nd-order (IP) and 4th-order (others) low-pass optimized for 45 degrees phase margins.
The filters are implemented as AC (FM2), damp (FM3), DC (FM4), and notch (FM5) following the IX/EX convention.
AC+DAMP turns the overall filter into a simple damping filter, while damp+DC turns it into a DC control loop (PID basically).
In addition, null filters with gain(0) and 5-seconds ramp time are put in FM1 to enable the FLOAT states in Guardian.

GAS filters LVDTs have very high coil2LVDT coupling at higher frequencies.
For most of the loops, I had to manually reduce the gains (by 2 or 4 times) so the OLTF doesn't peak out of the unity gain which can cause instability.
In the case of F1, I had to set the low-pass filter manually and sacrificed some phase margin.

For now GAS filter setpoints are arbitrarily set to 2680, -1300, 650, -300, and -1700 for F0, F1, F2, F3, BF respectively.
These are just values close to where they are sitting without DC control.

BF L and T fluctuations are close to unobservable when IP L and T loops are engaged. Or rather, the noise floor of the BF LVDTs is really high.
BF L and T loop gains are reduced 4 times manually due to high sensor noise at higher frequencies. They might saturate the coils at higher frequencies if the gain is not reduced.
I don't plan to engage them for alignment because they practically don't do anything besides injecting noise and saturating the actuation.
They are there just in case we need them.

All filters are generated from /kagra/Dropbox/Subsystems/VIS/vis_commissioning/etmy/notebook/control/

As for the Guardian, it's configured to engage all control loops described above except BF L and T. So, MN loops are not engaged for now. The signals aren't fluctuating a lot anyway.
When going to ALIGNED state, it engages MN_OPTICALIGN_P_OFFSET with 1625 counts, which is the value we found in 19439 for good pitch alignment.
For yaw, it's assumed to be aligned with 0 set point at the BF.
In the ALIGNED state, the OPLEV reads (30-seconds average)
K1:VIS-ETMY_TM_OPLEV_TILT_VER_INMON: 0.087206  (Good value 0.006388)
K1:VIS-ETMY_TM_OPLEV_TILT_HOR_INMON: -0.339861  (Good value -0.299486)
which are not too far away from the good values.

Then Guardian enters the ISOLATING state, it waits until the RMS (60s time constant, hardcoded in the real-time model) of the error signals to become lower than 5.
The BF yaw control has an integration time constant in the order of 1e-2, so it takes quite a while (say a few minutes) for Guardian to exit the ISOLATING state.
We can put an offset in OPTICALIGN block to sort of speed up the process but I am a bit worried about this approach:
The static position of the suspension can change and could require a different offset in the future.
So if we hardcode an OFFSET now it might work not but not certainly in the future.
For now, I settle at putting a -330 offset (determined from the DC control signal) in the BF_OPTICALIGN_Y_OFFSET, which is engaged by the guardian at some point.

See figures for the fitted transfer functions for the IP, GAS filters, and the BF.
For the BF, I only modeled L, T, and Y DoFs, which are the DoFs that we might want to control.
For IP L and T, they are somehow coupled to the yaw chain mode at 0.023 Hz. I ignored them when modeling the transfer functions.
As usual, GAS filters and BF LVDTs have huge coil2LVDT coupling at higher frequencies (typically above 10 Hz).
These are not modeled. The peaks at higher frequencies, which are physical resonances, are not modeled as well.

The measurements can be found in /kagra/Dropbox/Subsystems/VIS/vis_commissioning/etmy/transfer_function/
The fit can be reproduced at /kagra/Dropbox/Subsystems/VIS/vis_commissioning/etmy/notebook/transfer_function_fit/

The modeled transfer functions are put into FM10 of the control filters. (DCCTRL for the IP, DAMP for others).

I modified the BF_SENSALIGN matrix to reduce yaw to longitudinal and transverse coupling.
Fig. 1 shows the transfer function and cross transfer functions with yaw actuation. References are measured before changing the matrix.
As can be seen, there were huge Y2L and Y2T coupling (especially at ~0.023 Hz) which was preventing me to measure good transfer functions.
The yaw coupling has been basically eliminated as shown in the figure.
Fig. 2 shows the updated SENSALIGN matrix.

It's worth noting that for some reason, the L and T diagonal elements are not 1 from the beginning. I didn't change them.

The matrix can be generated from /kagra/Dropbox/Subsystems/VIS/vis_commissioning/etmy/notebook/diagonalization/bf_lvdt.ipynb