Today new versions of the single arm IR locking nodes, NARM_LOCK and WARM_LOCK, have been put in operation and tested. This serves three purposes:

1. allow a (partial) lock with reflection signals, namely SPRB_B4_6MHz_I (for both arms), as fallback configuration in case the B7/B8 beams are not available (up to a point, more on this later);
2. ease the integration with the ALS nodes, ALS_NARM and ALS_WARM, and the forthcoming management of the four of them by the single "arms" node;
3. polish the code, that was not done in a very long time.

Here are the main changes:

• we now import virgotools as we are supposed to (from virgotools.everything import *);
• the state indexes have been rethinked and changed;
• the state names have not changed aside the two handoff states, which are now HANDOFF_B{7/8}_TO_B4 and HANDOFF_B4_TO_B{7/8};
• some states have been removed from the drop down menu (request = False) as they should not be requested manually; if really needed, the usual full states list window should be used instead;
• the path for the lock with B{7/8}_56MHz_I has been removed entirely, as it was basically never used;
• the lock with the B4_6MHz_I signal has been re-implemented; a few notes about it:
  • since that we are forced to use the Guided Lock to acquire the lock of the arm, it is not possible to lock an arm directly with the B4 beam, as the Guided Lock is strongly dependent on the availability of B{7/8}_DC; instead, we can lock with the usual B{7/8}_6MHz_I signals and then handoff the control to B4_6MHz_I; this has been enforced by breaking the symmetry of the two locking paths by severing the edge between the ACQUIRE and the ACQUIRE_WITH_B4 states; so, selecting directly the LOCKED_WITH_B4_BOOST_ON state will lock the arm as usual and then perform the handoff (see next point);
  • selecting the LOCKED_ON_B4_BOOST_ON state will ask the transition through the HANDOFF state; in the final state, aside the obvious change of sensor, the drift control is disabled (as it relies on B{7/8}_DC) and the decorator which checks if the arm is still locked is changed from a "min" check on B{7/8}_DC to a "max" check on B4; due to the logic of Guided Lock triggers and flags, it is also needed to force a relay to be always 1 (LSC.B{7/8}_GL_DC_TRIG_LO);
  • the inverse handoff is possible, selecting back the usual LOCKED_ON_B{7/8}_BOOST_ON state; given that this should be done after verifying the availabilty of the B{7/8} beam, doing so will enable again the drift control, change the decorator back to the B{7/8} one and put the aforementioned relay back to automatic mode (-2); this very last thing is done also in the DOWN state, in case of unlock;
  • the automatic check of the availability of the B4 PD (which should be B4_PD2) will be tested and added tomorrow;
  • the lock on B4 was tested on both arms and the logic works, disabling B{7/8}_DC did not unlock the arms;
• the ACQUIRE state has been made a little more beam-independent (aside the Guided Lock parameters), as one would want the two paths with B{7/8} and B4 available; as said above, it will be studied if this is really possible or if the initial lock is forced to be using B{7/8};
• there are two new states, ALIGNING_{NORTH/WEST}_ARM and ALIGNED_{NORTH/WEST}_ARM, which realign the arms; this was earlier embedded in the ACQUIRE state, but it had to be separated and anticipated in order to ease the forthcoming management of all the four single arm nodes;
• there is a tentative WIP state, named for now OFFLOADED_TO_MANAGER: in this state nothing is done or checked, and this is the point: it will be used as a "parking" state for the node once everything is under the direct control of the single "arms" node; it can be only reached (but not requested manually) from the "final" state for these nodes, LOCKED_ON_B{7/8}_BOOST_ON.

The new node layout is attached, PLEASE RESTART ANY INSTANCE OF THESE NODES CURRENTLY OPEN, as the states list in the drop down menu has changed.

The aim of the shift was to further study the CARM offset reduction process, especially check the normalised error signal for MICH. We first locked the arms on IR and then on the green, using the Metatron nodes. We moved then to green locking with the offset on CARM and DARM with the usual procedure, and we went all the way without issues. After setting the offset of 3kHz on CARM, we moved to check the pre-alignment of CITF, which was sufficient.

The test we did with the normalized signal for MICH was not that successful, as it did not speed up the lock acquisition and the normalization made, when we were correctly triggering with low powers, the error signal becoming huge and, consequently, the correction signal as well. The issue here is that, given that we have a mixed driving due to geometry, a high MICH_CORR will impact on all three DRMI actuators, BS PR and SR.

We moved back (also in Metatron) to the standard sensing for MICH and we could relock the DRMI. Then, after we were locked on 1f with the PR alignment in full bandwidth, we checked the signals coming from SDB2_B1p_QD2 for the drift control of the BS mirror: their demodulation phase seemed ok, and they were sensitive to manual movements of the BS mirror. We made two tests (see Figures 1 and 2) closing the drift control loop, but in both cases after some time the control diverged and we lost the lock. The Galvo loops seem to work, up to our knowledge, but there are some spikes on the corrections (Figure 2).

With the BS loop left open, we proceeded to the lock on the 3f signals, which needed an adjustment in the mich_1f_3f_ratio parameter in order to keep the same UGF after the handoff, then use the gain handle to tune it not to saturate the corrections; probably the same could be done to the SRCL equivalent.

At around 17:00 UTC we had the Arms+DRMI locked, CARM offset 3 kHz and DRMI on 3f. At 17:04 we reduced the offset to 300 Hz, doing the usual adjustments needed in this phase (B1p_DC and B4_DC triggers forced to 1, MICH and PRCL gain reduced by ~ 33 %). In this stage we adjusted by hand the amplitudes of the UGF lines, which are not needed that high in this phase (note that the DARM one is off by default): MICH = 0.3 / PRCL = 5e-3 / SRCL = 0.6 / DARM = 5e-3.

At 17:10 UTC we made the handoff from the beating signals to the B7/B8 reconstructed ones, CARM first and then DARM. The 300 Hz CARM offset was -405 in the new units; we changed the offset to -430 at 17:22:36 UTC, and to -455 at 17:23:56 UTC, and we unlocked shortly after. Unfortunately, in the following trials we unlocked before reaching this point again.

In the central part of the shift, having to relock everything from scratch in any case, we moved to the study of the lock of the single arms on the IR beam, in reflection (using the B4_6MHz_I error signal, see Figure 3); we managed to do it on both arms and more technical details on the Automation will follow in a separate entry.

A note about the SWEB_dbox_bench process: given that we wanted to make sure that we could stay locked with the End benches out of position, a fast way to simulate this is by closing the B7/B8_PD1 shutter; this worked well for B7, while it did not work for B8, as closing the shutter did not bring any results (we tried both PD1 and PD2 shutters, with no success); looking at the process log, there are the following three lines, however not followed by the usual logging info about the telnet connection, as if there is some communication problem towards the PDs:

• 2021-05-17 19h34m08 UTC Cm>ServeDoConnect> Connection in progress - CmName NULL, port 0, Host olserver52.virgo.infn.it, owner virgorun, socket_id 11 (16384,32768), local 0, alive 1
• 2021-05-17 19h43m51 UTC Cm>ServeDoConnect> Connection in progress - CmName NULL, port 0, Host olserver52.virgo.infn.it, owner virgorun, socket_id 11 (16384,32768), local 0, alive 1
• 2021-05-17 19h46m15 UTC Cm>ServeDoConnect> Connection in progress - CmName NULL, port 0, Host olserver52.virgo.infn.it, owner virgorun, socket_id 11 (16384,32768), local 0, alive 1

In any case, we mimicked the same effect by putting the weight of B8_PD1_Blended to zero in the B8_DC weighted sum, and restored it to 1.2855 afterwards.

In the latest part of the shift we moved back to the full lock acquisition, but starting from around 20:13 UTC the Injection system started to suffer from fast unlocks, progressively increasing over time up to the end of the shift; we alerted the INJ on-call.

We leave INJ_MAIN in the target state IMC_RESTORED, and the arms nodes locked on the IR beam.

This afternoon was dedicated to ISC, the work wen on all the shift without any major problem, from 20:05UTC to 20:35UTC the injection system passed through a period of instability (ISYS on-call informed), activity concluded at 21:00UTC, we left the two cavities locked on the infrared.


Others activities:

- SQZ: PSD installation on SQB2 (Polini, Vardaro);

DET
At 16:43UTC I moved SDB1 in a safe position for the CITF locking activity.

FFT of NN channels from each building CEB,NEB and WEB reading from the raw_ll.ffl files

An interface to align FC mirrors has been prepared and tested (Pic.1). It is now accesible from the main menu (Pic. 2).

The code is in saved in: virgoDev/PyGUIs/fcUI 

Today we changed the crate of FCIM to the spare one sent from Nikhef. After changing the crate, all the monitor signals including optical lever and the bench seemed more noisy than the previous crate (fig 1). By zooming in the plots, we found all the signals have oscillation around 9Hz with different amplitude(fig 2). The jumps on the coil drivers still exist. 

The aim of the shift was to complete the work started in 51795  about the normalization of the MICH errsig.

At the beginning of the shift we locked the ALS sytems and noticed that the green lock was unstable (unlocking every few minutes) on both cavities due to the alignment not being optimal. We then tuned the BPC working points in order to maximize the transmission and minimize the reflection while locked the new working points have been saved on the config file. After this procedure, a lowering of the ALS fast loops gains to suppress high frequency glitches was necessary.

• NEB_fast.GAIN 1.5 --> 1.2
• WEB_fast.GAIN 2 --> 1.5

We also investigated a bit the weird behaviour noted on the North end green reflection signal (SNEB_PD_GREEN_DC / ALS_NEB_PD_REFL_LP) which show small power steps since 12/05/2021 (see figure 1, 2 and 3). Currently the jumps are not large enough to hamper the lock of the system and are not present on other signals (transmission signal, BPC PSD signals, PDH error signal).

The rest of the activity was postponed due to the instabilities of the INJ system that caused frequent unlocks and the relative activity to mitigate them. (See entry 51813 and 51810)

The system has been left with the ALS nodes in DOWN and the arms locked on the IR signals.

The planned commissioning activity started around 7:00 UTC.

Unfurtunately the ISYS was not collaborative as we suffered of many fast unlocks (see attached plot) which prevented the smooth running of the activity.

At around 9:00 UTC we contacted the ISYS on-call, see On-call intervention.

Other activities:

- TCS: CO2 guardian system ON, all the related flags have been unshelved on DMS under request of Ilaria;

- SVS: Optimization of mode matching on EQB1 for SQZ measurement (Sorrentino, Garaventa)

- SIN: PSD installation on SQB2 (Polini, Vardaro, ?)

ISYS
(17-05-2021 09:00 - 17-05-2021 10:45) On site
Status: Ended
Description: IMC unlocks
Actions undertaken: see entry 51810

Following the IMC baffle test, this morning we have analyzed the throughput losses of the cavity in two different moments: when the High Order Modes (HOM) have appeared and when the system was in normal conditions. 

On May 16 at 00:40h UTC ("normal" period), we had:

•  mismatch = 7.4%
• Output of IPC on EIB = 35.7 W (calculated with calibration of #50848)
• IMC_TRA = 26.4 W;
• "Other losses" = 0.5% (VIR-0225A-18)
• In this conditions, we have 18.15% of throughput losses, corresponding to a round-trip loss of 605 ppm.

On May 15 at 13:18:49h UTC (HOM period), we had:

• mismatch = 7.1%;
• Output of IPC on EIB = 36 W;
• IMC_TRA = 23 W;
• "Other losses" = 0.5%;
• Here, the calculated throughput loss is 28.5% (more than 10% compared to "normal" conditions), corresponding to a round-trip loss of 1048 ppm.

Figure 1 highlights the impact of the electrical grounding work. It shows three times when the input mode cleaner is not locked, with several photodiode channels after the input mode cleaner. So times when there is no or little light on these sensors, and electronic noise is dominant. The purple curve is in the night before the electrical grounding work, and in brown is the night after. This decreased the 50Hz and its harmonics by factor between 2 and 5, depending which sensor we look at. In addition it reduced also the wide shoulders around the 150Hz, 250Hz, etc. There have been some other small improvement in the electrical connection on May 4, but this had no noticeable impact. The blue curve shows data on May 11, which are comparable (slightly higher 50Hz harmonics) than just after the electrical grounding work.

This morning we were called by Berni due to frequent IMC unlocks.

As an attempt to mitigate the problem we performed the following actions:

- we moved the electronics of the ML away from the same breadbord of the optical setup (1st attached picture).

- we tuned the input ML polarization by turning the waveplate before the collimator from 182° to 185° (2nd plot) while checking the stability when moving the fiber. By turning the waveplate we noticed that the output power dropped a lot on both polarziation. This behaviour could be due to the fiber connector between the two PM fibers.

- we installed a beam dump on the SL input in order to get rid from possible backreflection when SL unlocks occur.

- we finally lowered the PSTAB gain in order to have the usual UGF of 75 kHz (3rd picture). We had in fact more than 100 kHz.

We relocked everything and we will keep monitoring hte system.

We also checked the UGF lines amplitudes on the CITF error signals I and Q components during the lock at 12.10 UTC of 17/03/2021  1305029518 , to see if the demodulation phases of the CITF error signals needs to be retuned.

For the LSC_B2_6MHz signal (see table below), used for PRCL, the 67.1 MHz line injected on PRCL shows a ratio of 38 between the I and Q  ASDs at this frequency, meaning the phase is very well tuned to optimize the PRCL gain. The amplitude of the MICH line is relatively high on this signal (compared to the PRCL line), and this could be related to a non optimal driving of MICH.

For B2 56 MHz (I used for MICH, Q used for SRCL) we have instead a less optimal I/Q ratio on the MICH/SRCL lines, but the phase is close to being optimal (~12.4°)  for a minimization of the PRCL coupling to MICH.

This morning at 10.55 LT the CO2 guardian system has been activated (see attached figure).

All the flags in the DMS related to this system have been un-shelved.

ITF State: Infrared cavities locked
Quick Summary: IMC and RFC locked; all Suspensions loops closed; all SBEs loops closed, SDB2 tracking loop Open; NEB_Eurotherm device not responding.
No Activities ongoing.

I think there is a "typo" in this entry:

you wrote

"On Wednesday 20210418 we replaced the old ADC7674 adc boards ..."

but maybe the date should be

"On Wednesday 20210428 we replaced the old ADC7674 adc boards ..."

This seems not important, but while searching for works related to the INJ lab (electrical works done the days after) I have been unable to correlate actions.

Following the recovery of the original set points for SQB1 suspension, we checked the relative alignment of the BAB retro-reflected from SQB1 and the LO beam on the EQB1 HD. We found the two beams quite close to each other on the HD cameras. We set the biases of HD AA mirrors actuators to zero, and we manually overlapped the two spots on both cameras, acting on the HD AA mirrors. We could see the interference between BAB and LO, see picture, and after engaging the automatic alignment we measure a visibility of about 80%. Mode matching needs to be tuned. We steered the EQB1 delay line so that now the BAB-LO alignment is equally good in the two paths of the BAB (i.e. delay line or SQB1 retroreflector).

The shift was dedicated to the two planned activities: SLC - tests on IMC instrumented baffle, carried out by Menendez, Karathanasis, Chiummo, and TCS - Check of CO2 DAS, carried out by Nardecchia. 
From 8:50 UTC to 9:24 UTC all the INJ Loops were opened to allow the IMC mirror to cool down for the measurements, the rest of the activity was carried out from the Control Room.
Nardecchia worked during the morning on the NI CO2 DAS and on the WI one during the afternoon.
The SLC activity was concluded at 14:30 UTC (see report #51803), while the TCS test was concluded at 17:50 UTC (see report #51801).
Infrared cavities left locked.

SUSP
From 13:29 UTC to 13:40 UTC Boschi, from remote, restored the missing data from Sa_SR ( see report #51802). During this period, from around 13:30 UTC to 13:38 UTC, we also experienced a loss of signals from all the Suspensions on the DMS (see attached image #1), while instead, the data were available online (see attached image #2). Boschi reported that this behavior was not expected, further investigation needed.

In the TCS chiller room I increased the flow of the fan coil (up to level II of three) and I turned ON the chiller guardian in order to reach the thermal steady state during the weekend. 

NI CO2 BENCH
The DAS quality has been checked after the realignement of the pin hole performed Monday 10th (51721).
As expected, the distribution of the power appeared quite bad (see fig 1 and 2). Thus, we acted on two different mirrors along the inner and outer rings layout to recover a good power distribution on both rings (see fig 3).

WI CO2 BENCH
The same check has been performed on the WI bench. The DAS is shown in fig.4.

Today in collaboration with Antonino we studied the performance of the IMC baffle in different situations.

1º We switched on the baffle until the temperature of the ADCs reached a plateau. Once this point was reached the loops of the IMC were opened and corrections unplugged from the piscina. We let the IMC to cool down for ~25 min. This cooling down was also visible in the temperature sensors of the baffle (See Temperature_Holes.png from 11:00 to 11:30). Everything was restored back to the nominal condition at the end of this test.

2º We displaced the impact point of the beam in the mirror by modifying the offset of the quadrant photodiodes behind the IMC, this is also visible by looking at the spot on the mirror hitted by the beam using the infrared camera (e.g. Camera_103038_Diagonal.gif ). These movements had a clear effect on the sensors of the IMC as can be seen in figure Temperature_Holes.png or Sensors_2part.png in which we observe rises and falls in the number of counts corresponding to different impact positions. During one of the displacement high order modes appeared and remained during part of the test (See Camera_131859_Vertical_HM.gif). We restored everything back to the nominal condition once we finished our test (See 210515_instr_baffle_test_summary.png).

Analysis will follow.

SR Data are now back. Top crate has been rebooted. All controls are now closed.

ITF State: Infrared cavities unlocked
ITF Mode: Commissioning.
Quick Summary: IMC and RFC locked; all Suspensions loops closed, SR IP data missing; all SBEs loops closed, SDB2 tracking loop Open; NEB_Eurotherm device not responding.
Activities ongoing since this morning: Preparation work for tests on IMC instrumented baffle (Menendez, Karathanasis, Chiummo);

The collimation of the green beam was improved yesterday by acting on the EQB1 telescopes, and then aligned again on the center of the SQB2_M1 and SQB2_M2 mirrors on SQB2. This morning we found the green beam sensibly misaligned on SQB2, because the SQB1 suspension was left uncontrolled by yesterday. Before engaging the SQB1 local controls again, we balanced the bench in order to recover the reference set points on which the IR beam from EQB1 was originally aligned. We added  temporary masses at symetric positions to lower the bench down to the nominal height in air; we updated the TY set point, we disabled the TY low noise mode, and we engaged the SQB1 local controls, Fig 1.

We then moved the steering mirrors on EQB1 manually to get the green beam well centered again on SQB2 optics. We found the SQB2 too high by more than 3 mm w.r.t the nominal position, so we added symetric masses close to the northern and southern sides of the bench to lower it. Fig 2.

We worked to align the green beam towards FCIM. At the beginning, on visual inspection of scattered light in the SQB1-FC link, the beam was pointing a bit eastwards. We first tried to align the beam along the S-N direction by acting on the SQB2 suspension. We changed SQB2_LC_TY by more than 8 mrad, Fig 3, then we realized this was not enough and we decided to block the bench and move the SQB2_M1 steering mirror manually. As the blocking system is not available at the moment, we used some temporary spacers, resulting in a bench position with large residual tilts, Fig 4. By manually steering SQB2_M1 while looking at FCIM from the sout-western viewport, we eventually found the beam hitting the mirror cage, Fig 5. However, when we tried to center the beam on the filter cavity input mirror, we found that the beam got cut in the upper part of the pipe. We then started to move SQB2_M2, looking for a symetric configuration in which we could steer the beam on the upper and lower sides of the FCIM cage with SQB2_M1. We found that this is only possible when the beam on SQB2_M1 is much lower the the center of mirror, Fig6. This is suggestive of a vertical mismatch between SQB1 and FCIM, requiring further investigation.

We centered the beam on FCIM, and then we looked for the retro-reflected beam on SQB2. We moved FCIM by acting on the marionette. We applied horizontal and vertical tilts by a few mrad, Fig 7, but we didn't find the retro-reflected spot. So we gave up for the moment, and we decided to try next week with a more systematic approach by applying a spiral to the FCIM tilt angles.

The aim of the shift was to test the lock of the DRMI during the CARM offset phase with a normalized error signals for MICH, in an attempt to further increase the reliability of the lock acquisition.

According to simulations, the most promising signals for the normalization of the MICH errsig (B2_56_MHz_I) are B4_DC^2 and B2_112MHz_Mag. The simulations took in consideration both the capability of the normalization signal to extend the linear region of the PDH signal (see figure 1),  and the increased robustness of the normalized error signal against optical gain loss caused by imperfect alignment of either PR or input mirrors (see figure 2 and 3).

The new normalized error signal have been added in Acl and in the DAQ and are temporarily named LSC_B2_56_MHz_I_NORM_B4_DC_SQ and LSC_B2_56_MHz_I_NORM_B2_112 respectively.

After a series of unlocks of the injection system (15.45, 15.53, 16.11, 16.19 UTC) , at 16.30 UTC a first set of attempts have been performed by locking the DRMI (with -3kHz of CARM offset) with the LSC_B2_56_MHz_I_NORM_B4_DC_SQ signal for MICH. Considering that the value of B4_DC during the previous locks 2.45 mW (see figure 4), an initial scaling factor of ~6 (2.45^2),   has been used for the new normalized error signal with respect to the non normalized one. This led to an overall error signal calibration factor (mich_1f_norm_cali) of  1411.

However, such signal generated immediate saturation of the BS, PR and SR mirrors during the lock acquisition (see figure 5) . The reason of this is due to the current setting of the triggers, that often trigger when B4_DC is still quite low (~0.1 mW DC), making the normalization with B4_DC^2 magnify the error signal too much near the edges of the trigger. Attempts in tuning the gains (aggressively) and to increase the restrictivity of the triggers to avoid the saturations were unsuccesfull.

We then proceeded testing the normalization using B2_112_MHz ( LSC_B2_56_MHz_I_NORM_B2_112) instead . Using the data in figure 4 we obtained a scaling factor of ~1/115, leading to an overall error signal calibration factor (mich_1f_norm_cali) of  2. This error signal resulted in lower saturation issue than the previous one, but a second series IMC unlocks at 18.53 interrupted the first tests. A relatively long period of stability of the IMC system allowed us to perform a few additional tests with the LSC_B2_56_MHz_I_NORM_B2_112 that led in a lock of the DRMI at 20.01 UTC (which then unlocked during the handoff of to the 3f signals, see figure 6) and a second lock at 20.06 UTC, which led instad to a successfull handoff to the 3f signals. In both successfull cases, the MICH gain was lowered from 2 to 0.7 and then increased to 1.5 before the handoff. At 20.10 UTC the IMC started unlocking repeatedly again and did not allow to conclude the activity. It is not clear yet how much this normalization improves (or hampers) the reliability of the DRMI lock acquisition.

The interferometer has been left with all nodes in DOWN to avoid issues in case IMC instabilities recur again.