Workers: Kohei Mitsuhashi, Dan Chen

We performed monthly Pcal-Y calibration on 2026/03/17.

After the calibration, we updated EPICS parameters related to the Pcal-Y system. No issues were found.

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

Operator: Kohei Mitsuhashi, Dan Chen

Update Time: 2026/03/17 15:31:24

I offloaded the F2 and BF GAS filters with the FRs.

[Washimi, Takahashi]

We replaced the broken signal generator (SG) NF1956 with the embedded SG card for the LVDT driver (S1807787) in TMSX. We opened the driver chassis and inserted the SG card. The output of the SG card was connected to "REF IN" port on the front panel. The signal was set to 10kHz in frequency and 7Vp-p in amplitude.

The new Pastavi server was purchased with the SEO budget and delivered to the KAGRA site. I performed the initial setup on the server and copied the data viewer system. After completing the preparation at the KAGRA site, I sent it to Kashiwa last Friday. I will replace the server in Kamioka this Thursday. 
Please note that it may take some time to recover the noise budget mode. I expect it to be restored within a few days.

Note

• Because the OS version jumped significantly from 9 to 12, many modifications to the settings were necessary.
• The cgi package has been deprecated from Python 3.13. Although it is possible to use legacy-cgi, its behavior was slightly different from what I expected. Therefore, I decided to use Python 3.12. In the future, it may be necessary to redevelop Pastavi using another framework or tool.
• Installation of the frameL was not as straightforward as in the previous environment  (actually I just wanted to use the FrChannels command). The procedure for installing lal (or framel) using apt is described on the following pages: [wiki] [IGWN debian repo]

ITMX has successfully reached the LOCK_ACQUISITION state.
We will leave it in this state for a while to verify that no issues arise with the new V2 I/O chassis.

I offloaded the IP H2 with the FR for recovery from the saturation. The guardian state could be switched from READY to ISOLATED.

[Takahashi, Washimi, Hirata]

We replaced the window mirror (SRM-W) with the 70% mirror (SRM-M). This is a demonstration for training in replacement. At the beginning, we confirmed the mirror's wedge direction: the X- side (Picture 1). We removed the black cylinder from the rear side and removed the mirror holder (Picture 2) by pushing with a Teflon bar from the rear side (Picture 3). We attached the assembled 70% mirror to the test mass (Picture 4), confirming the wedge direction (Picture 5).

We checked the extracted window mirror. The mirror thickness was 9.6mm at the wedge marker. Two 0.5mm shims were used as the spacer for the flange.

## Abstract

In order to unifiy the design philosophy of Type B local controll filter, I began to modify the control filter in the SR2 tower part (IP_IDAMP_Y, {F0,F1,BF}_DAMP), especially their roll off filters so that the same filters as SR3 are used in SR2.

Feedback spectra above 10 Hz seem to be reduced thanks to this modification. However, there seems to be gain peaking at 3-4 Hz in GAS signals. so we need to adjust them.

## What we did

In order to unifiy the design philosophy of Type B local controll filter, I began to modify the control filter in the SR2 tower part (IP_IDAMP_Y, {F0,F1,BF}_DAMP) so that the same filters as SR3 are used in SR2.

Basically,  the difference between SR2 and SR3 controll fitlers was the roll-off filters. SR3 usually used the 4th order elliptical lowpass filter. On the other hand, SR2 used 2nd or 3rd order pole. Therefore, the gain of the SR2 filter above 100 Hz is typically more than 20dB larger than the one of SR3. This indicates the SR2 filter transmits the 20dB larger sensor noise above 100 Hz to the suspension. So, I tried to impliment the same roll off fiter to SR2 control.

Fig. 1 shows the example of the modification for SR2_IP_IDAMP_Y. Fig.2 shows the medm screen of SR2_IP_IDAMP at that moment. I modified the roll off filter for GASs in the same manner.

After the implementation, I confirmed that SR2 could be reached LOCK_ACQUISITION state. Then, I measured the spectra of error/feedback signals of their controls in the LOCK_ACQUISITION state. Fig.3 and 4 shows the results of IP and GAS, respectively. Red curves in figures are current SR2 error/feedback signal spectra and magenta curves are the spectra before the modification. Blues(and Cyans) are the spectra of SR3 as references. Above 10 Hz, the feedback signals' spectra were reduced successfully. On the other hand, there seems to be gain peaking at 3-4 Hz in GAS signals. So we need to fix them.

## Next

• IM_DAMP_{L,T,V,R,P,Y}

ICX6610-24P Switch>show ip
     Switch IP address: None
           Subnet mask: None
Default router address: None
   TFTP server address: None
Configuration filename: None
        Image filename: None
                IP MTU: 9216

ICX6610-24P Switch>enable
ICX6610-24P Switch#configure terminal
ICX6610-24P Switch(config)#vlan 1
ICX6610-24P Switch(config-vlan-1)#ip address  
ICX6610-24P Switch(config-vlan-1)#exit
ICX6610-24P Switch(config)#write memory
Flash Memory Write (8192 bytes per dot).
Flash to Flash Done.

telnet@ICX6610-24P Switch>show media
Port 1/3/1:  Type : 10GE LR 10km (SFP +)
telnet@ICX6610-24P Switch>show interface brief
Port    Link    State   Dupl Speed Trunk Tag Pvid Pri MAC             Name
1/3/1   Up      Forward Full 1G    None  No  1    0   cc4e.24f9.762b     

telnet@ICX6610-24P Switch>show license
License record empty

telnet@ICX6610-24P Switch>enable
telnet@ICX6610-24P Switch#configure terminal
telnet@ICX6610-24P Switch(config)#interface ethernet 1/1/24
telnet@ICX6610-24P Switch(config-if-e1000-1/1/24)# inline power

When I tried to migrate a copy process of Weather Station Web pages on k1script, I found k1script couldn't access Weather Station@Atotsu.
It was since Feb. 9th according to the last update on the Web page.
Weather Station@Atotsu cannot be launched due to CMOS error now.
A replacement of a CMOS battery (or PC itself in the worst case) seems to be required to recover it.

-----
A copy process of Weather Station Web pages was migrated from the old k1script1 to the new k1script0. This process for Weather Station at Mozumi works fine. On the other hand, it for at Atotsu couldn't access Weather Station PC. Because the old process on k1script1 was also failed, I concluded that it was a problem on the Weather Station PC not on the k1script.

Any remote connections wasn't available, so I checked a situation by plugging a display to the Weather Station PC at Atotsu and found that gathering hardware information before booting OS up failed due to the CMOS trouble as shown in Fig.1. This message is also shown when a CMOS battery (often use CR2032 on common server) becomes empty. Of course, there is some possibility about a malfunction of a motherboard. Anyway what we can try at first is opening PC chassis, and then replacing the CMOS battery if there is no obvious malfunction on the mother board.

[Nakagaki-san, YamaT-san(Support), Ikeda]

This task is related to K-Log#36129.
We measured the noise levels of the ADC and DAC using the new V2 IO chassis.
(We used different ADC and DAC cards than those used with the old IO chassis.)
For the DAC measurement, We used an empty port (on the Board 2 side) of the Whitening Chassis (S1909739).

During DAC measurement, the signal is boosted by 42 dB using whitening.

[Data]
/users/DGS/measurements/ADC/K1IX1/{ADC,DAC}/20260316_V2_IO_CHASSESS/
 

[Alex]

Investigating if the Beam Jitter Coupling has Changed

Summary: The beam jitter coupling has become worse since July 2025, especially around 350 Hz. This seems to be a real change in the coupling and not just an increase in the amount of beam jitter. We're not sure why it has increased but it may have something to do with warming up the detector or reducing the power.

Background

A few weeks ago we (Carl and Alex) made measurements of the beam jitter coupling from 250-500 Hz we noticed that the beam jitter projections were very close to darm and even over-projected at some points. We were confident that the coupling function was reasonable accurate since we had a high SNR is our noise injections in the witness sensors and DARM. This was concerning we know that DARM was lower when the detetector was operating at 10W in July but the detector noise was also lower in the regions that we were projecting beam jitter.

We manually fitted the shot noise component to DARM in order to estimate the techincal noise in this region.

This also implied that the technical noise was much higher now than in July. This left us two posibilities: the beam jitter has increased or the detetector is now more sensitive to beam jitter.

Witness Sensors

I chose to look at the IMC-REFL_QPDA1_DC since we had engaged the whitening filters on the RF channels for the time period I was looking at, as far as I am aware has have been no changes to DC sensors since O4. The noise spectra of the sensors did not appear to show any major differences between July 2025 and Februrary 2026. The spectra in february do show an increase level of noise in in the peaks around 350Hz but the increase does not correspond with the same increase in DARM.

I then looked at the coherence of these sensors with DARM which showed that the coherence had increased significantly in this region.

This implies that the coupling function has changed so that the detector has become more sensitive to beam jitter. I then decided to try and estimate the coupling function in July and Compare it to now.

Coupling Function Estimates

Ideally to measure the coupling function we would perform a noise injection, however this is not possible since we are looking at past data. Instead I used the method given in Davis et al which describes on the ways that noise subtraction was done at LIGO Hanford. This uses the cross spectra density to estimate the transfer function from multiple witness sensors to DARM.

The cross spectral density is defined as:

$c1,2(f)=Y_1(f)Y^{\ast}_2(f)$

where Y1 and Y2 are the fourier transforms of the time series. The coupling function for each witness sensor is then defined as:

$\begin{pmatrix} CF_{0,1} \\ CF_{0,2} \\ ... \\ CF_{0,N} \end{pmatrix} = \begin{pmatrix} c_{1,1} & c_{2,1} & ... & c_{N,1}　\\ c_{2,1} & c_{2,2} & ... & c_{N,2}　\\ ... & ... &...&... \\ c_{1,N} & c_{2, N} &...&c_{N,N} \end{pmatrix}^{-1} \begin{pmatrix} c_{0,1} \\ c_{0,2} \\ ... \\ c_{0,N} \end{pmatrix}$

where CF is the coupling function and 0 is the target channel for noise estimation/subtraction and indices > 0 represent different witness sensors. In the case of N=1 this reduces to:

$CF_{0,1} = \frac{c_{0,1}}{c_{1,1}} = \frac{Y_0 Y^{\ast}_1}{Y_1 Y^{\ast}_1}=\frac{Y_0}{Y_1}$

which is the conventional definition of a transfer function. In our case we use a 2 channel estimation which witness sensor 1 being pitch and witness sensor 2 being yaw of the IMC-REFL QPDA1 DC. 

I took 1000 averages from 02/07/25 01:25:00 UTC for the July measurement and 1000 averages from 24/02/26 14:53:00 UTC for the February measurement. This gave the following coupling function estimates:

and the corresponding jitter noise projections:

While there is obviously some overprojection these seem to match DARM reasonably well and show that the beam jitter noise has changed.

Causes and Solutions?

We don't know what has caused this but we know that the arm finesse has changed since the detector has warmed up so maybe the balance of the arms has also changed? As for a solution, not sure if what the solution is if we don't know the origin of the change, it's possible that going back to cryogenic temperatures and increasing the power again might reverse the change. It highlights the need for any noise subtration to be able to cope with changes in the coupling function over time.

(Work on 13th, detail report)
[Takahashi.R, Hirata]

We assembled new 2inch mirror for SRM(pic1), and we reallized the new mirror was thinner than previous one, so we couldn't secure it properly.
At the end, we re-assembled SRM-M mirror again.

0. Preparation
We set two desktop COACH filters on the working desk. 

1. First contact(FC): new SRM mirror
・I put FC on AR and HR side of new SRM mirror.
・First, FC was pour into the stainless bottle.
・5min later, pour FC to the center of mirror. I used brush to enlarge the FC area.(The tip of brush can touch FC only.)
・Few minutes later, I put tag on the surface of FC. The direction of the tag should not cover the triangle mark on the side of mirror.
・15min later, I changed surface and put FC again.

2. Disassemble SRM-M
I disassembled SRM-M mirror assy. It stored inside the decicator at the OMC area.
The shim plates for the six M3 screw were kept separate so they wouldn’t get mixed up.

3. Assemble New SRM mirror
・Align triangle mark which is on the side of mirror with scruch line of the black parts(parts#2).
・Assembled parts#7 and tightened M4 screws with 1.5Nm.
・Put O-ring on the mirror surface.
・Assembled parts#5, #9(shim plates). And realized the gap between parts#5 and parts#7 seems almost zero.
・Took some photos to chech thickness.
pic2: side view of new SRM mirror
pic3: side view of SRM-M mirror (old mirror)
pic4: new SRM mirror Assy without shim plates
pic5: SRM-M mirror Assy without shim plates (old mirror)

・Takahashi-san measured thickness of both mirrors by vernier caliper gage. It was reported on klog36579.
・Takahashi-san talked with Ushiba-san, and we decided to re-assemble SRM-M again.

4. First contact: SRM-M mirror
・I put FC to the SRM-M mirror: AR and HR side. Procedure was same as "1. FC:new SRM mirror".

5. Assemble SRM-M mirror
・Procedure is the same as "3. Assemble new SRM mirror" up to a certain stage.
・Assemble parts#5 and parts#9(Shim-plates), tighten M3 screws with 0.6 Nm.

(Work on 13th)

[Hirata, Washimi, Takahashi]

This is a rapid report. Details will be reported later. We started assembling the new mirror. We found that the new mirror (85%) is thinner than the old mirror (70%). Measured thickness with vernier calipers was 7.9mm for the new mirror and 9.6mm for the old mirror at the edge marker. The thickness of the shims (#9 in Picture) was 1.5mm. Therefore, the flange (#5 in Picture) couldn't fix the mirror even if the shims were not used.