I offloaded the IM V OSEMs with the picomotors for pitch. It was performed in ALIGNED mode.

[Miyakawa, YokozaWashimi]

We installed 4 ahakers installation in the PSL room.

1. Near the final periscope
2. Near the ACC3 and the Laser source
3. & 4. Near the suspect mirrors

This update proposal was applied in klog#36305.

Spectrograms for the  large shaker tests.

Spectrograms for the  small shaker tests.

[Tanaka, Dan, Ushiba, Komori]

Abstract:

We attempted to implement a new filter to achieve high-bandwidth control of DHARD yaw, but have not succeeded yet.
As the next step, we will pursue a more moderate-bandwidth control scheme.

Details:

We began implementing a new filter to control DHARD yaw with a higher unity gain frequency (UGF).
This approach is expected to significantly reduce fluctuations in the arm transmission, which currently seem to be limited by TM yaw motion.

Based on the measured transfer function from DHARD yaw to the TM oplev signal (klog:36289), we designed a new filter consisting of ten zero–pole phase compensation stages.
The figure shows the estimated open-loop transfer function, with the UGF set to 4 Hz.

When we enabled DHARD yaw control using this new filter, an oscillation around 20 Hz occurred immediately, leading to lock loss.
To suppress this oscillation, we had to reduce the gain by a factor of approximately 3e3.
However, with such a low gain, the control was insufficient to drive the error signal to zero, even for a long period.
In addition, another oscillation appeared at approximately 80 mHz at this reduced gain, which also resulted in lock loss.

One possible cause of the high-frequency oscillation is coupling from other degrees of freedom.
The ten zero–pole phase compensation stages significantly amplify signals at high frequencies, which likely enhances such couplings.
Therefore, our next plan is to implement a moderate-bandwidth control scheme.
The target UGF will be between the second and third yaw resonances, approximately 2.5 Hz.

I restarted the automeasurement scripts on the remote desktop session of k1ctr5.
I hope the script will finish by tomorrow morning but ​​​​​if someone would like to stop the script, please send the command Ctrl+C on the terminal opened on the remote session of k1ctr5.

[Kimura and Yasui]
 On February 5, the helium gas inside the cold heads of the duct shield cryo-cooler unit were replaced.
The cold heads whose helium gas was replaced were and Yfa (arm side of the IYC) and Xfa (arm side of the IXC). 
Yfa worked from 10:30 AM to 11:40 AM, and Xfa worked from 1:30 PM to 3:30 PM.
The pressure of the helium gas in the cold heads is 15 bar. 
The grade of helium gas used was G1.

[Tanaka, Komori, Dan]

1. CHARD_P

At first, We worked with only the CHARD_P loop mainly engaged, while the other P DoFs were kept weakly locked.

• The CHARD_P filter was created by copying from the DHARD_P filter and modified a little.
• The OLTF was measured and compared with previous results.

OLTF file: /users/Commissioning/data/ASC/2026/0205/OLTF_CHARD_P_20260205.xml

• Brown: updated (measured in the morning)
• Blue: yesterday’s result
• Green: before notch tuning
• Purple: remeasured CHARD_P OLTF

The purple measurement was performed in the RF_LOCKED state, where all other DoF loops were also closed. A large response was observed in DHARD_P signals, suggesting that residual coupling remains from CHARD_P to DHARD_P.

Result: See OLTF_CHARD_P.png (fig001).

2. DHARD_P

Previously, the DHARD_P gain was set to −3 externally. In this work, the gain was incorporated into the filter bank FM6 = HBtest.
After this change, an oscillation appeared around 2 Hz. Therefore, the gain was temporarily reduced from gain(3) to gain(2).

Compared with yesterday’s results, a larger bump around 10 Hz was observed. After repeating the measurements several times, this bump seemed to occur when the alignment was poor. In particular, the bump around 10 Hz was prominent in measurements taken when the transmitted power was low. 

OLTF measurements were performed with and without the DHARD_P FM2 (notch filter). After that, the IFO was brought to the RF_LOCKED state, and the same measurements were repeated.

Measurement file: OLTF_DHARD_P_20260205.xml

Result: See OLTF_DHARD_P.png(fig002).

3. ASC GRD update

Based on the above results, the ASC GRD was updated.

• The appropriate filters for DHARD_P and CHARD_P are now automatically engaged.

Filters to be engaged for DHARD_P and CHARD_P:

• ASC_filter_after_GRD_change.png (fig003)
• DC_HARD_P_FB.png (fig004)

VIS MN LOCK filter configuration:

The FM4 = HBtest filter (marked by blue circles in the figure) is now turned ON via GRD instead of the conventional FM9 and FM10.

See VIS_MN_LOCK_FB.png (fig005).

To derive power recycling cavity length in klog #36249, I used the finesse of PRX/PRY to be 2.00(3), which is a theoretical value with measured BS T and R, without losses.
In reality, from carrier PRG measurements PRX had ~10% loss and PRX had ~15% loss.
Also, sideband PRG for PRFPMI sugested ~13% loss in PRC.
If we add optical loss of 13(5)% loss, PRX/PRY finesse will be 1.89(4).
This slightly changes the linewidth I used to calibrate the PRC length (the effect is negligible compared with other uncertainties).
Updated values will be

Estimated f2 detuning of PRC    0.071+/-0.004 MHz
Sideband PRG for PRX    0.0797+/-0.0031
Sideband PRG for PRY    0.0654+/-0.0026
Average PRC length      66.523+/-0.006 m

So the measured PRCL is shorter from the design (66.591 m) by 6.8(6) cm.
If we are to keep the current f2 modulation frequency (44.9946924104 MHz), PRCL needs to be shortened by 10.5(6) cm.

Using these values, the calibration factor for POP90 sqrt(I^2+Q^2)/Pin will be 6.16(18).
This gives the following (see attached).

Sideband PRG for PRFPMI 7.97+/-0.23
Sideband losses in PRC for PRFPMI 12.43+/-0.33 %

Note that this loss includes the effect from f2 detuning.
Carrier losses in PRC for PRFPMI should be estimated independently from carrier PRG (to be done next).

I checked the measured TF of BS for the noise budget.
Following local controls have a large coherence to DARM, so precise projection should be performed.
The other TF have low coherence, so the noise contribution to the current DARM should be small.

TM_OLDAMP_P: fig1
TM_OLDAMP_Y: fig2

[Tanaka, Komori]

We measured the transfer function from the MN to the TM in the yaw degree of freedom as the first step toward achieving higher-UGF yaw control.

We paid careful attention to reproducing conditions as close as possible to the interferometer locked state; for example, the OLDAMP yaw control on the MNs was turned off.
In addition, we turned off the TM lock filters, since our initial approach is to control yaw using feedback applied only to the MNs.

The measured transfer functions for each suspension are shown in the figure.
The results are consistent with expectations for a triple-pendulum system.

[Tanaka, Komori]

Abstract:

We successfully decoupled the pitch {D, C}SOFT modes from the HARD mode by adjusting the sensing matrix of the arm transmission QPDs.

Details:

After tuning the output matrix of the arm WFS signals based on the calibrated TM oplev signals, the next step was to decouple the {D, C}SOFT and CHARD modes from DHARD.
To measure the couplings, we injected a 50-mHz sine wave into the DHARD output while keeping the arm WFS control engaged with the conventional filter.
The excitation frequency was chosen to be 50 mHz in order to avoid feedback to the TMs, which is enabled in the conventional filter, and to isolate the response of the MNs.

The initial result is shown in Fig. 1.
From the beginning, we did not observe a clear 50-mHz peak in the CHARD signal.
This suggests that the TM oplev calibration is still accurate and that DHARD and CHARD are already decoupled at the level of an order of magnitude.
In contrast, clear peaks were observed in the DSOFT and CSOFT signals.
We therefore decoupled these modes from the DHARD excitation by tuning the sensing matrix of the arm transmission QPDs, which serve as the sensors for the SOFT modes.

Figures 2-4 show a comparison of the four arm signals during excitations of DHARD, CSOFT, and DSOFT, using the previous (labeled as “REF” in the figures) and the updated sensing matrix.
After the adjustment, the peak amplitudes in the DSOFT and CSOFT signals during DHARD excitation were significantly reduced, while sufficient responses remained in each degree of freedom when the corresponding excitation was added.
The updated sensing matrix is shown in Fig. 5.

Komori, Tanaka (remote)

As reported by Komori-san in klog, we decoupled HARD/SOFT modes in terms of SOFT sensors. After decoupling, we measured OLTFs of {D, C}SOFT modes. This time, we set the {D,C}SOFT_P_GAIN to -3. And, BPCs were engaged to controll their modes in their DC regions. Also, other ASCs were used not new filters but previous filters.

Fig. 1 and Fig. 2 show the results. The UGF of the CSOFT_P control seems to be 30 mHz and the gain shape becomes flat below 30 mHz. This time, the UGF of the DSOFT_P seems to be not measured with enough coherence but it seems to be similar with the one of CSOFT mode.

After then, we engaged DHARD_P controls with new filters. Then we measured the OLTF of the DHARD_P control. fig. 3 shows the results. The green line shows the OLTF before decoupling and the brown lines is after the decoupling. As you can see, the peak height around 10 Hz become smaller.  

However, we reminded that CHARD_P control with new filter was engaged when the green lines was measured. On the other hands, in this time, CHARD_P control was engaged with previous filters. So we suspected that the peak around 10 Hz came from the new CHARD_P control. In order to confirm this, we engaged the new CHARD_P controls and remeasured the OLTF of  DHARD_P. Red lines in fig. 3 shows the results. the peak height around 10 Hz is still smaller even though the new CHARD_P control was engaged. That is, the decoupling b.t.w SOFT and HARD makes the peak height become smaller.

Then, we measured the OLTF of CHARD_P. The UGF seems to be around 2 Hz but the phase margin at UGF seems to lower than 20 degrees! So it is necessary to improve the margin by modifying the filter design.

I tuned the template files and ran the automeasurement script on the remote desktop session of k1ctr5.
If someone would like to stop the script, please send the command Ctrl+C on the terminal opened on the remote session of k1ctr5.