last night's situation. Still, there are many unlocks of IMC (Fig.1). 

Fig.2 shows a quiet situation. Fig.3 shows the unlock timing. Fig.4 shows a slightly noisy situation. Fig.5 shows a huge glitch without unlock.

Ushiba, Tanaka

## Abstract

Frequncy noise seems to sometimes become noisy and it makes IMC lock acquisition difficult. We changed the temperature of the laser crystal but the situation is not changed. We need more investigation.

## What we did

we found that IMC lock acquisition becomes difficult when PMC error signal (K1:PSL-PMC_MIXER_MON_OUT_DQ), which reflects laser frequency noise, becomes noisy. Also, the frequency noise sometimes becomes noisy. fig.1 shows the spectra and time-series of PMC related signals at the moment when the PMC error signal becomes noisy. As for spectra, blue lines show each spectra  when the PMC error signal was relatively quiet, and red lines show each one when the PMC error signal became noisy like the time-series in the upper left panel in the ndscope window. As you can see, the spectra of the PMC error singal (K1:PSL-PMC_MIXER_MON_OUT_DQ) and the feefback signal (K1:PSL-PMC_PZT_HV_MON_OUT_DQ) above 100 Hz becomes large. On the other hand, Intensity spectra of a laser source (K1:LAS-POW_FIB_OUT_DQ) and a PMC transmitted light (K1:LAS-POW_PMC_OUT_DQ) seem not to be changed significantly.

Last night, we swept the setpoint of the temperature control for the laser crystal by sweeping K1:IMC-SERVO_NPRO_TEMP_BIAS_OFFSET from 0 to 1 to find the temperature region where the PMC error signal seemed to be relatively quite. This time, I set the offset value to 1 and turned on the null filter with 10000 ramp time. As long as I remember, when Miyakawa-san and I changed the BIAS_OFFSET value from 0.4 to 0, the setpoint of temperature control was changed by 2 degrees, that is, we scanned the laser temperature within 5 degrees. (Miyakawa-san, please give some comment if you remember) 

Anyway, Fig.2 shows the overview of the sweep. Unfortunately, at last ~4 mins, PMC lock was lost due to the actuator satutaion. Moreover, around "-15h" PMC seemed to be lost. Fig. 3 shows the time-series around "-15h". According to this figure, the PMC trans. power drastically decreased before the lock loss even though actuators were not saturated and the FIB_OUT power did not decrease. This may indicate that the mode-hop occurred aroung this temperature. 

Using this result, we found the temperature region where not only the PMC error signal but also PLLX/Y error signals seemed to be relatively quite. Fig. 4 shows the region where the BIAS_OUTPUT value is from -6200 to -4200 cnts. From this, we set the BIAS_OFFSET value to 0.3175 so that the OUTPUT value becomes -5200 cnts (= 0.3175*(5*3276), (5*3276) comes from the gain value labeld "cali" in FM1 of the K1:IMC-SERVO_NPRO_TEMP_BIAS filterbank). Also, the feedback amplitude from IMC is allowed within +/-0.06 (= +/-1000/(5*3276)) so that the temperature keeps in this range.

After today's finesse measurements for X/Y arms, we locked only PMC and monitored the PMC and PLL error signals in this state. Unfortunately, however, after ~40 mins, all of errror signals became noisy (fig.5). 

We changed the BIAS_OFFSET value to 0.425 by obtaining it with the same procedure. But this value also seems to be bad (fig.6).

We need more investigation.

## Next to do (plan)

• monitor the PMC error/feedback signals with Moku:lab to measure their higher frequency region
• change the applied current to the power amp or the master laser?
•  

## Note

We found that the error signal behaviour seems to be often changed suddenly like fig.7   

In the dark condition, a red LED light? from the Mephisto and a white indicator of neoLase seem to illuminate.

Violet spots are the IR spots and scattered light. The back-scattered light from the Amp is expected to be mitigated at IFI in photo.2

There seems to be ~ 4Hz oscillation in PMC trans. (Fig.1)

One of the glitch noises in K1:PSL-PMC_MIXER (Fig.2). The width is ~ 500usec level.

I increased ETMY IM heater for further warming up (fig1).

I increased ETMX IM heater for further warming up (fig1).

>When the PRFPMI+RF lock on December 23, 2024, the K1:PSL-PMC_MIXER signal amplitude seemed to be half compared with the present quiet state, and no glitches existed during the long lock.

Fig.1 is the data at the time.

While, Fig.2 shows the recent data. At this timing, K1:PSL-PMC_MIXER signal can be comparable in the last year. In addition, the amplitude of IMC-SERVO_SLOW is also comparable. While FAST is larger than before. We need to consider the crossover frequency for the precise comparison.

Information

• Measurement date: 2025/8/6, 2025/8/21
• Analysis date: 2025/9/2
• Members:
  • Misato Onish
  • Dan Chen

Background

• The TSK, recently calibrated at LIGO and transported to Japan, was used to measure the output ratio against the GSK.
• Measurements were carried out three times on two separate days.
• The purpose was to cross-check the consistency between the two standards.

Measurement

• The raw results are available here.

Quantitative results

• Toyama (3 measurements, average): TSK/GSK = 0.9429 ± 0.0011
• LIGO calibration: TSK/GSK = 0.9410 ± 0.0013

These values agree within their respective uncertainties.

Key points

• Both TSK and GSK had been independently calibrated at LIGO. When compared at Toyama University, their output ratios showed consistency within error.
• GSK was calibrated several years ago, while TSK was calibrated at LIGO just before this summer.
• The agreement indicates that the long-term drift of the GSK is small.

Summary graph: Attached

Last night, PMC was locked for ~ 9 hours. Although the lock itself was kept, we can see several noise excitation in the K1:PSL-PMC_MIXER signal as reported klog#34978. The excitation seems to be kHz-level glitches, not sinusoidal excitations as Fig.1 and 2.

When the PRFPMI+RF lock on December 23, 2024, the K1:PSL-PMC_MIXER signal amplitude seemed to be half compared with the present quiet state, and no glitches existed during the long lock.

Now, I'm sweeping K1:IMC-SERVO_NPRO_TEMP_BIAS_OFFSET from 1 to 0 to find the range where the mode-hop is not occured. According to the medm setting of the BIAS_OFFSET, since we can change the value from 0 to 1 I decided to sweep the value in this range. In order to keep PMC lock, the sweep speed is very slow so it will take 2.8 hours (10000 secs) to finish the scan. 

Today, I left PMC in this state. So please do not touch PMC, especially the K1:IMC-SERVO_NPRO_TEMP_BIAS filter bank and the k1ctr3 PC until 20:00 JST without any contact to me.

Ushiba, Tanaka

We began to investigate the reason why IMC lock is unstable, and we found that the RMS of the PMC error signal (K1:PSL-PMC_MIXER_MON_OUT_DQ) when the IMC lock acquisition become unstable (top in fig.1) seems to be 4x larger than the one when IMC lock acquisition seems to be relatively stable (top in fig.2). This indicates that the laser frequency noise sometime become noisy for some reason. (Originately, the error signal seems glitchy even when IMC lock become stable... )

We locked only PMC and started to monitor the noise to investigate the cause of increasing the RMS.

As for PLLX, current ALS-X_LASER_TEMP_BIAS_OFFSET value was close to the edge of the range. According to Ushiba-san, current PLLX guardian sweeped the temperature of GRX laser crystal with this offset and checked whether the PLLX error signal crosses 0. However, in this situation, the BIAS_OFFSET value which the error signal was 0 seemed to be sometimes the out of the sweep range. If so, PLLX could not be locked.

So, we tweeked the setpoint of the laser crystal temperature control so that the PLLX error signal could be observed when the BIAS_OFFSET value is around 0.  After that PLLX lock acquisition seems to become stable. 

We removed the lid of the NeoLASE box to improve the ventilation around the NPRO laser box (fig.1), according to Miyoki-san's advice . We hope that the situation about temperature controls is improved.

PMC HEATER control became crazy when IMC unlocked as Fig.1. Then SERVO_NPRO became suddenly around zero from 0.05.

Also, SERVO_SLOW shows ~ 0.1 again. In addition, SERVO_NPRO_TEMP did not show any changes as before. In these conditions, the relock IMC might be difficult.

PLLX was set down due to numerous failures yesterday. Sorry for my late report.

IMC showed unlock and successive failure of relock around 5:30. I easily noticed that BOOST in the IMC LOCK GUARDIAN failed every time.

So I manually requested DOWN, GRAB_RESONANCE, DOWN, BOOST, and IMC_Locked, then checked several signal situations. Consequently, IMC could be relocked soon.

Fig.1 shows each signal during "GRAB_RESONANCE, DOWN, BOOST, and IMC_Locked". SERVO_SLOW(PZT path) showed signals around zero. Then the IMC lock was recovered. The two dashed lines in  SERVO_NPRO_TEMP show the difference between the relock of IMC and the repeating time of relock and unlock.

Fig.2 shows the longer time range from "several failures of boost" to manual " DOWN, GRAB_RESONANCE, DOWN, BOOST, and IMC_Locked".  It showed the SERVO_SLOW changed from ~ 0.1 to around zero, and also that SERVO_NPRO_TEMP drifted up. This shows that  SERVO_NPRO_TEMP drift contributed to reducing the DC value in SERVO_SLOW, then it enabled BOOST and IMC relock.

Fig.3 shows several PT temperature changes for the last 3 days. PT06 is smoothly approaching the equilibrium temp. PT02/04 shows a strange up and down. These are because of unstable current supplier or some contacting point error in the cable that used to be observed as before. So, the smooth reduction is an expected phenomenon.  SERVO_NPRO_TEMP also shows the same kind of drift.

Fig.4 shows the longer time span from several hours IMC lock to unlock and recovery. Before unlock, SERVO_FAST and SLOW showed larger signals. Also, SERVO_NPRO_TEMP drifted up.

Consequently, I guess the IMC unlock was caused by the signal saturation in somewhere because of

• the insufficient balance between thermal path and PZT path,
• temp drift in the PSL room after the activities.

Also, I guess the several times relock failures were caused by the large DC values in SERVO_SLOW during boot.

I reduced the current for ITMY IM heater.

This is just report.

Around 15:00, IMC unlocked and continued to fail to lock. So I just requested DOWN state and wait 2 hours. Then I requested IMC_LOCK state, then IMC locked smoothly. During this ~ 2 hours, no lock loss of PMC. So IMC lock losses triggered the lock loss of PMC.

However, IMC SERVO signals started showing spike-like noises as Fig.1 after several minutes of the relock. The lock itself seemed to be kept. After several minutes, these spike-like noises vanished.