we are entering CEB clean room, 18:40 LT.

Before measuring the locking bandwidth, the OLTF of the IMC was measured, and a 14° phase margin was measured.

We measure the locking bandwidth, which is the linearity zone between the SL cavlty length and the error signal. Initially, it was measured at 0.9 MHz

After turning the LB_M2 mirror horizonally, we managed to increase the bandwidth up to 1.3 MHz.  After that, we measured the IMC OLTF again, and found a 19° phase marging.

After that, we tried to realign LB_M2 vertically, we managed to reach a 1.0 MHz bandwidth. The UGF of the IMC OLTF was 92kHz with a 23° phase margin. In order to increase the UGF, we diminished to IMC loop attenuator value from 11dB to 9dB, which increase the main gain of the loop. The new UGF is 120kHz with a 20° phase margin. The two plots shows the OLTF of the IMC with the attenuator at 11dB and at 9 dB.

Other subject: In order to stabilize the TY degree of freedom EIB, we moved the the water tubes of IPC1. It improves the TY value and releases a bit the actuator 0.

ITF DOWN in UPGRADING mode.

planned activities communicated to the control room:

• Realignment of Master Laser wrt Slave Laser, OLTF measurements (Gosselin, Lagabbe)
• WI tower baffle installation (Ballester, Erbanni, Francescon, Gherardini)
• EIB step-motor adjustment. EIB - SAS motor control not responding, issue temporarily fixed but if will be addressed again by Bertolini during the next maintenance (Bertolini, Carbognani, Gosselin)
   

Figure 1. The B4 phase camera has an apparent offset in the sideband power measurement when no light and no reference beam is present on the phase camera. The offset is around 25e-6 in the not normalized channels. This offset has been present for at least the past three years, and has been relatively stable over that period.

Figure 2. That offset is comparable to the sideband power in single bounce, especially when the input power is low, as in this example with 11W of input power. The signal of the phase camera for the 6MHz increases only by a factor 2, in the single bounce configuration, compared to the dip in the middle which is at 25e-6 corresponding to no beam arriving on the phase camera. 

If that is a real offset, then in single bounce the power of the sidebands is overestimated by a factor, and the sideband gain is under estimated by a factor 2. It would also affect the measurement of the sideband modulation depth that use single bounce beam data and the ratio between the sideband and the carrier power as measured by the phase camera.

Analyzing data from the reference period at 11W used in the O4 detector paper (VIR-0710G-24), on Dec 10, 2023. The 6MHz modulation depth without compenstating for the offset is 0.21, while by subtracting the offset it becomes 0.15. However, the modulation depth measured simultanously with the B1p camera is 0.19, much closer to the value without subtracting the offset, and the B1p camera is not affected by the offset issue because the power is much higher, so the offset is always negligible compared to the sideband power in single bounce.

Similarly the 56MHz sideband has modulation depth of 0.21 on B4 without correcting for the offset, and 0.14-0.16 after correcting for the offset, while on the B1p phase camer it is 0.19-0.20.

/users/mwas/PC/PC_modulation_depth/SB_mod_depth.m

Figure 3-5 The OMC was scanned a few days later on Dec 19. The carrier TEM00 power is 0.0272 and the 56MHz sideband power is 0.29e-3, which correspond to a modulation depth of sqrt(4*0.29e-3/0.0272) = 0.206, which agrees with the B4 phase camera measuremetn without offset correction. 

It appears that the offset visible in the phase camera does not affect significantly the sideband power measurement despite the offset being equal to half of the sideband power measured value on a single bounce beam.

Looking at the phase camer code and raw data, this offset is likely the average value of numerical noise. 

Figure 6 shows the raw phase camera data at a time when no beam is present, and it is clearly a digital noise with an integer number of counts between 0 and 3. Looking at the computation of the sideband power in /virgoApp/PyALP/v3r0/src/main.py, it is the average value of the square of that channel over one second. 

Figure 7 for comparison shows the same raw data when the interferometer is full locked. This is only available in raw full data so we do not have recent data, but there were some times which have been archived. I did not search the archive extensively, but I do not expect we have archived single bounce raw full data. The power of the sideband in single bounce should be a factor 5e-2*0.5^2/20 smaller, where 5e-2 is the PR transmission, 0.5 the BS reflectivity and 20 the sideband gain, which means the amplitude as measured by the phase camera should be a factor 40 smaller in single bounce than with a full interferometer, corresponding to up to 15 counts. Given that the average is looking at the square of the PC raw channel, mostly the times when the value is large count in the average, and these will be less affected by a digital noise of a few counts around 15, especially if that noise becomes symmetrical once the signal is no longer at the zero rail of the digital output. 

This might be an explanation why the offset is not affecting so much the measurement of the sideband power, but that depends on how that digital noise behaves when there is a small but not zero signal. It will be easier to confirm once we get back a single bounce beam to have access to PC raw data in single bounce.

Luca, Angelo and Fabio are entering the WI tower to install the 'tower baffle'

On July 6, we tried to improve the AOM alignment by turning the LB_M10 mirror, located just before the AOM on the laser bench. The first order beam power generated by the AOM was measured using a calorimeter. The PCM was re-aligned. We managed to get 2.6 W on the 1st order beam. The setup to measure the 1st order power is shown in the following image.
 
It is not clear if we were actually improving the 1st order power, since the separation between the 0th and 1st order as small. We though that the 1st order beam was clipped by the prism located just before the PMC. We tried to move the prism, in order to make sure that the whole 1st order, and only the 1st order beam arrives on the calorimeter. 
This reduces a bit the RMS of the PSTAB_PDd_AC photodiode signal. However, we still don't know if the improvement was due to a better alignment in the AOM, or if the PMC alignment was improved. 
 
On july 7,
The oltf of the pstab was measured using the spectral-analyser application of the Moku-Pro. The Moku generates a chirp signal whose amplitude and frequency range can be tuned by the user. The chirp signal was injected in the "perturb" channel. 
Initially, the UGF was measured at ~45kHz. By turning the "piezo gain" controller in the image below, we increased the gain of the loop, and so the UGF. The new ugf value is 67kHz, with a 60° phase margin.
 

The morning was dedicated to weekly maintenance started at 6:00UTC, here a list of the activity reported to the control room:

- standard vacuum refill from 6:00UTC to 10:00UTC (VAC Team);
- cleaning of central building (Menzione with external firm: from 6:00UTC to 10:00UTC);
- PAY: NI payload assembly;
- INJ: IMC lock/alignment commissioning, in progress;

Air Conditioning
at around 7:55UTC I put back the injection area air conditioning system in "portata ridotta", I set back the "portata nominale" at 13:55UTC.

On July 22 the INGV team (Filippo Greco, Alfio Alex Messina, Luca Timoteo Mirabella) with the EGO support (Lorenzo Lunghini, Fabrizio Rossi)  will install the quantum gravimeter (presently at 1500W fiber lab:  https://logbook.virgo-gw.eu/virgo/?r=68692) inside the WEB experimental hall. The foreseen location is along the South wall of the building. The operation might take the entire day.

Yesterday I recovered a bit the SQZ system in view of the DAC firmware upgrade.

1) PLLs. I managed to lock only CC PLL. SC PLL need a check from site, I don't finde the beat note remotely. Main PLL: in order to relock it we shoulde change by about 2GHz the frequency of all the SQZ laser, because according to INJ crew the last FMODERR tuning moved by 1.2mm the end mirror of IMC. This is a longer work because we should also check if the CC laser in the new position will have again multimode behaviour. I did not do it yesterday

2) HD AA, and CC loop. Are perfectly working and we managed to measure 8.5dB of ASQZ and 6dB of SQZ deflecting the SQZ beam on SQB1 retroreflector

3) FCIM, FCEM and SQB1 bench are recovered both position and angular loops.

Some sidenotes:

After the upgrade from cent OS to Alma Linux I had to partially rewrite all the SQZ GUIs because the PyQt4 now are no more available. I did the porting to PyQt5

Olwin does not work now anymore. It works only on ctrl1

ITF DOWN in UPGRADING mode.

planned activities communicated to the control room:

- 08:20 - 10:10 UTC - NI tower baffle repositioning (VAC team, Gherardini)
- 13:50 UTC  - BPC and IMC loops recovery activity in Laser Lab  (Spinicelli, LaGabbe, DeRossi). UTA INJ set in Portata Nominale.

We continued to deploy the DAC1955 v3r3 firmware on all the DAC1955 boards  used from the rtpc6 .

Due to some contraints related to the EPRB phase camera, sharing the same DAC1955 board with the SPRB SBE control,  only the 2 DAC1955 boards used for the SPRB LC controls and the SPRB Quadrant galvo loops have been updated.

• On the rtpc6, the Tolm output data flow was reduced from 12.6MB/s  to 6.2MB/s 

Operations performed between 2026-07-03-07h04m45-UTC and  2026-07-03-08h07m33-UTC

• all the loops were opened during the time of the operations
• and successfully closed at the end

Below the detailled list of the operations performed:

• Use DaqBox  v17r11 for SPRB_dbox_bench and SPRB_dbox_rack server 
• SPRB_LC and SRPB_Quadrants : 2 DAC1955 boards
  • SPRB_DAC_LC_LVDT_in              : DAC1955_SN008 -  DBOX_SN005 - SPRB_dbox_bench      : rtpc6   shared between the LC and Quadrant galvo loops
  • SPRB_DAC_LC_LVDT_out            : DAC1955_SN016 -  DBOX_SN023 - SPRB_dbox_rack         : rtpc6  used only for the LC loops
  • the LC and the SBE  loops were successfully closed after the upgrade 
  • We checks that the Tolm packect, related to the Quadrant galvo loop, was correctly received by the shared DAC1955 board (SN008) 

This morning we entered the NI tower to reinstall the tower baffle that was temporarily removed at the beginning of the NI upgrade intervention. While inserting the first part of the baffle, we found that its frame was interfering with the recently installed cables, log_69271
We then skipped the installation, dismantled the component and made a small modification to the aluminum frame by adding a slot at the cable location (not visible in the attached picture).

The plan is to reinstall the baffle on monday starting at 11h LT.

Due to the change of the 22 MHz EOM with its spare we had retuned the longitudinal demodulation phase but the RF QPD NF and FF demodulation phases had still to be retuned.

The phase shifter (in the LNFS rack in piscina) was in local, so we swapped it to remote control and we tuned the demodulation phases by injecting a line at 2Hz:

• on MC tx (does a shift on EIB) --> and tuned IMC_QPD_FF_PHI  maximize the 2Hz on INJ IMC QD FF I H / V
•  on IB tz does a tilt (tx for BPC) --> tuned IMC_QPD_NF_PHI to maximize the 2Hz on INJ IMC QD NF I H / V 

With the new demodulation phases we checked again the offsets on the galvo (fig 1 and 2). 

We zeroed the AA error signals and measured again the sensing matrix with the AA loop open. 

After a manual adjustment on the H direction we left the IMC AA loop closed in full bandwidth.

Moreover, this morning we had a lot of glitches on the galvo NF H (fig 3). The corrections are quite high (more than 4V), but they are like this since months. However, it seems to have disappeared as soon as the corrections went just below 4.2 V (they are now around 3.9 V). To be monitored.

We will continue the work on monday (check the FFTs) and maybe manually look for another wp, since looking at the rfc reflection it seems that we are far away on the horizontal with respect to the past.

Below the list of the activities communicated in control room:

- DAQ: DAC1955 firmware update for EDB and SPRB;
- VAC: NI tower baffle installation;
- INJ: IMC lock/alignment recovery;

SBE
SIB2 vertical correction recovered with F0 step motors at around 6:00UTC.

We completed the DAC1955 firmware update for the EDB part:

• rtpc1:  Tolm output data flow reduced from 54.5MB/s  to 18.6MB/s
• Operations performed between 2026-07-03-09h30m38-UTC  and 2026-07-03-09h41m49-UTC

Below the detailled list of the DAC1955 boards updated and the related ACL's server:

• EDB : 2 DAC1955 boards
  • EDB_DAC                         : DAC1955_SN046 - DBOX_SN051 - SDB_EDB_dbox_rack     : rtpc1
  • EDB_DAC_DETEEroom   : DAC1955_SN053 - DBOX_SN051 - SDB_EDB_dbox_rack     : rtpc1
  • plot showing  the successfull EDB OMC temperature stabilization and the presence of the PZT modulation line
  • we checked  that the Tolm packects are correctly received by the 2 DAC boards

We noticed that the movements of the PMC piezo are correlated with the output power of the PMC. This correlation is due to an offset in the PMC lock signal, which keeps the the transmitted signal away from its maximum. 
At 9:52 utc, we changed the offset in the ISYS_acl config file from -0.12 to -0.015 V. This set the central value of the PDH error signal to 0 V as wanted. However the PMC control loop brings the control loop signal to -22 mV instead of 0 V. So the correlation is still present after this intervention.
To correct the offset, we used the method descibed in gitlab. Instead of injecting a 270 Hz line, the injected signal is at 123 Hz (frequency written in the config file) with 0.05V amplitude.
By setting the offset level to -0.025 V, the 123 Hz line on PMC_tra was minimized. And the coherence at low frequency diminished.
 
 

We continued to deploy the DAC1955 v3r3 firmware on all the DAC1955 boards used at the NEB and WEB, execpt the ALS ones:

• NEB - rtpc8 : the 7 DAC1955 boards are updated
  • Tolm output data flow is reduced from 29,4MB/s to  5.9MB/s  
• WEB - rtpc9 : the 7 DAC1955 boards are updated
  • Tolm output data flow is reduced from 29,4MB/s to  5.6MB/s
• We faced some issues with these rptcs  as 
  • we would like to run the DAC1955 using the synchronous mode 
  • and there are some Acl's servers running at a loop frequency lower than the Tpro's one and for the ones running at the same frequency if their elapsed time is lower than the Tpro period, they are interleaved
  • until we understand that the 1PPS flag was not correctly set in the Tolm packet header 
• As consequence for the DAC1955 Tolm packet sent by an ACL's task running  with  a loop frequency lower than the Tpro's one, the DAC1955 board Tolm packet reception have been configured as asynchronous.
• For the others DAC1955 boards, the Tolm packet reception is configured as synchronous

Operations performed between 2026-07-01-09h11m50-UTC and 2026-07-01-18h43m53-UTC

• all the loops were opened during the time of the operations
• and successfully closed at the end

Below the detailled list of the DAC1955 boards updated and the related ACL's servers  at WEB

• WEB_NCAL server was already done : 2 DAC1955 boards
  • WEB_DAC_NCAL_TCS      : DAC1955_SN053 - DBOX_SN121 - SWEB_dbox_rack       : rtpc9 
  • WEB_DAC_NCAL0             : DAC1955_SN073 - DBOX_SN175 - SWEB_dbox_rack       : rtpc9
  • NCAL loops successfully closed
•  WEB_PCal_fast server  was already done : 1 DAC1955 boards
  • WEB_DAC_PCAL                 : DAC1955_SN011 - DBOX_SN049 - SWEB_dbox_rack       : rtpc9
  • PCal loop successfully closed 
• SWEB_LC  and  SWEB_SBE servers : 3 DAC1955 boards running at 400kHz  in asynchronous mode
  • SWEB_DAC_LC_LVDT_out  : DAC1955_SN044 - DBOX_SN043 - SWEB_dbox_rack       : rtpc9 
  • SWEB_DAC_LC_LVDT_in    : DAC1955_SN038 - DBOX_SN014 - SWEB_dbox_bench    : rtpc9 
  • SWEB_DAC_SBE0               : DAC1955_SN039 - DBOX_SN062 - SWEB_dbox_rack       : rtpc9
  • loops succesfully closed. The TX and TY position changed as we restored the ones of the configuration file 
• SWEB_PSD_sensing server : 1 DAC1955 board  running at 400kHz  in asynchronous mode
  • SWEB_DAC_PSD                 : DAC1955_SN035 - DBOX_SN014 - SWEB_dbox_bench     : rtpc9
  • we checked  that the Tolm packect are correctly received by the DAC board

Below the detailled list of the DAC1955 boards updated and the related ACL's servers  at NEB

• NEB_NCAL server was already done : 2 DAC1955 boards
  • NEB_DAC_NCAL_TCS         : DAC1955_SN018 -  DBOX_SN119 - SNEB_dbox_rack       : rtpc8
  • NEB_DAC_NCAL0                : DAC1955_SN006 -  DBOX_SN118 - SNEB_dbox_rack       : rtpc8
  • we checks that the Tolm packect are correctly received by the DAC boards 
• NEB_PCal_fast server and NEB_NNC_Tilt server   were already done : 1 DAC1955 boards
  • NEB_DAC_PCAL                   : DAC1955_SN121 - DBOX_SN035 - SNEB_dbox_rack       : rtpc8
  • PCal loop successfully closed 
• SNEB_LC  and  SNEB_SBE servers : 3 DAC1955 boards  running at 400kHz  in asynchronous mode
  • SNEB_DAC_LC_LVDT_out    : DAC1955_SN031 - DBOX_SN034 - SNEB_dbox_rack       : rtpc8
  • SNEB_DAC_LC_LVDT_in      : DAC1955_SN003 - DBOX_SN013 - SNEB_dbox_bench    : rtpc8
  • SNEB_DAC_SBE0                 : DAC1955_SN102 - DBOX_SN127 - SNEB_dbox_rack       : rtpc8
  • loops succesfully closed
• SNEB_PSD_sensing server : 1 DAC1955 board  running at 400kHz  in asynchronous mode
  • SNEB_DAC_PSD                   : DAC1955_SN007 - DBOX_SN013 - SNEB_dbox_bench     : rtpc8
  • we checked  that the Tolm packect are correctly received by the DAC board

On Tuesday morning we completed the realignment of the BPC loop. We started by zeroing the corrections sent to the actuators and centering the DC offset on the amplifier, in order to maximise the actuator dynamics.

With the PMC locked and the BPC loop closed, we used the M6 and M8 mirrors on EIB to center the BPC QPDs. The loop closed easily in full bandwidth, with less than 1V of correction per DoF. The power measured at the output of EIB_M9 was around 29W, similar to what we measured before the April intervention.

We then closed the BCP at the usual setpoints, and coarsely realigne MC and SIB1 in order to relock the IMC. Despite the presence of peaks above 8-10W (the power threshold for the rampauto) we didn't succeed in relocking.

We found a phase shift of almost 45° on the demodulated error signal, most probably due to the swap of the 22 MHz EOM with its spare. The new value for the phase is 2.85. A finer tuning will likely be performed once the IMC recovery is complete.

Today, the OLTF of the IMC under these new conditions was measured (see Fig. 1), showing an UGF of less than 90 kHz and ~23° of phase margin, significantly lower than the last measurement at the end of April.  This may be a direct consequence of the chiller and the slave laser issue from early May, with the SL not being properly seeded by the ML beam. For the time being, we will not work on improving the SL loop.

With the IMC locked, we checked the IMC working point (new value: −0.85), realigned the IMC cavity, and measured the new AA sensing matrix and angular galvo offsets. The AA has been closed in full bandwidth. Moreover, the RFC locked directly after closing the IMC AA.

Some remarks, though:

• The longitudinal working point is smaller with respect to the last measurement.
• The AA error signals show quite large offsets across all channels.
• The IBz loop, which links BPC_X to the Z position of the SIB1 bench when the IMC is locked, is currently not working. The BPC X DoF, however, works correctly when using the BPC QPD error signal. For the time being, we have manually adjusted the X setpoint to minimise the IBz signal. We will investigate further to identify the origin of the problem.