Abstract

Details

$ lspci -nvvv | grep 10b5:9056 -A1
14:04.0 1180: 10b5:9056 (rev ff) (prog-if ff)
        !!! Unknown header type 7f
--
16:04.0 1180: 10b5:9056 (rev ff) (prog-if ff)
        !!! Unknown header type 7f
--

With Kenta Tanaka

We cleared the SFDs.

Memo
SR TM ICSINF had hodling values -10.657 for P and 29.068 for Y. We put these values at OFFSET of these filters and cleared the hold. (The offset switchs are kept at OFF.)
Later, we probably need to put these vaues to set flters.

Summary

I tried to perform the initial alignment, but I could not reach IRX_LOCKED. The initial TR_GRX was low, around 0.3. After manually adjusting the alignment, TR_GRX increased to about 0.6, and I requested IRX_LOCKED. However, the X arm still did not lock.

When the arm briefly approached the lock, the IR beam spots on ITMX and ETMX were clearly off from the center. I adjusted the ITMX and ETMX OPTICALIGN values to bring the IR beam closer to the centers, but the lock could not be established. Since the situation did not improve, I stopped the trial and restored the OPTICALIGN values of BSPR3, ITMX, and ETMX to the original values.

One possible issue is that the IR and GR axes were not well matched?

Preparation

• Opened the GR shutter.
• Moved PRM from LOCK_ACQUISITION to MISALIGNED.
• Moved ITMX from LOCK_ACQUISITION to MISALIGNED.
• Kept SRM in LOCK_ACQUISITION at first.
• Confirmed SRC flashes using K1:LSC-POP_PDA1_RF17_I_NORM_MON.
• Moved SRM from LOCK_ACQUISITION to MISALIGNED.
• Confirmed that ETMX and ETMY could reach LOCK_ACQUISITION.

X-arm alignment trial

I confirmed that FIB and PNC were locked as expected. I then requested IRX_LOCKED, but the lock was not acquired. The IR beam on ETMX was largely miscentered, and TR_GRX was only about 0.3.

I once set the initial alignment guardian to DOWN and manually adjusted PR3 to improve the GRX transmission:

• PR3 (P, Y): (62.8, -28.1) → (56.8, -28.5)
• TR_GRX: about 0.3 → about 0.6

After requesting IRX_LOCKED again, the lock was still not stable. When the beam was briefly visible during the lock trial, the IR beam was located around the upper-right side on ETMX and the lower-left side on ITMX. The displacement seemed to be mainly in yaw.

I then adjusted the ITMX and ETMX OPTICALIGN values:

• (IXP, IXY, EXP, EXY): (5.0, -13.0, -9.5, -11.0) → (5.8, -16.5, -8.2, -11.8)
• At some moments the IR beam looked reasonably good on the TCam, but the lock was immediately lost.
• TR_IRX was around 0.2 during such attempts.
• The transmitted beam often looked like a vertically split higher-order mode rather than TEM00.
• Tried (5.8, -16.5, -10.0, -11.8).
• Tried (5.8, -16.5, -7.0, -11.8); the TEM00 rate seemed slightly better.
• Finally tried (7.1, -16.5, -8.3, -11.8) because the beam still looked slightly high on ETMX.

Even after these adjustments, IRX_LOCKED could not be achieved. The lock attempts frequently ended up in non-TEM00 modes. This may indicate that the IR input axis, or the mismatch between the IR and GR axes, was the limiting issue?

Restoration

I stopped the trial and restored the alignment values:

• Set the initial alignment guardian to DOWN.
• Restored ITMX and ETMX OPTICALIGN: (IXP, IXY, EXP, EXY) = (5.0, -13.0, -9.5, -11.0).
• Restored PR3: (P, Y) = (62.8, -28.1).

I found some GAS filters are closing to the satuation values.
Do we need offload works?

Tanaka, Fujimoto

We found that IMC IP PZTs were applied strange offset during today's initial alignment. During initial alignment by ST students, IMC could not be locked because IMC alignment become worse. Whole we investigated the cause of the misalinment, we found the issue. 

PZT offset value is set to the middle (75 V) of the range (0-150V). On the other hand, offset value seems to be changed two times recently, May 11th and April 22th respectively (fig.1). In the PZT2, the offset value was changed to the negative value (-22V). The PZT driver could not be receieved the negative value because the negative voltage will break the driver or PZT itself. Since PZT offsets are not changed automatically, they are changed by human.

Fortunately, we restored PZT offsets to nominal value, then IMC alignment was restored.

Please do not touch PZT offsets if you cannot take care of the PZT driver or PZT themselves.

[Fujimoto, Tanaka, Takano]

Summary

We tried to lock DRMI for the first time since the last RSE trial. Several problems were found to be solved.

Detail

For the first time since the last RSE trial 6 years ago, we started the work on DRMI locking.

Preparation works:

1. The interferometer was first aligned by students from ST [Kawakami, Hasegawa, Abe] with the great help of Kenta.
2. We moved on to aligning SRM using SRY. We encountered a strange behaviour with the locking point, but Hiroki miraculously solved it.
3. After aligning SRM with SRY ADS, we offloaded the value using the offload function, but it didn't work well. Further investigation is necessary.
4. We misaligned SRM again and aligned PRMI. We manually aligned PRM to improve the POP90 buildup, but the maximum was ~ 0.4, which should be close to 1.
5. Once PRMI was locked with 3f signals, ADS was engaged, but it took a very long time, and the buildup never exceeded 0.5.
6. We stopped ADS and manually moved BS in pitch, then the buildup increased. This indicates that PRMI ADS doesn't work now.
7. Manual alignment tweaking for BS and PRM was done, and the buildup reached ~ 0.8.
8. While working, many glitches were observed and they disturbed (probably) MICH control. At the same time, bright beam spots appeared on the bottom of ITMX and ITMY. It seems that the reflection from SRM was close to the main beam and f1 sidebands barely resonated. We need to misalign SRM more, but unlike ITMs, MISLIGNED_BF state is not prepared for Type-B. Probably we have to move IPs.

DRMI trial

We realigned SRM and tried to lock DRMI in these ways:

1. Hiroki's idea: use 1f signals
   1. Plan to use REFL45I for PRCL, POP17Q for MICH, and POP17I for SPRCL.
   2. We estimated the expected gain for them using Finesse simulations and put these gains in the sensing matrix.
   3. Wait for a while. Noticed that the gain for SRCL was bigger by one order than the other and seemed to kick SRM a lot.
2. Satoru's idea: use 3f signals
   1. Plan to use REFL135I for PRCL, REFL51Q for MICH, and REFL51I for SPRCL.
   2. We intentionally misaligned SRM by changing the oplev setpoint and locked PRMI.
   3. We gradually aligned the SRM and hoped that PRMI would stay locked, but it turned out that MICH control was vulnerable to f1 resonance in SRC.
   4. Then we fully aligned SRM, and left all three servos on. We set triggers and waited for being locked. Sometimes AS34 buildup was held in < 1 sec, but it didn't last for long.

After several hours of trial and error, we concluded that we need to develop effective locking strategies.

Notes

While thinking of strategies, we found these issues:

1. The Schnupp asymmetry in KAGRA is too big for stable control of MICH
   1. In LIGO and Virgo, they are O(cm), and the Michelson reflectivity is almost 0 for f1, which makes the SRC undercoupled. Thanks to it, MICH signal with 1f and 3f is relatively insensitive to SRCL and easier to keep locking than KAGRA.
   2. For KAGRA, it's overcoupled (Michelson: R=15%, SRM: R=85%) and f1 buildup in SRC is quite big. This makes MICH sensitive to SRC resonance for f1.
2. We investigated the possibility of using f3. It is said that f3 doesn't resonate in any cavities, but Finesse simulations show that it actually resonates in both PRC and SRC. Are we wrong, or are the simulations?

We confirmed that this offset disappeared when ITMX was in MISALIGNED_BF state. For SRM alignment using SRX or SRY, we should misalign ITMY or ITMX by MISALIGNED_BF, not MISALIGNED.

Summary

I performed initial alignment for XARM, YARM, and OMC. XARM required small ITMX and ETMX adjustments, and YARM required a BS adjustment. For OMC, I also adjusted OMMT1 a little.

Details

Preparation

PRM and ITMX were changed to MISALIGNED. I tried to check the SRC flash by changing SRM to ALIGNED, but no clear SRC flash was observed. SRM was then returned to MISALIGNED.

Initial Alignment

I confirmed that FIB and PNC were locked as expected, and opened all laser shutters.

For XARM, IRX_LOCKED and GRX_LOCKED_WITH_IRX were achieved. During ALIGNING_XARM, I adjusted ITMX and ETMX: (ITMX P, ITMX Y, ETMX P, ETMX Y) = (5.0, -12.5, -7.9, -9.1) to (5.0, -13.3, -7.9, -9.6). I then recorded the good values and offloaded the alignment.

For YARM, IRY_LOCKED and GRY_LOCKED_WITH_IRY were achieved. During ALIGNING_YARM, I adjusted BS: (P, Y) = (6.8, -38.5) to (5.8, -37.2). I then recorded the good values and offloaded the alignment.

OMC

PRM was changed to MISALIGNED_BF to increase the laser power, and the GRX and GRY shutters were closed.

I adjusted the OMMT1 alignment to make the beam position on OMMT2 QPD at center. The OMMT1 TM OPTICALIGN was changed from (P, Y) = (-24000, -4400) to (-25000, -4600). The QPD sum was about 8.

OSTM was changed to MISALIGNED_FOR_LOCK_ACQ to avoid unnecessary OMC flashes. I then changed the laser power by rotating the PSL HWP from 150 deg to 162 deg, and then returned it to 150 deg. The beam on the QPD was confirmed to be the IR beam.

Recovery

OSTM was returned to LOCK_ACQUISITION, and PRM was returned to LOCK_ACQUISITION.

[Takano, Tanaka, Fujimoto]

Summary

While trying to lock the SRY on carrier resonance, we observed that the cavity occasionally locked to a dark state at transmission even though the sign of the feedback control had not been changed.
We found that this phenomenon is a carrier anti-resonant lock originating from the low finesse (~4) of the SRY and the FSR difference between the SRY and SRC.

Details

While aligning the SRM for DRMI locking using carrier-locked SRY (ITMX: MISALIGNED_BF, PRM: MISALIGNED, Guardian: vertex/SRY_1F_LOCKED), we observed that the SRY transmission ports (REFL, POP, and AS) sometimes locked to the bright state (carrier resonant, Fig. 1), but occasionally locked to a dark (anti-resonant) state instead (Fig. 2).
Therefore, we investigated the origin of this dark lock.

Fig. 3 shows time-series data obtained during a cavity scan by applying an offset of 30000 to K1:VIS-PRM_TM_TEST_L_OFFSET.
The blue trace is the error signal (POP17-I), while the orange, green, and red traces are AS DC, POP DC, and REFL DC, respectively.
From the transmission resonance peaks and the error signal, it can be seen that an error signal with the same sign as the carrier resonant lock exists even at the anti-resonant point.
The same behavior is also reproduced in the analytical simulation (Fig. 4).

The origin of this anti-resonant error signal can be understood as follows.
Since the current cavity configuration is SRY, its FSR differs slightly from that of the SRC (FSR_{SRY}=2.20 MHz, FSR_{SRC}=2.25 MHz).
Therefore, when the carrier is resonant in the SRY, the f1 sidebands are shifted away from the perfect anti-resonant condition (= half-integer multiple of the FSR), corresponding to

f_1 = 7.68 * FSR_{SRY}.

In addition, the SRY is a low-finesse cavity (F~4) formed by (BS ⇒ ITMY ⇒ BS ⇒ SRM ⇒ BS).
As a result, when the carrier is anti-resonant, the f1 sidebands are located on the side lobe of the cavity resonance peak.
Therefore, fluctuations of the laser frequency or cavity length produce asymmetry in the amplitudes of the sidebands transmitted through the SRY and entering the POP RFPD, generating an AM sideband component.
This AM sideband component beats with the carrier transmitted through the anti-resonant SRY, producing an error signal even at the carrier anti-resonant point.
This is the origin of the SRY dark lock.

If we wants to avoid this dark lock in the SRY, the feedback loop can be enabled/disabled using a trigger based on the transmission power.
As an additional remark, if the sign of the feedback control is inverted, the cavity locks on the side lobe of the transmission peak (Fig. 5).
The reason why this control point is slightly shifted from the sideband resonance frequencies (carrier resonance frequency ± 0.32 MHz) is presumably that the carrier light acting as the local oscillator is located on the side lobe of its resonance, introducing an additional phase shift and effectively changing the optimal demodulation phase.

[Yu, Shingo, Dan]

Summary

We installed the new laser unit in the rack near the Pcal-Y Tx module. A shelf was first mounted in the rack to support the installation work, and then the laser unit was fixed in place. No power-on test or functional test was performed today.

Work details

Today, we installed a new laser unit in the rack near the Pcal-Y Tx module.
Before installing the laser unit, we mounted one shelf in the rack. This shelf was added to make the installation easier and to support the laser unit during the work.
After preparing the shelf, we placed the laser unit in the rack and fixed it mechanically.

Current status

Today's work was limited to fixing the laser unit in the rack. We did not perform any power-on test or other operational checks.
The laser head has not yet been installed in its final position and is still placed on top of the laser unit.

[Yu, Hido, Dan]

Summary

We tested whether a QPD placed behind the steering mirror in the Pcal Rx module can be used to monitor the Pcal beam position. The transmitted laser power behind the steering mirrors was measured to be ~0.7 mW, which is consistent with the expected order assuming less than 0.1% transmission. A QPD signal response was observed when the ETMX TM Y setpoint was changed.

Background

During O4, the Pcal beam position was found to move over time. During weekly maintenance, we checked the beam position using the Tcam, and if the displacement was too large, the beam alignment was recovered using the pico-motors.

The Pcal beam position is the second largest factor contributing to the Pcal uncertainty. Proper treatment of the Pcal beam position is also one of the characteristic points of the KAGRA Pcal system. Therefore, we are trying to develop a beam position monitoring system. Ideally, we would like to monitor position changes of about ~+/-2 mm at the Rx side, which are difficult to detect clearly with the Tcam.
Also if we have this system, we can alwasy monitor it even during the IFO lock.

Purpose of today's work

• Check whether it is realistic to place a QPD behind the steering mirror in the Rx module.
• Measure the transmitted laser power through the steering mirror.
• Place a QPD and check whether the QPD signal responds when the ETMX TM Y setpoint is changed.

Power measurement behind the Rx module

We measured the laser power behind the steering mirrors in the Rx module using a power meter.

• Path 2, right side: 0.78 mW
• Path 1, left side: 0.67 mW

These values are consistent with the expected order. Since the input laser power is about 1 W, 0.1% transmission would correspond to about 1 mW. However, 0.1% is the upper limit from the mirror specification, so a smaller transmitted power is reasonable.

The measured values will be used as a reference for designing the QPD setup to be installed.

QPD test

We placed a QPD(PDQ80A) behind the Path 1 steering mirror. The signal was acquired using a KPA101.

We changed the ETMX TM Y setpoint as follows[urad]: -7.4 -> +20 -> -20 -> -7.4

The QPD output at the Rx module changed approximately as follows:

(X, Y) = (0, +0.1) -> (-0.6, -0.6) -> (+0.1, 0) -> (0, +0.1)

We confirmed that the QPD output changes in response to the motion of the TM.

In this test, no lens was placed in front of the QPD. Since the beam diameter is large, about 10 mm, it may be necessary to place a lens before the QPD in the final setup.

Due to limited time, today's test was limited to confirming that the QPD output responds to the TM motion even without a lens.

Conclusion

The transmitted power behind the Rx steering mirrors was measured and found to be reasonable. The QPD signal also showed a response to the ETMX TM Y motion. This result suggests that a QPD-based monitoring system behind the Rx steering mirror is a realistic option for monitoring Pcal beam position changes. Further work is needed to design the final QPD setup, including the possible use of a lens.

Summary

The estimated frequency noise induced by IMC suspensions is ~1 MHz peak-to-peak, which is inconsistent with the observed fluctuation of the beat note with the auxiliary laser. Other sources are suspicious.

Detail

During the PLL work, we have a feeling that it is hard to grab the linear range of the PLL signal because the beat note fluctuates a lot. Without anything, it moves O(MHz) in ten seconds and sometimes we saw a big jump in ~ 5 MHz. The error signal was almost always just white noise, and it seemed to be available for only a tiny moment (< 1 sec, ~ 0.1 sec?). We suspected that this fluctuation came from the main laser, because in these frequencies it follows IMC length and there are many resonant peaks of the IMC suspensions.

To verify this theory, I investigated the frequency noise using the calibrated channel: K1:CAL-CS_PROC_IMC_FREQUENCY is calibrated for the laser frequency fluctuation (up to a few kHz, according to Yamamoto-san). When IMC was locked to the laser frequency, I measured the PSD as attached, which indicates an RMS laser frequency fluctuation of ~160 kHz. Multiplying by a factor of 6, the peak-to-peak value of the fluctuation is ~ 1 MHz. This implies that other noise sources contribute to the beat note fluctuations. One possibility is phase noise induced by the suspensions, like the green lasers.

Notes

I also measured the PSD of the channel when MCL control was turned on after IMC was locked, i.e. both the laser frequency and MCL control were engaged. The RMS value was reduced to 11 kHz. But I'm not sure whether this value is an underestimation of the laser frequency fluctuation

I noticed a strange offset in POP17I used for locking SRY, as shown in the attached. It changed from -0.93 to 0.86 while SRY was freely swinging, so it was slightly off-centred. When locking the SRY using it, AS DC power increased to 0.005, while the maximum power when freely swinging was 0.007. It reached the maximum value when an offset was added to the error signal and it was locked around 0.17, not 0. Probably related to this offset, ADS didn't work well; when it was engaged after locking SRY without any offset on the error signal, AS DC power stayed the same, and nothing changed after tuning the offset.

Last week, we didn't see such an offset; SRY was locked to the carrier resonance without any offset tuning, and ADS seemed to work well. We need further investigation for stable locking of DRMI.

Sorry, I forgot to attach the figure 1.

With Shingo Hido

Summary

We performed initial alignment for XARM, YARM, and OMC using the initial alignment guardian. XARM was aligned only with ADS, while YARM required a small BS adjustment. We also confirmed the IR beam on the OMMT2 trans QPD and checked the SRC flash.

Details

Preparation

PRM was changed to MISALIGNED. SRM was already in MISALIGNED. We briefly changed SRM to ALIGNED and confirmed that the SRC flash was visible, then returned it to MISALIGNED. (fig_001)

Initial Alignment

We confirmed that FIB and PNC were locked as expected, and opened all laser shutters.

For XARM, IRX_LOCKED and GRX_LOCKED_WITH_IRX were achieved. Then we requested ALIGNING_XARM. No manual adjustment was needed, and the alignment was done only by ADS. We then recorded the good values and offloaded the alignment.

For YARM, IRY_LOCKED and GRY_LOCKED_WITH_IRY were achieved. During ALIGNING_YARM, we adjusted BS from (P, Y) = (9.6, -39.3) to (8.6, -38.3). We then recorded the good values and offloaded the alignment.

OMC

We followed the procedure in klog36759. PRM was changed to MISALIGNED_BF, and the GRX and GRY shutters were closed.

We tweaked the OMMT1 angle using coil-magnet actuators to adjusted the beam on the OMMT2 trans QPD. The OMMT1 TM OPTICALIGN was changed from (P, Y) = (-22800, -4200) to (-24000, -4400). The QPD sum was about 8.5.

OSTM was changed to MISALIGNED_FOR_LOCK_ACQ to avoid unnecessary OMC flashes. We then increased the laser power by rotating the PSL HWP from 151 deg to 162 deg, and then returned it to 151 deg. (fig_002) The IR laser power was therefore restored to the original 1 W. The beam on the QPD was confirmed to be the IR beam.

Recovery

OSTM was returned to LOCK_ACQUISITION. We also checked the SRC flash by setting only ITMY to LOCK_ACQUISITION among Type-A suspensions, keeping the others MISALIGNED, and changing SRM to ALIGNED. The SRC flash was confirmed, and SRM was returned to MISALIGNED. (fig_003)

Finally, PRM was returned to LOCK_ACQUISITION.

[Kawakami, Hasegawa, Abe, Tanaka, Fujimoto]

Summary

We adjusted position of LEN QPD. In the result, coupling from yaw motion to LEN sensor was reduced.

Details

We measured spectrum of the SRM OPLEV QPDs with covers but the coupling from yaw motion to LEN sensor became worse comparing with the no cover case.

Therefore we adjusted position of a QPD which measures LEN in order to reduce coupling. 

A 10 Hz sinusodial excitation(gain: 3000) was applied to the SRM in the yaw direction, and the micrometer controlling the longitudinal position of the LEN QPD was adjusted.
Subsequetly, the specra of the LEN and TILT QPDs were measured.
During the measurement, the centering of the LEN QPD drifted, so it was readjusted as neccesary.

When the micrometer was moved in the direction that reduced the coupling, it reached the limit of its range of motion. The spectrum of LEN Horizontal obtained at that point is shown by the red solid line in Fig.1.
As the result, the coupling improved compared with that before adjusting the LEN position. However, it could not be improved beyond the result obtained on May 13th.

A 10 Hz sinusodial excitation(gain: 30000) was applied to the SRM in the direction along with laser. The spectrum obtained at that point is shown by Fig.2.
In the result, a peak on 10 Hz appeared in the LEN and TILT VER (Fig.2 on the lower side) which can be regarded as a reasonable result. 
Otherwise, the value of a peak in the LEN and TILT HOR (Fig.2 on the upper side) is close to the noise floor.

SRM OPLEV is ready to use.

[Kawakami, Hasegawa, Abe, Tanaka, Fujimoto]

Summary

For the SRM OPLEV LEN QPD whose mount had been modified yesterday, we restored the mount to its original configuration, installed the LEN QPD cover, and reinstalled the QPD on the OPLEV table.

Details

During yesterday’s installation of covers onto the SRM OPLEV QPDs, it was found that the mounting structure of the LEN QPD differed from that of the TILT QPD, preventing installation of the cover onto the LEN QPD.
Therefore, modification of the LEN QPD mount had been necessary.
However, it was later pointed out that a dedicated cover for the LEN QPD existed.
After searching in the mine, we found the LEN QPD cover on a shelf along the IMC.
Therefore, the mount modified yesterday was restored to its original configuration, and the dedicated LEN QPD cover was installed.

Fig. 1 shows the modified LEN QPD (right) and the unused QPD left on the POS table (left), after yesterday’s exchange of the (adjustment stage + mount) assemblies.
The exchanged assemblies were swapped back again, restoring their original configuration, as shown in Fig. 2.
This also restored the longitudinal adjustment capability of the LEN QPD.

Fig. 3 shows the photo of the layout after installing the LEN QPD onto the OPLEV table and attached the cover.
The LEN QPD was installed at a position 273 mm away from the folding mirror so that its position remained unchanged from the previous setup.

We found that PRM misalignment state when REFL power bulidup was measured on May 13th, 2026 (klog36922) was differ from the state when the power was measured on May 19th, 2026 (klog36891).

Fig. 1 shows the timeseries the power on AS, REFL, and POP PDs, and the guardian state numbers of ITMX and PRM. Left panel shows them on May 19th and right panel shows them on May 13th. According to this figure, the following table is summarized the relation between the states and the REFL build up ratio.

So one of possibilities why the REFL power buildup becomes low is that PRM misalignment angle in the MISALIGNED state is also not enough. 

I have asked engineers in NAOJ/ATC for the latest 3D data of each QPD cover, and uploaded them to JGWdoc.

• For tilt oplev 17360.
• For length oplev 17361.

There should be the QPD covers for the length QPD setups. According to my e-mail records, the products arrived at Kamioka on Oct 12 2022.

Yesterday, I took TCam images for the Pcal beam position estimation for the first time in about one month. During this work, the TM reference position used in the image analysis was also adjusted.

Adjustment of TM reference position

• Pcal-X: The TM reference position was changed from (1942, 1410) to (1958, 1410). This corresponds to a displacement of (+16, 0) pixels, or approximately (+1.4, 0) mm.
• Pcal-Y: The TM reference position was changed from (2250, 1000) to (2250, 1010). This corresponds to a displacement of (0, +10) pixels, or approximately (0, +0.9) mm.

Discussion

Compared with the images taken about one month ago, the TM reference position changed by about 1 mm. On the other hand, the estimated Pcal beam positions showed little change and remained stable within approximately 1 mm.

Initial alignment was performed for XARM, YARM, and OMC. XARM and YARM alignment were completed using the initial alignment guardian. For OMC, the procedure in klog36759 was followed.
After the work, I conformed SRC flash with aligned SRM and ITMY.

Attached figure is the flash after the initial alignment works.