We started the shift opening the SDB1 shutter at 8:42UTC with +41000 steps

In the mean time we also checked the status of the squeezer with the HD detector: +8.7 dB of ASQZ and 6dB of SQZ at about 9:10 UTC

After that we opened SQB1 shutter with -500000 steps at 9:48 UTC we struggled to see any beam from ITF

We then understood that the SQB1 AA mirrors were not in the position of last time when we injected SQZ. I.e. 2nd september 2025. We then put them in their last position and this allowed us to see a beam into the SQB1 cameras

2026-02-20-10h24m00-UTC    info vardaro      'AA Mirror1:X ramp [Frequency=0.5,ampl=0,bias=5.28]' sent to SQZ_PLL_AA
2026-02-20-10h24m33-UTC    info vardaro      'AA Mirror1:Y ramp [Frequency=0.5,ampl=0,bias=-1.795]' sent to SQZ_PLL_AA
2026-02-20-10h25m20-UTC    info vardaro      'AA Mirror2:X ramp [Frequency=0.5,ampl=0,bias=0.606]' sent to SQZ_PLL_AA
2026-02-20-10h25m40-UTC    info vardaro      'AA Mirror2:Y ramp [Frequency=0.5,ampl=0,bias=3.22]' sent to SQZ_PLL_AA

Once the ITF was locked in LN3 aligned we started the realigment of the squeezer. At 10:40UTC we started the alignment with a magnitude of about 0.1mV or less. The maximum found in 2024 was about 2 mV.

We started to check the CC offset but this was ok

We inserted some ramp on SQB1 mirrors but this was not useful

We then decided to engage the SQZ AA system and we left it drifting. With that we managed to arrive at about 0.8 mV

We started to move SQB1 X and TX direction. We improved a bit going with SQB1 in TX from 300 to 120 urad

We arrived with a B1 4MHz magnitude of about 1.7mV and we injected SQZ into the ITF during lunch break at 12:42:30 UTC. Probably a bit less than one dB was visible in the ITF (fig 1)

Then we started to align with AA and pico motors on SQB1 in order to improve the magnitude. And after some iteration between SQB1 bench, SQZ AA actuators and SQB1 picomotors we mamaged to have a magnitude of about 2.6mV

But with the SQB1 AA mirrors near saturation SQB1 M1 X at +7V and SQB1 M1 Y at -8V, whereas M2 X was at -5.5V and M2 Y of about 0.5 V. Then the interferometer unlocked,

ITF relocked at 16 UTC, we waited up to 16:13 to open the shutter then we restarted with the alignment in order to discharge the AA actuators, We tried in closed loop, then the interferometer immediately unlocked.

ITF relocked around 17UTC but it ulocked quite immediately after the restart of the alignment

When it relocked the magnitude was very good and the SQB1 actuators far from the saturation. So we put in SQZ from 17:44:27 UTC. Fig2

Shot noise reference at 18:02:00 UTC

Phase scan run at 18:09 UTC Gain 50000 Shot noise reference 2 mins. (fig3)

The phase scan was too short to see the minimum of the magnitude. So we run onother one at 19:17 UTC between 80 and 290 deg (fig4)

After the scan we took 10 min of SQZ at 19:48:24 UTC with an angle of 4.2 rad 

We stopped at 20:02 UTC

Final position of the various DOFs:

• SQB1 M1 X 5.8V
• SQB1 M1 Y -4.99V
• SQB1 M2 X -4.94 V
• SQB1 M2 Y 1.78V
• SQB1 TX 0urad
• SQB1 TY -140urad
• SQB1 TZ 125 urad
• SQB1 X -3250 um
• SQB1 Y -550 um
• SQB1 Z 710 um 

At the end we closed only SQB1 shutter, we switched off all the dither lines and we rotated HWP2 of SQB1 to send SQZ toward the HD detector

ITF found unlocked with the north cavity unlocked; after a manual aligment of the cavity and then a manual prealignment of SR/PR I launched the lock acquisition.

We collected many unlocks; at 10:36 UTC we finally reached LN3_Aligned and the Squeezing team started working from remote.

Under request of Jean Pierre at around 8:30 UTC I  switched ON/OFF the light on DET base tower.

From  14:30 UTC to 14:50 UTC, taking advantage of an unlock, Valerio performed a check of the IB accelerometers.

Relock in progress; the Squeezing activity will resume once locked in LN3_Aligned.

As happens quite often, this morning the lock acquisition failed many times at the beginning of CARM NULL (fig 1). The typical behaviour is a large fluctuation of the dark fringe, sometime becoming too large. WI and WE optical levers show some unusual rotation in TX, but apparently too slow for being able to have an impact on DIFFp accuracy (fig 2). Moreove, this behavior of the optical levers occurs only in this particular moment of the lock acquisition; normally, when the lock is acquired and lasts hours, no SOFT loop malfunction or sensing glitch is visible.

I think the problem comes from DIFFp TX sensing. During the crisis, the accuracy of the in-loop signal looks good enough and cannot explain the fluctuation of the datk fringe, but in the quadrature (and also in the other B1p QD) a large noise is visible (fig 3).

Using the optical levers to reconstruct DIFFp TX, one can notice an oscillation well correlated to B1p fluctuation (fig 4). The amplitude is about 0.1 urad peak to peak, at the limit of OptLev sensitivity, but very large for DIFFp, and much larger that DIFFp_TX in loop. Sensing noise of this loop seems a good explanation of that behaviour.

This night the unlock occurred because of an increase of noise seen in SSFS signals, which at the end saturated the loop correction (fig 1). The problem started as a gradual increase of broadband noise, visible also in PSL_PMC_TRA_DC (fig 2, fig 3). When the noise was larger, just before the unlock, it appeared in SSFS_Corr as a series of fast glitches (fig 4).

Tonight, after each unlock, ITF_CONDITIONS got stucked in NOT_LOCKED, without transitioning again to LOCKING, therefore blocking ITF_LOCK in DOWN.

Tomorrow we will investigate this unexpected behaviour and the link to what was done earlier today. For tonight, I disabled the autorelock failsafe and let the ITF start locking again.

In order to align the metatron instance running in Cascina we have removed the dev version of metatron (/virgoDev/metatron/v0r3p1) used to run ITF_CONDITIONS node.

We have then promoted the conda env updated to Fd 8.61.1 (/virgoDev/mamba/env/automation) into production under /virgo/conda/envs and then restarted the same version of metatron (/virgoApp/metatron/v0r3p0) for all automation nodes on top of the updated conda environment.

We first tested only the ITF_CONDITIONS and ITF_LOCK nodes, in order to verify that all channels supposed to go in the DAQ actually do, as this was the showstopper reported in v68573. After verifying this, we stopped and restarted all the nodes with the new underlying conda environment, without changing the version of metatron itself.

We tested a full lock acquisition up to LOW_NOISE_3_ALIGNED, with no issues.

A side issue we found before starting the activity is the presence of a few duplicated channels, only on the online data and not the stored one. These are the channels which are pushed to the DAQ by ITF_CONDITIONS (DQ_META_AUTORELOCK_*, DQ_META_ITF_LOCK_REQUEST and DQ_META_NOMINAL_LOCK). As they don't appear duplicated in the stored data, we don't know if this behaviour started today after a couple of tests with the previous metatron configuration, or it was already the case since some time. We'll try to figure this out offline.

ITF found relocking up to LN3_ALIGNED in COMMISSIONING mode and the planned "Update of Metatron conda environment (Bersanetti, Carbognani)" activity in progress...
Activity went on without particulary issues until 17:30 UTC. ITF stable locked at LN3_ALIGNED.

PREPARE_SCIENCE Mode and AUTORELOCK_FAILSAFE set. 

The side issue has been fixed by restarting the Fd dataDisplay providers:  Fbm{Det, Main, Moni, Susp}Users, FbmPyAccess and FbmStDy{,Bis} servers.

The problem seems related to the channel list built by these servers and used by the DataDisplay.

As already reported, there weren't duplicated channels in the DAQ streams

ITF found in LN3_Aligned.

At 8:59 UTC the ITF unlocked; as requested by the pending action I reloaded the configuration for SDB1_Quadrants.

At 9:30 UTC Matthieu and Camilla started working at the planned activity of ML cristal temperature investigation; activity concluded around 14:30 UTC.

After that Diego and Franco started working at the update of Metatron Conda environment.

In order to find a good crystal temperature setpoint, working both for the master laser and the squeezing laser, we investigated temperatures from the current value, which is 23.7C, up to 31.4C.

The mode hops appears as an additional error signal while scanning the SL piezo.

In conclusion, a temperature around 29.7 C seems the best candidate since there are no spurious peaks and it is in the middle betwwen two mode hops. Also 27.3 C has a good shape but it is closer to the 28.1 C mode hop.
 

ITF Found in COMMISSIONING Mode and SINGLE_BOUNCE_NI State with Gouaty and Romero working on OMC Mode Matching activity (see #68707 and #68715).
All times are UTC.
17:49 OMC Mode Matching activity concluded (see #68707 and #68715).
17:53 ITF set DOWN State and then requested CARM_NULL_1F after added +30 urad to PR TY (see #68709). State reached at 18:48 after usual cross alignment.
18:50:00 ITF UNLOCK from CARM_NULL_1F.
19:01:19 ITF UNLOCK from LOCKING_CARM_NULL_1F.
19:38 ITF in LOW_NOISE_3_ALIGNED.
22:00 ITF left in COMMISSIONING Mode with AUTORELOCK_FAILSAFE On and LOW_NOISE_3_ALIGNED State.

DET
(18-02-2026 20:30 - ) SDB1_Quadrants configuration file to be reloaded when ITF will be DOWN (entry #68715).

Looking at the data after the ITF was relocked in Low Noise 3 aligned (attached figure), we can see that there is an offset between the B5 QD2 signal (in particular in H) and the SDB1 bench aligned position with the dither control. I already added an horizontal offset of -170 um in the SDB1_Quadrants configuration file, but the configuration needs to be reloaded to apply it. To be done when the interferometer will be DOWN.

The goal of the shift was to adjust the OMC mode matching. First we worked in Low Noise 3 aligned in order to find the position of the SDB1 bench that would optimize the mode matching. Then we worked in single bounce in order to reset the nominal position of the SDB1 bench that we had during O4, while adjusting the position of the meniscus lens to preserve the mode matching tuning.

When the shift started the ITF was still trying to lock. State Low Noise 3 aligned reached at 8h14 utc.

The initial position of SDB1 is SDB1_LC_Z = 1204 um and Sa_OB_F0_X = -400.

At 8h36 utc we start to scan the SDB1 bench going from F0_X = -400 to -1400 um, in 1000 s.

At 9h07 utc we start to scan from F0_X = -1400 to -2400 um in 1000 s.

At 9h28 utc we start to scan from F0_X = -2400 to 600 in 3000 s.

At 10h30 utc we start to scan from F0_X =600 to 1600 in 1000 s.

At 10h54 utc we start to scan from F0_X = 1600 to 2600 in 1000 s.

On Fig.1 and Fig.2 we see that the optimum of optical gain was reached around 10h00 utc when the bench position was Sa_OB_F0_X = -500 um.

We then unlocked the interferometer manually and started to work with the NI single bounce.

We displace the SDB1 bench to Sa_OB_F0_X = -500 um, where we found the best mode matching in LN3 aligned.

Then we adjusted manually the TX angular position of the bench to TX=225 urad  in order to center the B1p beam on the camera. Then we lowered the threshold on B5_QD2_sum in order to be able to engage the drift control.

We lock the OMC at 12h18 utc.

We start the OMC scan at 12h22m40 utc (See Fig.3). We measured 15 uW on the TEM00, 1.6 uW on the first order mode (10% of misalignment) and 0.8 uW on the second order mode (5% of mode mismatch). Given the relatively large misalignment we decide to relock the OMC and improve the alignment.

OMC locked again at 12h37 utc (see Fig.4). We tried to maximize the power on the TEM00 by adding offsets in the B5 QD2 H (-200) and V (+100) signals.

New OMC scan at 12h48 utc (Fig.5). We measure 16 uW on the TEM00, 0.65 uW on the first order mode (4% of misalignment), and 0.68 uW on the second order mode (4.2% of mode mismatch).

We restore the SDB1 bench position that we had during the O4 run (F0_X = 2600 um).

Since the optimal position of SDB1 that maximizes the mode matching in LN3 is -500 um, this means that the meniscus lens should be displaced by 3100 um. From a measurement performed in Feb 2024 ( https://logbook.virgo-gw.eu/virgo/?r=63269 ), we estimated that 4 mm correspond to about 113500 steps. Therefore we should displace the meniscus lens by about 88000 steps, and we should perform positive steps which are equivalent to a negative displacement of the bench.

After displacing the meniscus lens by only 5000 steps and upon Michal's suggestion, we locked the OMC again at 14h21m40 utc and adjusted finaly the alignment with the B5 QD2 offsets. Then we performed a new OMC scan at 14h32 utc (see Fig.6): we measured a TEM00 of 17 uW, a first order mode of 0.9 uW (5.3% of misalignment) and a second order mode of 0.26 uW (mode mismatch of 1.5%).

We displaced the meniscus lens by a total of +88000 steps (by +2500 steps at a time). In order to keep the bench balanced we also moved the BENCH_TX counter-weight by +117500 steps.

We then lock the OMC at 15h24m40 utc. We try to adjust the B5 QD2 offset to optimize the OMC alignment (H -150 and V +50).

We start a new OMC scan at 15h37m30 utc (Fig.7): we find 16.5 uW on the TEM00, 0.75 uW on the first order mode (4.5% of misalignment), and 0.55 uW on the second order mode (3.3%) of mode mismatch. As the mode mismatch is still a bit smaller than the one we found when the bench was at F0_X = -500 um, we decide to move the meniscus lens a bit further.

We displace the meniscus lens by another +15000 steps (by +2500 steps at a time), compensated with +17500 steps on BENCH_TX.

We relock the OMC at 16h07 utc. We try to adjust the OMC alignment with B5_QD2 offsets (H:-175, V:+100).

Started a new OMC scan at 16h18m53 utc (Fig.8): we find 16.5 uW on the TEM00, 0.86 uW on the first order mode (5.2% of misalignment), and 0.6 uW on the second order mode (3.6% of mode mismatch).

We displace the meniscus lens by another +17500 steps (by +2500 steps at a time), compensated with +22500 steps on BENCH_TX.

We relock the OMC at 16h44m30 utc. We try to adjust the OMC alignment with B5_QD2 offsets (H:-200, V:+100).

Started a new OMC scan at 16h58m43 utc (Fig.9): we find 16.4 uW on the TEM00, 0.83 uW on the first order mode (5% of misalignment), and 0.8 uW on the second order mode (4.9% of mode mismatch). The mode mismatch is now a bit higher than what we were aiming at.

We undo -8750 steps on the meniscus lens, compensated by -10000 steps on BENCH_TX.

We perform another OMC scan at 17h18m09 utc (Fig.10): we find 16.3 uW on the TEM00, 0.83 uW on the first order mode (5% of misalignment), and 0.74 uW on the second order mode (4.5% of mode mismatch).

We consider that the obtained mode mismatch of 4.5% is sufficiently closed from our target (4.2% with some error bar). Therefore we stop the tuning here. In total we performed 111750 steps with the meniscus lens, and 147500 steps with the motorized counter-weight.

To be noted that we have changed the threshold on B5_QD2_sum in order to be able to engage the B5 drift control on the single bounce beam, as shown below :

ACL_RELAY_CH    B5_QD2_safe    ""    LC_LOOP_FREQ    20    0.0    -2    "B5_QD2_sum"    1.6    1.4    ">="    # 2.5 1.8

Some of the unlocks we've been having in this last week (Figs. 3 and 4 in v68695) are preceded by a large deviation of MC_MIR_Z_CORR, but especially on PR_X (and _Y), which are the signals of the input bench alignment in the locked configuration (Figures 1 and 2). As a result, the bench is misaligned and sometimes the North arm cannot lock again. The West somehow manages, though, maybe because the BS is fast enough to compensate enough that the lock is possible afterwards.

Looking at more signals, it shows up that this behaviour is caused by glitches at the F0 level of IB, that can be seen in all horizontal accelerometers (Figure 3). More investigation will be focused on understanding the underlying cause and the mitigation.

ITF found unlocked: west cavity locked and north cavity unlocked. To recover the lock I had to move the PR top stage. After that the ITF was locked in LN2 and then in LN3_Aligned at 8:13 UTC.

Once locked Romain and Alba started working at the OMC mode matching.

At 11:23 UTC the ITF was manually unlocked to continue the activity in NI single bounce with PR TY further misaligned of -30ur.

Activity still in progress at the end of the shift.

Yestarday as an exploratory study, we have performed some very rough measurements of the magnetic field generated by the hidden permanent magnets present in a Varian Noble Diode ion pump located in CEB. A similar measurement was also performed on a vacuum gauge.

Even if not accurate at all, these measurements done with a Galaxy A13 phone and the Pyhphox app have shown a magnetic field as large as ~2 mT close to the pump surface.

In order to extract detailed information about the magnetisation amplitude and direction to further simulate the field created by these devices at the location of the coil magnet actuators (mirror and marionnette), further studies are needed with dedicated magnetic field sensors. These are important since these ion pumps and vacuum gauges are possibly vibrating with the surrounding environment.

There was an error in the last two plots (RIN vs DCP and RIN vs OG), as thermal noise subtraction was not done for the whole data set. In attachment the correct plots.

The relevent code snippit, extracted from the previous elog, to produce the correct RoC in Finesse, in a hot interferometer, is this one:

# create a new Virgo model
virgo = finesse.virgo.Virgo("02_tf_sideband.kat")

# %%
model = virgo.model

def P_RH(I1, V1, I2, V2):
    return (I1*V1 + I2*V2)

# RH values taken at 1453136567 (17h 22nd Jan 2026 UTC)
P_RH_NE = P_RH(I1=0.268,V1=16.0627, I2=0.27125, V2=16.03)
P_RH_WE = P_RH(I1=0.2903,V1=17.4024, I2=0.2954, V2=17.4)

print(f"NE: {P_RH_NE} W, WE: {P_RH_WE} W")

LMA_measurements = dict(
    NI=1424.6,
    NE=1692.6,
    WI=1425.5,
    WE=1695
) # taken from slide 8, "Cold IFO", VIR-0907A-25

Pifo = model.i1.P

for k,v in LMA_measurements.items():
    v = v-4.9 # systematic error at LMA
    v = v-2.9 # effect of cryo trap
    v = v + 0.27*Pifo # VIR-0977A-25, Effect of YAG beams
    if k[1] == "I":
        v=-v #input mirrors are negative RoC in Finesse
    else:
        if k[0] == "N": # effect of the RH on the end mirrors
            v = v-2.92*P_RH_NE
        else:
            v = v -2.41*P_RH_WE
    getattr(model, k).Rc = v

# Also from VIR-0977A-25
model.NIsublens.f = 1/(Pifo*2.02e-6)
model.WIsublens.f = 1/(Pifo*2.02e-6)
model.beam_trace()

print(f"""
    The expected second order mode freq is:
    North: {2*model.cavN.mode_separation_x:.0f} Hz
    West: {2*model.cavW.mode_separation_x:.0f} Hz
""")

During yesterday maintenance we removed the three microphones and associated ADC channels. The Microphone GRAS 46AZ ns 601362 from Milanio.Bicocca and BK 4193 SN 2927327 from APC are now in Irene's office.

At the beginning of the shift the ITF was relocking, it went back to LN2 at 14:57 UTC
SRCL detuning activity ongoing
16:29 UTC unlock, relock in LN2 at first attempt
17:18 UTC end of SRCL detuning activity with ITF in LN2, ITF set to LN3_ALIGNED
18:04 UTC unlock to test the lock acquisition after the changes as in entry #68706 (Bersanetti)
18:39 UTC ITF in LN3_ALIGNED at first attempt with the changes of entry 68706
22:00 UTC end of shift, ITF left in LN3_ALIGNED with AUTORELOCK_FAILSAFE
 

At the end of the day's activities, I made two changes to the lock acquisition, to be tested during the last part of the shift:

• I reinstated the BS_TX angular loop in full bandwidth in LOW_NOISE_2 (instead of LN3); this is a recurring thing, but lately we managed to close the loop manually soon enough in LN2, so it is worth checking if it can be automated (again); for reference/rollback, I uncommented line 5966 and commented lines 6505 -> 6507 of ITF_LOCK.py;
• I increased both gains (standard filter and boost filter) of the SSFS loop in CARM_NULL_1F: the loop UGF was quite low yesterday when I checked, and I tried the boost gain of 800 to improve it (apparently, it has been tested already many times). So I set this value in the ITF_LOCK.ini file, and changed of the same amount the gain of the initial loop; the old values are left, commented, just above the new ones (lines 227 -> 230 of ITF_LOCK.ini).

The issue could be related to the fact that one of the relays of the NE actuator (REL21) was not correctly set to the LowNoise2 mode in the second lock. If the GUI shows the relays in the standard ordering, this should mean that the DR coil was affected, and this confirms the checks done offline on the Hrec reconstruction and the NE TF monitoring in particular.

After the maintenance we set the ITF in LN2 and allowed about 10 minutes for the state to stabilize, before proceeding with the exploration of the stability range of SRCL_SET.

We closed BS_TX in full bandwidth and opened the SRCL_SET servo, then started changing SRCL_SET in steps of 10, one minute per ramp, allowing 20 minutes for each new state to stabilize and gather data. We managed to make a three points scan:

the ITF unlocked a few minutes after the last change of SRCL_SET (Fig.1).

We relocked the interferometer and decided to explore the margin of the scan more granularly. After reaching LN2 we set BS_TX in fbw and disabled the SRCL_SET servo again, then we set SRCL_SET to -20 with a two minutes ramp, allowed about 15 minutes to stabilize. We started changing the SRCL_SET in steps of 2, with one minutes long ramps, but realized after the first step that very little changes with such a small step while the duration of the test becomes considerable. For the next steps, we changed SRCL_SET in by 4, with two minutes long ramps.

The ITF unlocked about 10 minutes after step 4 (Fig.2)

The sensitivity was unexpectdly high for LN2, but the correct reconstruction of Hrec needs to be checked.

ITF found unlocked with WEST cavity misaligned; many red flags for ISYS.

I was able lo realign the WEST cavity but once locked it apperead a strong oscillation in the powers of the arms.

At 7:00 UTC started the planned Maintenance, below the listof the activities communicated in control room:

• check of TCS thermo camera reference (operator);
• check of TCS powers (Diana from remote);
• check on the ISYS, see: 68699;
• TCS chiller refill (Marco and Nicola);
• Remove microphones at NE, reconnect NNC arrays at NEB/WEB, measurements in CEB (Tringali, Fiori, Quemener);
• cleaning of the experimental areas;

At around 11:00 UTC I started to relock; after a manual prealignment of SR and PR the ITF reached LN3 at 12:16 UTC.

At 12:47 UTC it was manually unlocked to relock in LN2 as requested by the commissioning activity; activity still in progress at the end of the shift.

SUSP

At 7:30 UTC while testing the new Alignment interface I opened the SR LC loops; see SR LC tz saturating for wrong setpoint.

DET

The periodic task of OMC lock and scan in single bounce was skipped because of the problem of the SR LC that was fixed around 11:00 UTC.