We checked that the loops were working well after the ACL update.

We sent SQZ_MAIN SQZ_LOCKED_NO_FC and it went at the first attempr

We then aligned the filter cavity but we had to modify some of the code lines in the FC Alignment interface.

We locked the FC both with only the mirror and also with the laser. We updated the gains. SQZ_CC GAIN=4 and LFC_Z_GAIN = 6 instead SQZ_CC GAIN=1.55 and LFC_Z_GAIN = 6

We also run a SQZ CC Phase can with SR tuned at 17:34 UTC CC GAIN 40000. We also noticed that the CC sensing noise is dominant above 100 Hz

ITF Found in LOCKING_ARMS_BEAT_DRMI_1F State and COMMISSIONING Mode.
All times are UTC.
07:00 MAINTENANCE started.
07:09 Acl Update step 3 Started (Masserot).
07: 20 INJ beam blocked for INJ Acl Update (Gosselin, remote), then I requested DOWN to INJ_MAIN Metatron node and I paused ITF_LOCK.
07:27 Acl Update step 3 Completed for INJ and SQZ, while TCS, and SAT postponed afted the shutdown (Masserot, #68939).
07:35  INJ beam block removed (Gosselin, remote).
07:43 INJ in IMC_RESTORED.
10:43 ITF in LOCKED_ARMS_IR.
11:00 MAINTENANCE ended.
12:57 ITF in LOW_NOISE_3_ALIGNED;

Task performed by the operator during the maintenance:
- From 08:25 to 8:35 OMC Lock;
- From 08:38 to 08:59 OMC Scan;
- 09:08 TCS Chillers Refill;
- 09:18 TCS Thermal Camera Reference;
- 11:15 TCS Power Checks at the Pickoff:

This line's amplitude continuously decreased and cannot be clearly seen after midnight today.

In the injection alignment strategy, the vertical setpoint of the BPC is still defined by the DC NF QPD located on the EIB.

This morning, we performed a quick test to check whether this vertical setpoint has any impact on the RFC alignment. We shifted the beam by approximately 1300 µm and observed an increase in the RFC transmission of about 15% (see figure attached).

During the final steps, we also noticed that IMC_TRA increased to 18.2 W, then dropped to 17.4 W before the system eventually unlocked. Observing the reflected beam, it was clear that the beam was misaligned on the IMC, and its transmission was unstable.

We then returned the BPC vertical setpoint to its initial position.

Although this does not necessarily mean that the RFC misalignment originates from this vertical shift, it would be worth dedicating more time to investigate whether a different IMC alignment, combined with a higher BPC Y setpoint, could lead to improved RFC transmission.

All NCals have been stopped during the maintenance.

I expect this bump to be due to the squeezer shutter being opened but the squeezing not being injected. From what I remember the shutter on SDB1 towards squeezing was openend on March 12-13, then closed for the weekend and reopned in the middle of the week. And it has remained opened since then.

This wandering line was likely already present in the March 13 data, but not in the March 12 data, i.e., during the first recovery after the electrical shutdown. The onset of the source may therefore be uncorrelated with the shutdown.

I am attaching two plots showing the median-normalized spectrogram of the strain: one covering March 12–13, and another starting from March 18, where the spectral line is particularly visible with the better sensitivity in the detuned SR configuration achieved in recent days.

Could be related to this issue found in the past: #61815?

The Acl's servers running on the following rtpcs have been restarted with the Acl release v1r39p17

• ASL rtpcs : rtpc22 (trend plot), rtpc24(trend plot)and rtpc10 (trend plot)
  • Operations performed between 2026-03-23-06h22m00-UTC and  2026-03-23-06h39m45-UTC
  • the ITF has been unlocked at 2026-03-23-06h30m21-UTC  to performed the update of the Acl's rtpc server.
  • After these operations the ITF has been successfully locked upto LOW_NOISE_3_ALIGNED  at  2026-03-23-07h10m46-UTC
  • ITF unlocked  2026-03-23-07h23m08-UTC to continue the Acl deployment 
• SPRB and LNFS rtpc: rtpc6 (trend plot)
  • LC and SBE loops openned 
  • Operations performed between 2026-03-23-07h25m14-UTC and 2026-03-23-07h33m42-UTC
  • LC and SBE loops successfully closed 
  • After these operations the ITF has been successfully locked upto CARM_NULL_1F  at  2026-03-23-07h45m21-UTC
  • ITF unlocked  2026-03-23-07h47m16-UTC to continue the Acl deployment
• SWEB  rtpc: rtpc9 (trend plot)
  • LC, SBE, PCAL and NCAL loops openned
  • Operations performed between 2026-03-23-07h47m55-UTC and 2026-03-23-07h59m27-UTC
  • LC, SBE, PCAL and NCAL loops successfully closed
  • After these operations the ITF has been successfully locked upto LOW_NOISE_3_ALIGNED at  2026-03-23-08h30m48-UTC
  •  ITF unlocked  2026-03-23-08h31m17-UTC to continue the Acl deployment
• SIB2 rtpc: rtpc18 (trend plot)
  • LC and SBE loops openned 
  • Operations performed between 2026-03-23-08h32m26-UTC and 2026-03-23-08h34m43-UTC
  • LC and SBE loops successfully closed
  • After these operations the ITF has been successfully locked upto CARM_NULL_1F at  2026-03-23-08h48m12-UTC
  •  ITF unlocked  2026-03-23-08h48m58-UTC to continue the Acl deployment
• SDB1, SDB2  rtpcs: rtpc1(trend plot) and rtpc13 (trend plot)
  • LC(SDB1,SDB2), SBE(SDB2) and OMC  loops openned 
  • Operations performed between 2026-03-23-08h50m27-UTC and 2026-03-23-09h00m26-UTC 
  • LC(SDB1,SDB2), SBE(SDB2) and OMC  loops successfully closed
  • After these operations the ITF has been successfully locked upto CARM_NULL_1F at 2026-03-23-09h17m-UTC , and to LOW_NOISE_3_ALIGNED at  2026-03-23-09h42m24-UTC

Acl 's upgrade  second block complete 

A faint but large spectral artifact (bump) can be seen wandering around the 50 Hz mains spectral lines as one can see on figure 1. This artifact can go down to around 40 Hz and seems to have appeared around march 17th/18th. It was not present at the beginning of february for instance.
Recent BruCo reports do not seem to find any coherent channels with it.

ITF found at CARM_NULL_!F, relocking up to LN3_ALIGNED. COMMISSIONING mode.

The planned activity on "Acl Update step 2 (Masserot)" went on without major issues from 07:00 to 10:00 UTC.
Infrastructure civil works by external company, close to MC building, from 11:00 to 13:00 UTC. ITF remained stable locked at LN3_ALIGNED.

considering all the channels:

DQ_BRMSMon_BRMS_SEA_SEIS_100mHz_200mHz_ENV_CEB_SEIS_N_mean. 

DQ_BRMSMon_BRMS_SEA_SEIS_100mHz_200mHz_ENV_CEB_SEIS_V_mean

DQ_BRMSMon_BRMS_SEA_SEIS_100mHz_200mHz_ENV_CEB_SEIS_W_mean

DQ_BRMSMon_BRMS_SEA_SEIS_300mHz_500mHz_ENV_CEB_SEIS_N_mean

DQ_BRMSMon_BRMS_SEA_SEIS_300mHz_500mHz_ENV_CEB_SEIS_V_mean

DQ_BRMSMon_BRMS_SEA_SEIS_300mHz_500mHz_ENV_CEB_SEIS_W_mean

we can set the limits for which it is more probable to be not locked than locked

In the past, a 1 Hz comb was observed to originate from the TCS chillers when operating outside the power-save mode (elog #40113, #63357, #63358)

Following the electrical power outage on Feb. 28th, a similar comb is observed again. The comb is visible in the current and voltage monitors UPS_CURR_* and UPS_VOLT_*, as well as in the magnetometers CEB_MAG (CEB hall) and IB_MAG (IB tower base), Figure 1.

The noise comb is visible in the following time intervals, Figure 2:

• time slot 1--> Mar 2, ~09:53-09:55 – 10:30-10:35 UTC

• time slot 2 --> Mar 2, ~14:54 – Mar 6, ~12:20 UTC

• time slot 3 --> Mar 6, ~14:30 - 14:50 UTC 

According to the logbook, several recovery activities were carried out on Mar 2, shortly after the power outage. Among them are the TCS lasers and chillers recovery (elog #68787) and the INJ recovery (elog #68793). 

Time slot 1 

During the morning, the INJ chillers were switched on first (top), followed in time by the CO2 main laser chillers (middle), Figure 3. 
The switch-on involved the EE room chiller (*CHILLER_NEO*, for NEOVAN electronics) and the chiller in the chiller room (*CHILLER_HEAD*, for the NEOVAN head), while the chiller serving the beam dumps and slave laser was already on and is not shown.

Two distinct sets of 1 Hz combs are observed in the UPS channel (bottom), each appearing and disappearing at different times, indicating that they are likely associated with the operation of different devices. The INJ chillers are switched on before the occurrence of the first comb, although no direct temporal coincidence is observed. The CO2 main laser chillers are switched on before the occurrence of the second comb; however, a clear correlation is observed at switch-off, as the disappearance of this comb coincides with the switch-off of the CO2 chillers.
The INJ chiller in EE room is a spare unit of the TCS system. A dedicated test was performed on Mar 9, during which the power-save mode of the INJ chiller was temporarily disabled (16:30–16:35 UTC). Following this change, the 1 Hz comb reappeared in the CEB and IB magnetometers, Figure 4. Thus, such device can generate this type of noise when operating outside the power-save mode.

Overall, the switch-on times do not show a strict one-to-one correspondence with the comb occurrence, while a clearer coincidence is observed for the switch-off of the CO2 chillers. 

Time slot 2

The 1 Hz comb is observed to persist from Mar 2 to Mar 6. During the afternoon of Mar 2, as reported in elog #68787, several activities were carried out in the TCS area, including the identification of a malfunction in one of the RF driver of the NI CO2 laser. As shown in Figure 5, the switch-on of the CO2 chillers does not show a clear coincidence with the presence of the 1 Hz comb. In addition, the elog entry reports that the chillers were verified to be operating in the correct configuration (power-save mode). This suggests that the observed comb in this time period may not be directly related to the chiller operation, and could instead be associated with the RF driver malfunction.

Time slot 3

On Mar 6, activities were carried out in the TCS area, including the replacement of the NI CO2 laser RF driver (elog #68263). As shown in Figure 6, the comb disappears around 12:30 UTC, in coincidence with the switch-off of the NI chiller, although the chillers were previously verified to be operating in the correct configuration. However, during the subsequent activities (Figure 7), including tests and the RF driver replacement, no clear one-to-one correspondence is observed between the chiller operation and the presence of the comb. In particular, after the intervention, the system was fully re-powered, with the lasers switched on at 14:31 UTC, and normal operating conditions were progressively recovered.

These observations suggest that, while the chillers can generate a 1 Hz comb (as also supported by dedicated tests), the RF driver malfunction may represent an additional or independent source contributing to the observed comb structure. Notably, the comb is no longer observed following the RF driver replacement.

Friday we performed the injection with the ITF locked in LOW_NOISE_3_ALIGNED.

We tested three current levels:

• 11:19 UTC: 5V, 0.5 A (readings on amplifier display)
• 11:32 UTC: decreased amplitude by a factor 2: Amplifier's display readings 2.5V, 0.3 A  (note, the accuracy of the amplifier display is 1 digit)  
• 11:39 UTC: reduced the amplitude of the injected signal from 1.2 V  to 0.24 V, tush reducing the amplitude of the injected signal by an additional factor 5

We kept injecting this signal until the ITF unlocked at 12:39 UTC.

The comb is well visible in hrec up to a few hundred Hertz. In addition to the current and the ddp (ELECTRIC) monitors,  the comb is also visible in all magnetometers in CEB (Fig. 2). In some places (BS, CEB, ...) the injected 11 Hz line exceedes 10 nT (Fig.3)

The comb amplitude in hrec has a steeper frequency slope than in the monitors and magnetometers, roughly 1/f^2 steeper (Fig. 2).

From a quick look the amplitude of the signal scales proportionally in hrec and all sensors (Figure 4).

In order to test if magnetometers were sensing a true magnetic field or they were fooled by a fluctuating ground

• at 13:15 UTC, we set the BS Bartington magnetometer on battery supply disconnecting it from the mains: the comb signal did not change (Figure 5)
• then, we read the BS magnetometer signals with a portable spectrum analyzer: the comb was there and the reading was the same as from the Virgo ADC. Again, we repeated the reading with the spectrum analyzer on the mains (Fig.6), and on batteries (Fig.7), the reading was the same.

We trust the magnetometers reading is a true magnetiec field produced in the CEB hall.

I ve runned the statistical analysis in 1 year of data understanding which are the ocean conditions that allows to lock the ITF.

I ve used the DQ_BRMSMon_BRMS_SEA_SEIS_100mHz_200mHz_ENV_CEB_SEIS_N channel to evaluate the distribution of the weather condition and evaluate which is the number of succesfull locks (normalized on the total number) against the unsuccessful ones.

The ratio of these quantities gives an idea on the lock probability.

In figure 1 it is visible that for 1.4e-6 the probabilty to dont reach the lock (META =135) is 10 time bigger than the probability to lock.

A reasonable value should be then 1e-6. to be addressed on all the channels

At 13:43 UTC, all NCals and associated PCal lines have been moved from near 36 Hz to near 18 Hz.

I have noticed the lock acquisition being stopped in down. Checking DET_MAIN it was stuck in checking shutter closed, the slow shutter was actually open (no command was sent to close after the unlock). After pressing force check closed, and the force check open. I could go to the shutter open state, and the shutter closed state of the automation. When I started the B1s photodiode had its shutter closed, so likely a flash closed the photodiode when the automation was trying to check the state of the slow shutter, and it confused it. 

It is good that the automation waited in this confused state, instead of allowing the lock acquisition to continue. But we should add some logic to let the automation get out of checking shutter closed after a minute, and retry a few times before giving up.

This happened again this afternoon. Around 16:39 UTC I have opened the shutter of the B1s photodiode and enabled to voltage bias. This was enough to unstuck the DET_MAIN automation without any other action.

I have added up to 4 attempts of enabling the B1s bias again during SHUTTER_CHECK_CLOSED, and reloaded the DET_MAIN node, to see if that solution can be automated.

I checked the effect of the phase scan in different range of Hrec. The following plot are composed by three subplots:

• First line: Hrec BRMS in different bandwidth normalized in dB, 0dB means shutter closed of the SQZ
• Second line: Blue plot B1_4MHz_mag i.e. the magnitude of the beat note between 4MHz CC beam and B1 Beam on B1_PD
• Third line: BNS Rang

I did the plots with three different GPS:

• Figure1: 19 March - 16:00, SR tuned Best alignment of the squeezing performed during the week
• Figure2: 20 March - 16:10, SR detuned First phase scan
• Figure3: 20 March - 17:33, SR Detuned Second phase scan

Not clear if the SQZ injection improved the range with SR Detuned, interesting that:

• with SR  tuned the maximum of the 4MHz mag correspond to ASQZ in high frequency and that the phase dispersion between frequencies range in the bucket and at higher frequency is quite negligible
• with SR detuned the maximum of the 4MHz mag correspond to SQZ in high frequency and that the phase dispersion between frequencies range in the bucket and at higher frequency is big, they have quite opposite behaviour

Unfortunately the plot of Fig1: the long stretch of SQZ was performed with too high gain in the CC loop this is why that part is more noisy

Figure 1. In the detuned SRCL configuration last night the alignment of SR and SDB1 was quite unstable, with SR moving by +/-0.3urad on time scales shorter than one hour, and SDB1 moving by +/-3 urad.

Figure 2. in the tuned SRCL configuration on March 18 the fluctuation of SR alignment compared to the ground were a bit smaller at +/-0.2 urad, and especially for SDB1 they were a lot smaller at +/-1 urad. 

We had seen during the shift this week that SR and SDB1 alignment changes during the detuning. And we would make sense for the SR alignment signal to be affected as it looks at the demodulation of the phase of the 491.3Hz line at frequencies of the SR alignment dithers.

Figure 3, 4 and 5 show analytical models of the optical response for a detuning of respectively 1nm, 10nm and 20nm. For a small detuning the phase at 491.3Hz decreases with increased losses, and the SR alignment signal tries to maximize the phase, so reduce the losses due to misalignment. But for a detuning of 10nm the phase at 491.3Hz stop depeding on losses, and for 20nm detuning it reverses direction, higher losses increase the phase. That would explain the SR behavior, the current alignment signal looses SNR with larger detuning, and eventually for higher detuning it will flip signs.

This would also explain why the squeezing phase scans last night had a lower effect for larger detuning. If SR alignment becomes erratic, there will be more losses in the SRC, and squeezing will become less efficient.

This should be tested and if confirmed there can be several solution. Use a line at DARM line at lower frequencies for the SR alignment, or use a line at higher frequencies with an opposite sign once the detuning is large enough, or think about possible other error signals for SR alignment in the detuned conifguration.

/users/mwas/detchar/toySensitivity_20260225/toySR_TF.m

ITF found in LOW_NOISE_3_ALIGNED and COMMISSIOING Mode.
All times are UTC.
19:55:03 ITF Unlock;
20:40 Unsuccesfull locking attempt at LOCKING_CARM_NULL_1F;
20:49 Unsuccesfull locking attempt at LOCKING_DARM_IR;
21:25 ITF in LOW_NOISE_3_ALIGNED;

ITF Left in LOW_NOISE_3_ALIGNED State and COMMISSIOINING Mode.
Activity concluded at 21:26;

The interferometer has unlocked with SRCL INPUT running away from -20 down past -30, and also with SR and SDB1 alignment moving away. I have reduced the optical spring set point from 12 to 11, so that is further away from a potential instability.

Today we aimed to inject again squeezing in the interferometer, while working in the SRC-detuned configuration.

At first we tested the lock acquisition with the SRC already detuned from the beginning: usually we get to CARM Null with a non-zero optical spring, then the SRCL_SET servo brings it and keeps it to a setpoint, which is 0.5 (roughly in Hz). Today we immediately put a setpoint of 10, meaning that the servo's job was to just keep the optical spring in place.

This happened at 12h53m34 UTC, then we unlocked at 13h07m54 UTC as we were testing this in anticipation during other activities.

We repeated the procedure with the lock in CARM Null starting at 14h11m23 UTC, with the servo setpoint at 10 starting at 14h12m48 UTC.

Later, while still in CARM Null, we made a noise injection on DARM (to be analyzed later) in order to understand the DARM plant and check the optical spring: DARM_noise shape, amplitude = 20, 180 seconds starting from 14h39m00 UTC.

Then we continued the lock acquisition while keeping the detuned configuration, and that worked with no issues.

We reached LOW_NOISE_3_ALIGNED at 14h59m44 UTC.  At 15h04m05 UTC we lowered the SRCL_SET setpoint down from 10 to 9.

Then we made another noise injection on DARM with the same parameters of the standard injection for Science mode: DARM noise shape, amplitude = 0.2, 180 seconds starting from 15h11m30 UTC.

At around ~15:20 UTC we started the squeezing injection using the SQZ_MAIN Metatron node, going to the SQZ_INJECTED_NO_FC state.

At 15:35 UTC we changed the squeezing phase from 3.8 to 4.1;. 4.1 was the value used yesterday.
At 15:37 UTC we changed the squeezing phase to 1.1
At 15:40 UTC we started a 1000s long scan over -7 radians of the squeezing phase, with a manual ramp. We then understood that the automatic alignment is not supposed to be enabled all the time, but only when anti-squeezing is injected.
At 15:56 UTC we set the phase to 3.1 rad for anti-squeezing, and we let the automatic alignment run for 10 minutes.

We went back to the SQZ_LOCKED_NO_FC state and run the /virgoDev/AEI_SQZ/coh_scan.py script, to make the scan in the usual way. Initial GPS time: 1458058066.74 (16h07m28 UTC).

Then we increased the detuning of the SRC by progressively increasing the SRCL_SET setpoint, targeting a pole of ~200 Hz (read by Hrec) as yesterday:

• 16h37m35 UTC SRCL_SET_SET = 11;
• 16h50m34 UTC SRCL_SET_SET = 12;
• 16h56m11 UTC SRCL_SET_SET = 14;
• 17h06m20 UTC SRCL_SET_SET = 13.

The small rollback was due to the pole reading starting to drop to 170 Hz, so we got back to around 200 Hz.

At 17h33m09 UTC  we started another scan of the squeezing phase. Initial GPS time: 1458063207.03.

None of the two scans reached the same increase in sensitivity as yesterday's first one, but additionally the second one showed a lower effect overall. Additionally, something else was changing in the interferometer at the time, as the frequency noise and the PSTAB couplings we higher. More analysis on that will follow.

WIth the same conditions as the second scan, we performed a calibration measurement using the CALI node, state CALIBRATED_DF_DAILY; the procedure started at 18h08m11 UTC.

We decided to keep the detuned configuration for the current lock and for the weekend, using the loop and the automation: we set the SRCL_SET_SET to 10 in CARM Null (line 274 of ITF_LOCK.ini) and to 12 in LOW_NOISE_2 (line 522 of ITF_LOCK.ini), which are the only places where it is set in the automation. I added the reset to zero of the setpoint in the DOWN state of the DRMI_LOCK node.

The SRCL_SET setpoint was also put manually to 12 at 19h13m46 UTC. In the figure a snapshot of the figures of merit of the interferometer.

18/03/2026 

We started in low noise3 alignment with a very low magnitude. This was due to the fact that the day before we aligned the SQZ with the SR in a poisition in wich we had a very low 4 MHz magnitude

We managed to improve the 4MHz magnitude by using SQB1_MIRROR 21-22-23-24 (ie two aa actuators and two picomotors) The maximum magnitude that we obtained was about 2.3-2.4mV.

We decided to run a phase scan of the CC loop that started at 12:20 UTC (see fig1)

In the afternoon we decided to go in SR single bounce.

Procedure:

• ITF_LOCK in SR Single Bounce
• SQZ_MAIN in SQZ_LOCKED_NO_FC and then Paused
• SC_PLL unlocked both FAST and SLOW 
• Set DET_MAIN in Shutter OPEN
• SDB1 LC in High bandwith and fixed set point
• SR position the same of last lock in low noise 3
• SDB1 LC position the same of last lock in low noise 3
• B1 PD1 and B1 PD 2 rearmed and open by hand (SDB2_Dbox_bench
• SQZ fast shutter Open
• SC Shutter Open

We started to ramp SC PLL. We did a ramp between 35400 and 35100 steps of the pelter in 5 seconds and we looked on B1 PD1 the height of the peaks. We started somewhere around 3V but we noticed a strong clipping. Probably in the beam reflected by SR and entering back into FI. We started to align but having the actuators saturated we could not use the autoalignment of SC into OMC. We tried to lock the OMC to the SC beaam but it was too noisi probably due to the beam clipping

Using the picomotors we managed to arrive to an alignment of about 4V but still clipped. Looking the modes we had 8% of TEM00 and 4% of mismatch.

We went in low noise 3 alignmend and we run another phase scan at 20:15. (Figure 2)

19/03/2026

We started the activity around 10:30 UTC

Given the high wind the interferometer did not lock in low noise 3. So we decided to start again in SR single bounce.

Before the unlock Michal tested again the SR detuning. So we decided to not align in this configuration but in the tuned one. Thus we have chosen the SDB1_LC_TX and SDB1_LC_TY and SR_TX and SR_TY set point of about 30 before the unlock.

Once we arrived in SR single bounce we had a similar alignment situation of the day before, i.e. a SC transmission to OMC of about 4V.

We started with quite all the DOFs of the AA saturated so we started in scan mode by discharging the worse one, i.e. SQZ_MIRROR_2_Y that was at +9V.

SQZ_MIRROR2_Y is SQB1_M31_Y that is in front of to M32_Y. So to discharge it we mainly used SQB1_M32 to do it. After that we discharged also SQZ_MIRROR_1_X that is SQB1_M33_MT1. We discharged with the pico SQB1_M33_H and V but was quite difficult.

When the actuators were below 5V in absolute value we locked the OMC to the SC beam by following the procedure in this wiki pagehttps://wiki.virgo-gw.eu/Commissioning/Detection/OutputModeCleaner?validation_key=383c00b313bb289236e6666bd7b3ab8e#OMC1_manual_locking_on_direct_or_squeezer_beam. We only chosen different lock and unlock thresholds i.e. 0.8V lock threshold and 0.2V as unlock threshold. The lock works very well only when the PZT loop engages, but it happens only when the system is very quiet.

With this configuration we run the autoalignment and we tried to continue to discharge the actuators. We iterated a couple of time between lock mode and scan mode (when the actuators saturated). 

At 12:25 UTC just after lunch we took the first measurement (fig3) and we obtained:

• TEM00 4.44mW ->92.6%
• TEM10 0.115 mW -> 2.4%
• TEM20 0.238mW -> 5%
• Total power 4.8mW

We continue to try to discharge the actuators, then we did a step of SQB1 by 50um in vertical, i.e. from -600um to -650 um

Figxx shows another check in scan mode at 13:26:44:UTC Fig4

• TEM00 4.49 mW ->91.0%
• TEM10 0.21 mW -> 4.5%
• TEM20 0.21mW -> 4.5%
• Total power 4.91mW

Alignment worsened but more total power.

We engaged again the AA with the OMC locked on the SC and we continued in moving SQB1_LC_Y in negative direction up -800um

You can see from the long trend figure5 that this was the most effective DOF to unclip the beam. At 14.18:22 we performed the last SC scan before going in carm null 1f and we obtained (fig6)

• TEM00 5.15 mW ->94.0%
• TEM10 0.1 mW -> 1.9%
• TEM20 0.225mW -> 4.1%
• Total power 5.48mW

Fig 7 shows some screenshot of the cameras with OMC locked on SC and in reflecton from OMC we are dominated by mismatch mode

We decided to stop and start to see in higher state of the interferometer. Respect to the day before the TEM00 power is improved by 5.15/4=1.3 times. We expect that the B1_4MHz mag will improve by a factor sqrt(1.3)=1.14. Considering that the maximum of the magnitude the day before was 2.5mV we expect something around 2.8mV.

Before going in LOW_NOISE_3 aligned we stopped the lock acquisition at CARM_NULL_1f. And we performed again a scan of the SC with the OMC and looked the modes in transmission on B1_PD. Note that when we arrive inCARM_NULL_1F the OMC is parked near the 56MHz TEM00 at a temperature of 22.556. We moved the temperature of the OMC until the camera of SDB2_B1_DC was quite dark (we also increased the integration time) we had to move in total by -0.01 deg ie to 22.546. At this temperature we performed a scan of the OMC after opening the B1_PD1 and PD2 PDs by hand. Fig8 at 15.25:03 UTC. We onbtained

• TEM00 8.42 mW ->92.5%
• TEM10 0.2 mW -> 2.2%
• TEM20 0.47mW -> 5.2%

So the MM is a bit worse with the hot interferometer but we confirmed that SR single bounce is a good working state.

So we decided to lock the interferometer in LOW_NOISE_3 aligned and we arrived at about 15.45UTC. The magnitude of B1_PD1_4MHz was about 2.5mV before running the AA. With the AA engaged we managed to reach 2.8mV as expected (Fig9)

After that we took some stretches of SQZ, ASQZ, and Shot then we run a phase scan of the CC loop Fig10:

• SQZ phi 4.1rad at 16UTC (too high CC gain. At the end of the stretch we passed to 50000 instead 75000)
• ASQZ: phi = 3.1 rad at 16:28 -> 16:40
• Shot 16:41 UTC
• Phase Scan at 16:50 UTC up to 17:24 UTC Fig 11

After that we started the detuning of SRC. We left the SQZ shutter open to monitor the behavior of the magnitude and we saw that it passed from 2.8mV to 0.18mV (fig12). N.B we reached 0.2mV acting again on picomotors.

After that we tried to go in SQZ mode and we reached 2 times more than 55Mpc. See fig12 to do. Between the 1 and 2 stretch of SQZ we tried to align with SQB1 and PICO Motors

Then we wanted to start a phase scan but the ITF unlocked

We checked so if the decrease of the mag was really a clipping/misalignement.

We went again in SR single bounce but with SR and SDB1 angular position as in LN3 with SR Detuned.

AA Mirror position:

2026-03-19 19h11m48 UTC    vardaro 'AA Mirror1:X ramp [Frequency=0.5,ampl=0,bias=3.8]' sent to SQZ_PLL_AA
2026-03-19 19h12m02 UTC    vardaro 'AA Mirror1:Y ramp [Frequency=0.5,ampl=0,bias=-0.55]' sent to SQZ_PLL_AA
2026-03-19 19h12m09 UTC    vardaro 'AA Mirror2:X ramp [Frequency=0.5,ampl=0,bias=6.92]' sent to SQZ_PLL_AA
2026-03-19 19h12m18 UTC    vardaro 'AA Mirror2:Y ramp [Frequency=0.5,ampl=0,bias=6.26]' sent to SQZ_PLL_AA

Fig14 to do is with the position of the mirrors and picomotors as during the last lock 19:1

2026-03-19-18h59m18-UTC    info vardaro      'AA Mirror1:X ramp [Frequency=0.5,ampl=0,bias=4.825]' sent to SQZ_PLL_AA
2026-03-19-18h59m31-UTC    info vardaro      'AA Mirror1:Y ramp [Frequency=0.5,ampl=0,bias=-7.6]' sent to SQZ_PLL_AA
2026-03-19-18h59m46-UTC    info vardaro      'AA Mirror2:X ramp [Frequency=0.5,ampl=0,bias=4.72]' sent to SQZ_PLL_AA
2026-03-19-18h59m58-UTC    info vardaro      'AA Mirror2:Y ramp [Frequency=0.5,ampl=0,bias=7.15]' sent to SQZ_PLL_AA

Fig13 to do is with the mosition of the AA mirrors as with the SR detuned but with the picomotrs steps reverted (19:13:10UTC)

Then we moved back SR and SDB1 angular position to the same position in wich the SR was tuned and we saw that our alignment is reproducible. See fig 14 taken at 19:09 UTC

We closed the SC and SQZ Fast shutters and Fabio relocked the interferometer.

Summary: the SR detuning configuration is promising, but we have to re-do the measurement with different alignment. We saw that we clip easily with the beam that comes back from SR to SDB1 FI. Don’t know if is the diafragram or just the FI small crystal. The mismatch of SQZ to OMC is 4% with cold interferometer and 5% with hot interferometer. SR single bounce is a good position for Align SC into OMC.

ITF found in down with WEST cavity misaligned. After a manual realignment of the cavity the ITF reached LN3_Aligned at 7:44 UTC, it unlocked at 7:53 UTC.

It relocked at 8:39 UTC.

The first part of the shift was spent to fix a problem on HREC see NCals frequency change.

From 11:00 UTC to 12:00 Irene, Maria amd Vincenzo went in CB to perform noise injection in LN3.

At 12:39 UTC Valerio changed the frequncy modulation of WI accelerometers.

At 14:45 UTC the ITF was relocked in CARM_NULL_1F to tune the optical spring set point.

ITF left locking in LN3_Aligned.

The 66 Hz was a bad choice since Hrec uses this frequency band to estimate the SNR of several lines, even at 71.5 Hz, instead of the adjacent bins as in the past. With the wrong line SNR, Hrec was not recomputing the optical gains and producing a wrong h(t) and BNS range.

It took us a while to remember/discover this feature.

All NCal lines have been moved to near 36 Hz, restoring the proper behavior of Hrec.