ITF in Science mode for all the shift.

No DMS alerts during the shift.

Guard tours (time inTC)

21:00-21:40; 23:40-0:20; 1:40-2:20; 3:40-4:20

Today we worked on the recovery of the FIS source after the Virgos's main laser frequency change.

We have been focusing especially on finding the origin of the glitches that appeared in the green line (see figure PLL_ON).  To check if they are related to the new PLL setup we first turned off both the PLL loops (MAIN and CC), figure PLL_OFF.  Then with both the PLLs OFF  we moved the temperature of the main laser by -0.51 Celsius , figure PLL_OFF_LowerLasTemp.  These operations do not appear to have changed the glitch rate.  We then also lowered the temperature of the laser CC obtaining a similar result.

We also turned on/off the lasers noise eater,  we reduced the SHG threshold of about 10% and  and we did an inspection of DetLab.  Even in this case the actions were not decisive.   Further tests will therefore have to be done to understand the origin of these spikes.

Another problem we encountered is that the MAIN and CC beat notes, either with  the CC loop open or closed, have a comb structure that wasn't present during the last shift (# 67444). Also in this case further test are required.

ITF found locked at LN3 in SCIENCE mode.
The BNS Range is fluctuating since this night in a visibly abnormal way. TBC.
At 15:00 UTC I set COMMISSIONING mode and the planned SQZ activity started with deLaurentis in Det Lab. Activity concluded at 20:50 UTC. SCIENCE mode set.
The fluctuation of BNS range is now disappeared. (TBC).

Guard tour (UTC):
18:00 - 18:40

We updated the DMS threshold for B1p_DC to account for the darker fringe of the past few weeks.

Considering a mean value of 0.0084 mW for B1p, the new thresholds at LN3 are:

• yellow alert: 0.0076 < B1p_DC < 0.0092 (+-10%)
• red alert: 0.0067 < B1p_DC < 0.0101 (+-20%)

LockMoni has been successfully stopped and started, with the ITF in Adjusting. We are now back in science mode.

We recently observed a periodic seismic noise in the CEB area affecting nearly all F0 suspension accelerometers.

Although this noise is effectively filtered by the SA chain and seems not visible in Hrec, its roughly three-hour cycle and characteristic double-pulse structure (two strikes separated by about 12 minutes, with minor timing variations) make it worth investigating.

Initial mapping with Newtonian sensors pointed to the northeast sector of CEB, and subsequent analysis narrowed the source region to the FCIM and FCEM zones.

Remarkably, each cycle unfolds as four pulses: an initial event in FCIM (or broadly the CEB area), a second in FCEM about 30 seconds later, then after roughly 12 minutes a third pulse again in FCEM, and finally a fourth back in FCIM/CEB about 30 seconds afterward. After three hours, the cycle repeats itself.

The implied propagation speed of ~10 m/s rules out a traveling‐wave origin.

Preliminary checks show that neither the (unlinked) HVAC units in FCIB and FCEB (per Davide Soldani, personal communication) nor the air compressor 150 m from CEB (which is under continuous monitoring) correlate with this noise, leaving the true source still unidentified.

It might be useful to place some seismic sensors near the HVAC units in FCIB and FCEB to shed light on this problem 

ITF found in SCIENCE

08:37 UTC ITF set in ADJUSTING - see entry

08:39 UTC ITF set in SCIENCE

13:00 UTC ITF left in SCIENCE

These are the best noise budget I could achieve for the data of July 24 during the latest IPATSiA test. In each case using the last 10 minutes of data of a given state of IPATSiA, in order to avoid the initial few minute transient from turing the power of IPATSiA on or off. The best shape to fit all the data had a slope of 1/f^0.55.

Figure 1 shows the noise budget for the first position (with 40mW of power)

Figure 2 shows the noise budget for the second position.

In both cases the plotted mystery noise level is the same, and it is 10% higher than in the case of LN2 data with IPATSiA off.

Figure 3 shows the noise budget for IPATSiA that was taken in between the two times shown in figure 1 and 2 

Figure 4 shows the noise budget for one hour after the IPATSiA test, with the interferometer back in LN3. Where the mystery noise is a factor 1.7 lower in terms of mW/rtHz, as expected from the change in HOM signal recycling gain.

For this injection (April 1), we injected filtered white noise (8–1000 Hz) into the WEB using two loudspeakers. Although the loudspeakers were unable to generate acoustic waves below approximately 50 Hz, Hrec nonetheless exhibited an increase in noise between roughly 10 and 50 Hz.

Simultaneously, the "in-loop" optical lever signal (Sc_WE_MAR_PSDT2_Y2) was heavily disturbed by the acoustic injection. This disturbance may explain the rise in low-frequency noise observed in Hrec (to be confirmed).

On July 21, we repeated the same injection using only one loudspeaker, halving the acoustic power. Under these conditions, Hrec showed a much smaller response (colour scale are unchanged).

We plan to upgrade the acoustic injection system next months. The enhanced setup will allow us to vary sound levels across the terminal buildings to determine whether the Hrec low-frequency noise persists at higher amplitudes and whether its behavior is linear or shows a threshold effect. It remains to understand and correct the behaviour of this optical lever system, to check if it is really the source of the Hrec noise.

A question raised during a discussion with Paolo Ruggi was whether the coupling observed on the optical lever originates from the magnetic field.
This scenario appears unlikely: the standard Metronix magnetometers (middle row in the attached spectrogram) do record a field increase, but they sit immediately adjacent to the loudspeaker cables (see first photo).
By contrast, the triaxial fluxgate (bottom row) located near the tower—and thus close to the optical lever—shows no measurable change in magnetic field.

Here's the noise budget for the SSFS injection of August 4th.

ITF remained locked in LOW_NOISE_3 for the whole shift with SCIENCE mode (Autoscience ON); BNS Range ~55 Mpc. No relevant DMS events.

Guard tour (UTC)
21:48 -> 22:20
23:33 -> 00:11
01:33 -> 02:13
03:43 -> 04:20

ITF in Science mode for all the shift.

Guard tours (time in UTC)

18:00-18:40

ITF found locked at LN3 in SCIENCE mode.
At 11:29 UTC ITF unlocked due to an earthquake occurred in Russia (pict1).
Aurorelocked at first attempt. Auto SCIENCE mode.

Looking at the figure the error signal used for the PyTCS loop, TCS_WI_CO2_POWER_CH_PSTAB, is unaffected by the waveplate. On the other hand the total power of the DAS+CH correction, TCS_WI_CO2_POWER_CH_PICKOFF, sees the waveplate rotation. So either the actuator is wrong and it is the DAS power that was changed, which seems to me very unlikely as the interferometer in steady state looks unchanged. Or it is the waveplate that is located after the TCS_WI_CO2_POWER_CH_PSTAB power meter pickoff, which means that the error signal for PyTCS need to be changed to look at a signal after the actuator, for example change the error signal to TCS_WI_CO2_POWER_CH_PICKOFF.

ITF remained locked in LOW_NOISE_3 for the whole shift with SCIENCE mode (Autoscience ON); BNS Range ~53-55 Mpc.

Guard tour (UTC)
21:36 -> 22:13
23:32 -> 00:11
01:37 -> 02:12
03:42 -> 04:19

DET
2:30UTC - SDB2_B1p_DC power returned below its red threshold on DMS (fig.1) after initial transient starting around yesterday's activity on SR at 11:40UTC (entry #67442)

Today, with the help of a spectrum analyzer, we restore the squeezer MAIN  and CC  PLLs. The previous entry's (67359) setting probably   referred to a spurious beat signal which was very faint (about  -25 dBm).

With the new settings:

MAIN  PLL:  31594

CC PLL:       31437

the PLLs lock are stable in time. While we had no  time to work on the SC PLL, which however is not used in this period.

We also used part of the shift time to calibrate the laser temperature versus the DAQ bit count and, in the case of the main PLL, also the frequency shift to the bit count. The results are shown in the attached figures.

With these calibrations we were able to estimate that to restore the PLLs we had to increase the temperature of the lasers respectively by

MAIN PLL Temp increment 1.49 Kelvin

CC PLL Temp increent 1.43 Kelvin

which is compatible with the increase in temperature of the main laser reported in #67348

ITF found in Science mode.

At 14:05 UTC ITF in Commissioning mode for the planned activity of Squeezing; activity concluded at 18:35 UTC.

Guard tours (time in UTC)

18:00-18:40

Figure 1. Today during the relock after the maintenance I have added in the ITF_LOCK automation that an SR TX offset is added at the same time as the SR TY offset when acquiring LN3. This has worked successufully, and added the offset that of 0.015 that has been working well for the past few days.

Figure 2. The BNS range this afternoon has not been as good as before the maintenance, it may be due to some drift in alignment that occurred during the maintenance, SR TX changed by ~7urad, and the drift control is slowly recovering that with the SR TX error signal not actually managing to arrive back to zero.

This morning we performed a set of loudspeakers injections with the purpose to measure the acoustic isolation 

of the two (equal) fire-protection doors located in CEB, L2 at the entrance of the Clean room AHU room (see picture). 

The two doors are parallel approx. 2 m apart, and are structurally the same. 

Two subwoofers were positioned in the AHU room, as well as one microphone (GRAS 46AZ) named "IN", while the other 

one (GRAS 46AZ) named "OUT" was positioned on the other side of both doors (see picture) or in between. 

The used generator sent uncorrelated white noise to each speaker.

The log file with times and actions is attached. 

The attached pictures compare the two mics spectra in different conditions:

(a further analysis will follow to convert the spectra in acoustic dB versus octave bin)

During INJECTION

• Huddle test (both mics closeby)
• both doors open
• only door IN closed ("IN" is the door closer to the AHU room)
• only door OUT closed ("OUT" is the other door) 
• both doors closed
• both door closed and "OUT" mic in between the two doors

No injection, only QUIET ambient noise

• both doors open
• only door IN closed
• both doors closed

ITF found in SCIENCE
6:00 UTC ITF in MAINTENANCE

Activities communicated to CR:

• Standard IMC, CEB, NEB entrance, WEB entrance cleaning (Ciardelli, external firm) - 6:00 UTC
• TCS: chiller refill (Ciardelli) - 6:30 UTC
• TCS: WI, NI CH active servo loop test (Nardecchia, Cifaldi) - 7:30 UTC
• SQZ: laser checks (Zendri, Tofano, De Laurentis, Lunghini) - 7:00 UTC
• SQZ: replacemnt of the dust collection wafers on EQB1 in  DET Lab (De Laurentis, Gargiulo) - 9:20 UTC
• ENV: CEB Acoustic Injections (Fiori, Tringali, Loche, Paoletti) - 7:00 UTC
• VAC: Liquid nitrogen refill, NE tower large valve nitrogen tank switch, PR tower turbo pump switch (Vacuum team) - 7:30 UTC
• SUS: PR checks (Boschi) - 9:00 UTC
• SUS: WE optical lever check (Majorana, Pinto) - 9:20 UTC

Activities performed by the operator:

• OMC scan 6:00 UTC
• Thermocameras Pictures Acquired 6:50 UTC
• Power Check skipped by the operator, done by Nardecchia

10:12 UTC MAINTENANCE ended

10:54:51 UTC LOCKING_CARM_MC_IR lock failed

11:49:44 UTC ITF in SCIENCE

As agreed during the daily meeting, I disabled the servoloop at 12.17 UTC.

In order to stabilize the WI CH power, we attempted to reactivate the PyTCS-based process developed by Bas (see entry   52930 / vpm)

Just as a reminder: the probe and setpoint used by the loop correspond to the power measured by the signal TCS_WI_CO2_POWER_CH_PSTAB.
Note: decimals must be written with a dot, not a comma!

When the CH flip mirror closes (i.e., at CARM Offset Reduction 2 of 3), the loop goes automatically into pause.
When the flip mirror reopens, the loop automatically resumes.

To check if the servo is active or not, you have to look at the line in the VPM section (see in figure 1).

As for the gain, I kept the default value originally set by Bas.
After some trials, I enabled the WI CH servo at 09:26 UTC with a setpoint of 2.67 W (on the PSTAB power meter) in order to maintain a constant power of 0.64 W on the PICKOFF (DAS+CH). The servo’s behavior during the first two hours is shown in Fig. 2. It does not appear to have performed as expected. To be investigated.

To be monitored and evaluated whether it should be turned off or not.

Due to a moderately high correction on the F0 vertical control, and the anomalous vertical position of the F7 crossbar, we recentered all the filters of the PR chain with the DC motors. Now the DMS PR_vert flags are all green.

Start of operation: 11:07 (local time)
End of operation: 12:09 (local time)

ITF found in LOW_NOISE_3 and in Science Mode. It kept the lock for the whole shift.
ITF left locked.

Guard Tour (UTC):
21:11 - 21:40
23:33 - 0:13
1:34 - 2:11
3:36 - 4:14

ITF remained locked in LOW_NOISE_3 for the whole shift. ITF found in SCIENCE mode (Autoscience ON); BNS Range ~54 Mpc. Set to CALIBRATION mode between 15:30-16:34 UTC for SSFS and scattered light injections planned for today. ITF back in (auto)-SCIENCE at 16:34UTC.

Guard tour (UTC)
18:13 -> 18:51

Calibration
15:31-15:52UTC run SSFS_injection.py
15:53 and 16:02UTC restart of CALI node from VPM and scattered light injections (entry #67432) by Haevermaet. End of calibration activity at 16:32UTC. CALI node found in CALIBRATED_DF_SCATTEREDLIGHT, set to DOWN at 20:29UTC.

We have performed a statistical study to investigate whether specific CH power conditions might prevent the interferometer from reaching CARM NULL 1f. The analysis covers 600 days of data starting from April 10, 2024.
We analyzed the values of the CO₂ pickoff power (DAS + CH) for both NI and WI, measured in CITF, and we labeled each lock attempt based on whether it reached CARM NULL 1f stably (for at least 5 minutes) or unlocked beforehand.
Although unlocks can occur for many different reasons, if there were a specific condition of the CH powers preventing the lock, it should emerge statistically from the data.
The attached plots show the power trends for successful (blue) and unsuccessful (red) lock attempts. No clear separation, threshold, or instability region appears: the distributions largely overlap. This suggests that CH power alone is unlikely to be the limiting factor in the lock acquisition process.