Pastavi came back online on Monday. (I forgot to report)
Totally, I investigated three locked loss during last night.
1. 1443273610 13:19:52 UTC
Fig.1. 7.5 Hz oscillation detected in ETMY TM oplev, but no change in MN TILT SUM, so not the usual oplev glitch, but actually sudden excitation of the 7.5 Hz.
Between 1 - 2, large earthquake detected and we need longer time for IFO recovery
2. 1443301314 21:01:36 UTC
No critical hint can be detected, no significant ground motion etc
3. 1443303898 21:44:40 UTC
Fig.2. I am not so sure the PMC related channel, something happened in PMC. We need to ask Yuzurihara-san or expert of the PMC to check the PMC related signal more.
[miyoki, ushiba]
Before installing SR560+Notch+ 50ohm teminator, we checked the pahse dely of the SR560 for 10kHz and 100kHz with legacy methods (photo1,2).
Assuming Gain=2, DC coupling, no filters, 50 ohm output and 50 ohm terminator, roughly, 3.6 degrees delay for 10kHz and 36 degrees delay for 100kHz. (In other words, 1 useconds delay for 100 usec(10kHz) and 10usec(100kHz)). So we judged the pahse delay effect can be accepted.
By the way, SR560 now costs over 5000 USD! we should prepare more cost effective solution :(
Abstract:
I modified the ASC filters to reduce the residual RMS of effor signals for ARM ASC.
It seems to work well, so I would like to keep IFO in PRFPMI_RF_LOCKED state to check the stability of IFO.
Detail:
As reported in klog35093, ASC loop gain seems to decrease from the nominal value, so I changed the overall gain of ASC by checking WFS error signals.
Based on the measurement in klog35093, I increased the gain of DHARD_P, DHARD_Y, and CHARD_P by a factor of 2 and CHARD_Y by a factor of 1.5.
Gains are set at FM7(6dB), FM3(6dB), and FM7(6dB) for DHARD_P, DHARD_Y, and CHARD_P, respectively.
For CHARD_Y, I modified the FM8 from 3dB gain to 6.5dB gain to increase the CHARD_Y loop gain by a factor of 1.5.
In addition to this, WFS error signals had a large peak at 0.08Hz, so I implemented resG at 0.08Hz with Q=3 and depth =20dB.
These filters are implemented at FM1, FM9, FM1, FM6 of DHARD_P, DHARD_Y, CHARD_P, and CHARD_Y, respectively.
I also tested resG for CSOFT_Y and DSOFT_Y but it doesn't seem to make significant changes, so I turned them off today.
Figure 1 shows the error signals of HARD loops before (blue) and after (red) filter modification.
As you can see, RMS of all error signals are significantly reduced.
I checked if the IFO relock canbe done automaically after my changes and confirmed that it can reach PRFPMI_RF_LOCKED state.
So, I keep this condition tonight and see the long-term stabiity.
It takes a long time to damp 1.69Hz resonance in L DoF, I modified the NBDAMP filter of ETMX L5.
Origina filter was moved to FM2 (BP1.69_old) as shown in fig1.
New filter was implemented at FM1 (BP1.69) as shown in fig2, which BP frequency was shifted 0.04Hz.
After that, OLTF (fig3) seems to be healthy and damping performance was improved.
I modified the IMC guardian so that the IMC UGF becomes 130kHz by increasing the IMC CMS IN1 gain by 3dB (fig1).
In addition, I also increased IN2 gain by 3dB (total gain is +3dB as shown in fig2) to keep the relative gain between IMC loop and CARM loop.
Figure 3 shows the OLTF of ALS_CARM with the above condition.
Thin and thick line shows original gain and increased gain for both IMC CMS IN1/2, respectively.
So, the ALS_CARM phase is slightly improved due to the increase of IMC UGF.
Figure shows the PLTF of IMC loop with SR560 + 150k notch + 50ohm resistance.
Thin line shows the OLTF with the nominal IMC gain determined by the IMC guardian.
Thick line shows the OLTF with increasing gain by 3dB from thin line.
A peak at 150kHz was disappeared thanks to the notch filter and we can increase the UGF at 130kHz without any problems.
Abstract
We successfully installed the 150kHz notch filter in the path to the PZT in the IMC control loop, and also recovered the UGF of 120kHz for the IMC control loop without resonance at 150kHz.
Background
After replacing the neoLase laser, we could not increase the UGF for the IMC loop as FBL case upto 120kHz because of 150kHz resonance appeared. Ushiba-kun confirmed that the 150kHz resonance existed in the PZT path in the contol loop. A passive 150kHz notch filter was prepared, assuming 50 ohm terminator resistance at the PZT input. However, the present output circuit IC(AD829) for the IMC servo to the PZT can drive only 6mA current, then 50 ohm resistance cannot be directly connected with this notch filter. On the other hand, the actual voltage amplitide to the PZT seemed to be limited within +/- 1.5V because 1/10 attenuator was set at the output in the IMC control servo circuit to the PZT path. Consequently, a current driver amplifier with roughly ~ 30mA driving ability is enough to solve this situation.
Countermeasure
We decided to use SR560 to drive 50 ohm resistance with 150kHz notch because it has a specification to drive 50 ohm resistance with 5V output voltage using its 50 ohm output. According to Shimode-san, OP37 + LM6321 current buffer amp are used in the SR560 for the output circuit. It can drive 300mA according to its specification sheet.
Reforming the PZT path and IMC lock recovery
- confirmed that the notch filter was connected at the output of the IMC servo SLOW out (PZT) as photo.1,
- set IMC down (PMC was set resonance),
- disconnnected the BNC cable at the PZT input of the neoLase laser's master laser, then I connected only 2W type 50 ohm terminator,
- removed 150kHz notch filter from the IMC srvo slow output (photo.2), and connect it at the 50 ohm terminator input at the master laser side. (photo.3)
- set SR560 on the 19 inch rack cage that houses neoLase controller and DC voltages (photo.4). The power line of the SR560 was taken from the uninterruptible power supply that was also used for the neoLase amp, etc. (photo.7) set the input status at GND. connect the BNC cable from the IMC control servo at the "B" input in the SR560. The polarity was "non INV". set the gain of 2 to compensate the 1/2 voltage reduction due to 50 ohm output and 50 ohm load condition with the notch filter. set no filter in the SR560.
- check the output voltage from the 50 ohm output using oscilloscope. it showed very slight offset at the level of mV. I adjusted it around zero by rotating the offset adjustment resistnace? in the SR560. (photo.5, 6)
- connect a BNC cable between the SR560 50 ohm output and the notch filter. also connecte a BNC cable between 600 ohm output of the SR560 and the oscilloscope to monitor the feedback voltage to the PZT.
- change the input statsus from GND to DC coupling.
- I set "Providing Stable Light" in the IOO guardian. IMC could be locked.
Loop gain, etc, checks
- I checked the RMS amplitude to the PZT. Just after the IMC lock, RMS was ~ 1.5Vp-p. After ~ 20 minutes or so, it became smaller to the level of 0.5Vp-p or so, maybe because I exited from the PSL room and no air turbulace was generated.
- To measure the transfer function of the IMC control loop, I rearranged cables to measure it by using Moku. Ushiba-kun confiemed that the 150kHz bump disappered in the TF.
- We tried to increase the UGF for the IMC control. Although 140kHz was realized, the phase margin became ~ 27 degrees or so. We decide to keep the same UGF of 120kHz as before with the phase margin of > ~ 40 degrees. (Becasue of troubles, TF data was lost. We will measure later again.)
Additional Note
I change the target temp of the precision cooler for the PSL from 21.1C to 19.1C only during activities in the PSL room to mitigate the temp changes inside the PSL room.
I noticed that ETMX V1 actuators had been saturated since September 18.
So, I offloaded ETMX pitch signals by moving the CRY stepper motor.
Kamioka Mining Company
The trap for bears was moved between KAGRA and the Atotsu mine entrance.
A graph was created to compare the original GSK responsivity [V/W] with the GSK responsivity calibrated using TSK.
The difference between the two main values is about 0.2%, as reported previously, which is within the uncertainty and therefore consistent.
Original GSK responsivity [VpW]: [-4.700954, 0.00476]
Updated GSK responsivity [VpW]: [-4.69120523 0.00720338]
Tanaka-kun's report on April 30 2024 and May 1st was very similar to the present situation. Do we forget something to do about interlock???
Accroding to a mannual that was delivered from Mio-sensei today (this is the first time for me to see the mannual for FBL), FBL amplifier has an intelock connector that has 4 pins. 1-2 and 3-4 should be connected to operate FBL.
-----
Another minor issues.
- The slackpost function stoped some guardian nodes due to the HTTPS connection error during the network down.
It was recoverd by STOP/EXEC cycles. But this kind of error should be caught and properly processed without stopping guardian node itself from the viewpoint of operations of observing runs. I prepared modified version of the slackpost function in order to avoid fixing each guardian node in which the slackpost function was used. It will be replaced someday.
- Connection to Mozumi entrance workstation was lost due to a change in IP addr.
Because network connection on the mozumi entrance workstation is established by DHCP, IP addr was changed before and after the planned power outage. So all connections from DGS network were lost. As the result, auto-recovery, screenshot on the Web and some other services were stopped. I updated the connection settings on the script side and then all services came back online. Because using DHCP makes hard to keep these services, a static IP should be assigned for Mozumi entrance workstation if we want to keep these services.
Date: 2025/09/29
Member: Misato Onishi, Keisuke Sakanoue
We performed our usual WSK calibration at UToyama.
The results look no problem.
Results
| Case | Alpha (Main Value) | Alpha (Uncertainty) |
| Front WSK, Back GSK | -0.908612 | 0.000216 |
| Front GSK, Back WSK | -0.907487 | 0.000079 |
Comparison with Previous Results
Comparing with previous results, no significant issues were found.
Attached graph is the result summary including the latest measured data.
According to Mio-sensei, the most fragile part is the ADC/DAC board in the Amplifier. This might explain the strange data of temperature and so on.
We have no way to repair ADC/DAC at present.
[miyakawa, uchiyama, miyoki(remote)]
[Abstract]
The broken? FBL was replaced with a spare one. However, we could not operate it. At present, we suspected that the control software had bugs or that the spare FBL was also broken.
[What we did]
- The old FBL was removed from the 19 inch rack, and set the spare FBL in the same 19 inch rack.
- Hoses to the chilleres, and optical fibers to the master laser optics and output optics on the table were connected to the spare FBL. Very simple connections. The chiller was on.
- Proceeded the laser operation according to Nakano-kun's manual. However, we encountered the following troubles,
- When turning on the master laser and applying the current as in mannual, we got 13mW values as the input laser power for the amplifier in the software GUI. This was the same to the broken ? FBL case. In addition, actually, ~ 1W laser power can be detected at the out put of the FBL amplifier. This was also the same stituation to the broke FBL case.
- Before enabling the amplifier, 3C !! temperature was displayed in the NuAMP_RACK FBL controller software GUI. This was also the same situation to the broken FBL. This value should be 23C or so, according to the Nakano-kun's mannual.
- After enabling the amplifier, 13mW became 26mW and temperature became ove 50C!! However, no warning was displayed in the GUI. This was also the same situation to the broken FBL.
- We increased current gradually. However, no changes in the FBL output power. This was also the same situation to the broken FBL.
- We suspected that the mechanical interlock in the amplifier. So we mechanically shorted the interlock terminals to kill the interlock. However, the situation was the same.
- We suspected USB outputs on the PC side. So we changed the USB port. However, no FBL output.
- We suspected that windows PC bugs. So we rebooted the windos PC (Windows10) and tried. However, no FBL out put.
- We suspected that some initial data files that were used by the software. However, we could not find such data, except for the calibration data of the FBL power.
- Although we wanted to reinstall the control software, we could not find such installation devices near the windows PC.
- We checked the directry where NuAMP_RACK FBL controller software was installed. We found another software for the FBL control. We used this aftware. We found that this software apperacne was alsmost same to the previous one, except for warning messages as follows,
- Seed signal Too Low, master Fault, even 13mW was obtained.
- Temperature too high, even 3C was displayed in GUI.
- Back reflection too high. We could not understand this message.
We are now asking Mio-sensei any ideas about these situations and also asked him and Nakano-kun whetehr we can obtain the software installer.
Anyway, the displayed temperature of over 50C dgrees was strange and it could enable the software interlock to the amplifier. Or, both FBL were broken.
According to Mio-sensei, the most fragile part is the ADC/DAC board in the Amplifier. This might explain the strange data of temperature and so on.
We have no way to repair ADC/DAC at present.
Tanaka-kun's report on April 30 2024 and May 1st was very similar to the present situation. Do we forget something to do about interlock???
Accroding to a mannual that was delivered from Mio-sensei today (this is the first time for me to see the mannual for FBL), FBL amplifier has an intelock connector that has 4 pins. 1-2 and 3-4 should be connected to operate FBL.
I checked the past data as Fig.1, especially around the July 30th when Kamchatka earthquakes happened because Type-As were tripped and forcely misaligned for one day, and because IFO was set down on purpose for 1 week to avoid the after shocks. Fig.1 shows,
- date of zero is the earthquake event.
- During one day after the earthquakes, ALS-X/Y_PNC_SERVO_IN1_DQ should show the power from the PSL room only because the ITMXY were tripped and misaligned. PNC-Y side clearly showed power fluctuation! While, no power fluctuation in PNC-X side.
- After the ITMXY recovery to the aligne mode, both showed interference fringes. PNC-X showed constant fringe width, while PNC-Y showed fringe width fluctuation as reported Ushiba-kun. The same kind of fringe fluctuation can be seen before the earthquakes during IFO locking.
Consequently, the reported fringe width fluctuation did exist around July 30th, 2025.
As the next investigation, I traced back when this PNC-Y fringe fluctuation started. Fig. 2 shows the data around August 30, 2024, that was just after the start of the IFO recovery. The PNC-Y fringe fluctuation seemed to be much smaller than now. However, PNC-Y fringe fluctuations can be recognized around September 30, 2024! as Fig.3.
So, the PNC-Y fringe fluctuation seemed to exist for 1 year. We just found such phenomena, that were triggered by the recent many trouble shootings.
The larger fluctuation of ALS-Y(and X?)_LASER_TEMP_SERVO_OUT16 seemed to continue as Fig.1 and 2. Actually, PSL-PMC_ZPT_HV_OUT_DQ also seems to be a little bit noisier than the stable state as Fig.4. I checked the same data in July and June. The past data was comparable with the noisy level which is reported in this post. So, the fluctuation itself seems not to be problem
IMC trans is getting worse as in Fig.3. Fig.4 also shows the relation between the slight PMC trans reduction around -2 days and noisy PSL-PMC_ZPT_HV_OUT_DQ.
IMC and PMC were unlocked 6 times for 40 minutes as Fig.1 and 2. One or two of them seemed to be related with earthquakes.
Fig3 and 4 show the close up one of lock loss timings. PMC was unlocked 4 times, while IMC was unlocked 11 times or so. Especially, Fig.4 shows the glowing up noise in PSL-PMC_MIXER_MON_OUT_DQ.
Fig5 an 6 show the close up with seismic noise enhancement. Fig.6 shows the glowing "down" noise in PSL-PMC_MIXER_MON_OUT_DQ.
Fig.7 and 8 show the noisey timing including Fig3,4,5,6. Fig.8 shows glowing "up and down" noise in PSL-PMC_MIXER_MON_OUT_DQ.
On the other hand, IMC trans decreased after the big shock due to earthquake about 1 day ago. I don't know weather the IMC alignment changes were related with the IMC successive unlocks or not (Fig.9) even though the IMC-SERVO_OUT2_OUTPUT values (offset in other words) were almost constatnt.
Fig10 and 11 show PLLX/Y. In the recent half day, the ALS-Y_LASER_TEMP_SERVO_OUT16 seemed to fluctuate more compared with the past, while ALS-X seemed to be same woth the past. So, is it better to adjust PLLY?
Takahashi-san reported that the temperature at IYV decreased too much, then F0 GAS position became too low. I increased the heater power again in IYV by 600W. I will reduce it to 300W tomorrow.
Also, the temperature at IXV was decreasing too much. However, I could not increase the heater power because of network discinnection. I will check tomorrow.
More two lock losses around -7 hours. They had no correlation to the seismic noise. However, They are quickly recovered and no continuous lock losses happened. In addition, the IMC trans seemed to decrease and IMC trans power increased at the latest lock loss. So these lock losses were caused not by known crazy noise in the laser but other noises, perhaps.
Operators: Kokeyama, Nakagaki
Shift time: 09-17 JST
Check Items:
VAC: No issues were found.
CRY cooler: No issues were found.
Compressor: No issues were found.
Temp check: (at 10:00) FIELD_IYA changed by +0.5°C and COOLER_BS changed by +0.4°C. Reported to Miyoki-san and Hayakawa-san.
(at 16:00) No issues were found.
VIS GAS filter output check: (at 10:00 and 16:00) ITMY F0 over -25,000 counts. Reported to R.Takahashi-san.
IFO state (JST):
09:00 Start of shift: STANDBY
17:00 End of shift: STANDBY
Operators: Kokeyama, Nakagaki
Shift time: 09-17 JST
Power Outage at Kashiwa.
Check Items:
VAC: No issues were found.
CRY cooler: No issues were found.
Compressor: Not checked due to the power outage at Kashiwa and not accessible to the image.
Temp check: The temperature in IYA changed by -0.40°C at 10:00. Reported to Miyoki-san and Hayakawa-san.
VIS GAS filter output check: Not checked due to the power outage at Kashiwa and not accessible to the image.
IFO state (JST):
09:00 Start of shift: STANDBY
17:00 End of shift: STANDBY
Only one lock loss because of an earthquake near the site for the last 1 day and 16 hours as in Fig.1 and 2. Aftre relock, the IMC trans decreased a little bit.
You can see slight enhancement arout -21 hours in PSL-PMC_PZT_HV_MON_OUT_DQ. Fig.3 shows the close up around the enhancement in PSL-PMC_PZT_HV_MON_OUT_DQ. This enhancement is because of the very long oscillation (~ 5 minutes in this case). It had relation to PSL-PMC_HEATER_OUT16. So, some instabilities in the temperature control time scale could happen.
I decreased the current as Fig.1.
