I performed a phase scan to investigate the sqz and asqz level after the tuning of the ITF (log file attached)

The measured squeezing level in DARM and in HREC is about -3dB whereas the measured asqz level is about 5.6dB (black lines in fig1)

The maximum squeezing level correspond to a CC phase of 0.6 rad, whereas the minimum of the mag is 0.9 rad.

I left the squeezer with these phase parameters and the AA dither line amplitude equal to 0.007. After the tuning the ITF unlocked. After the relock the squeezing level was 3dB see attached plot

Phase scan (figure 1)

• squeezing level -2.4 dB, antisqueezing level 5.3 dB.
• Red line Darm 1350-1750 Hz, Blue Line: Darm 2600-2800 Hz, Yellow Line Hrec 2600-2800 Hz, Violet line Hrec 1350-1750 Hz, Green 2650-314 0Hz 
• The green line is a BRMS computed with window of 1s whereas the other BRMSs are computed with 20s windows
• The phase scan in DARM is less visible than in HREC: this was due to the bad CMRF of the ITF, so we had a lot of frequency noise
• The phase noise was performed in a not good alignment condition so the maximum squeezing level was 2.4 dB
• This is visible observing the B1_PD1_mag 3.0e-3V when was misaligned and 3.5e-3 V when I realigned it exploting the dither loop
• The brms channel computed with 1 s windows is faster than the other, so it allow to computed better the phase distance between mag and sqz (0.33 rad)

Night Injection (figure 2)

• The night injection, after the realignment started with -2.8 dB of squezzing in the bandwidth between 2650-3140Hz
• After 6 hours the sqz level was decreased to -2.2 dB, but this was probably due to a 56MHz RIN, see B1s2 DC photodiode (second line of the plot)
• After the locking seems that for long time we injected antisqueezing, but this can be due to hrec not updated or a bad working point of the itf. The phase error signal of the slow CC loop was in a good position and not drifted (third line)

Coherent control phase noise (figure 3)

I performed the CC phase noise calibration with two different methods

Method 1:

1.  open shutter, CC open,
2. Measure the Aplitude of the fringes and from them obtaining the V/rad
3.  PD1cal =  0.00399 V/rad,   PD2cal =  0.10984 V/rad

Method 2:

1.  Open shutter, Close CC,
2.  perform a phase scan
3.  plot SQZ_B1_PD{1,2}_Q as a function of the CC phase,
4.  fit the linear part of the curve 
5.  PD1cal =  0.00388 V/rad,  PD2cal =  0.11375 V/rad

The two methods are consistent. I computed the rms phase noise of the CC between 0.05Hz and 25 kHz and I obtained 20 mrad in loop (PD1) and 23 mrad out of loop (PD2).

Massimo and Davide ideed found the timer that was responsible of the daily 1s switch off of the CEB hall air conditioner. After an attempt to reprogram it which failed (the conditioner continued with the periodic switch off) today the timerhas been electrically bypassed.

The problem was in the power adaptor that broke. ELE team replaced it. Now the system is back in operation. The ENV_LB_MIC and ENV_EIB_MIC channels can be unshelved in DMS.

This morning (March 21 0730UTC) the BS loop was found open (as it was highlighted by the logbook entry from the nigth before, 45333). The BS_TX and TY were put in a configuration for minimum B1p.

Then we wanted to compare the two configurations: Low B1p and low CMRF.

Thus a sweep was conducted to determine the best working point for the BS TX and TY in terms of CMRF.

In a well functioning ITF the two working points should coincide. However, these two metrics were inversely proptional with the position of BS_TX and TY. Two set points were determined for each requirement. At each setpoint, the position of DIFFp TX TY and PR TX TY were tuned for minimum power on B1p and maximum power on B4_2f (also B5) respectively.

1. Minimum CMRF: BS_TX: -0.013 BS_TY : -0.006
2. Minimum B1p power: BS_TX: -0.017 BS_TY : 0.005

The attached plot (190321_BS_TY.png) shows a the behavior of the CMRF, B1p, and B4_2f  through a scan of BS_TY at a gps of 1237189790 for 30 minutes. It is shown that for the minimum CMRF the setpoint of BS_TY should be around -0.006, althought B1p power is at a maximum for this angle. Conversely, the minimum B1p power is for BS_TY > 0.002 and the CMRF is steadily increasing in this region.

The second plot is compares the two set points. The blue line is the low CMRF working point and the purple line is the low B1p working point. For the low CMRF, there is a lower difference between HrecClean and Hrec, 0.5 Mpc, which means low 56MHz RIN coupling. The low B1p shows 2 Mpc of difference. There is negligible difference in the B5 56 sidebands power between the two setpoints. The B4 sidebands are slightly lower for low CMRF. The structures from the detection are visible in the low CMRF Hrec, meaning that diffuse light is visible on the detection bench (which means not optimal alignment). Moreover, on the low CMRF working point the alighnment is less stable as shown by a larger fluctuation of power on the dark fringe.

To conclude, the low B1p point is more stable and robust. But it spoils the squeezing injection effect (high frequency noise). The low CMRF is not a stable working point.

This morning I found the ITF locked in Science Mode, the activity scheduled (ISC) started without troubles.

7:46 UTC - Commissioning Mode activated.

13:58 UTC - ITF left locked in LowNoise3. All the subsystems worked properly, no particular problems or events to report.

Some details of the INF interventions:

On 18th March, I went in the Cleaning Lab (where is the tank of the ultrapure water) to check the leak of water reported by M. Perciballi. Once there I contacted D. Soldani for instructions. He instructed me to close all 5 valves in order to stop the leak. ~ 19.30 UTC I did it.

Following today's (Mar 20th) activities on the WE suspension, we could not relock OMC2; looking at the DARM spectrum I found out that the violin modes of the WE payload were quite excited, making B1 to saturate. We tried then to damp them with the already used mechanism and GUI interface:

• this is the first time we did that after the DSP->Acl migration of the ASC code, which reshuffled also the channels related to this issue;
• the GUI in the user menu is out of date for the same reason; we checked out and used the updated version from SVN, which we should also push in /virgoApp as soon as possible;
• we managed to lock up until LOCKED_OMC1_B1s2_DC, where we did the damping;
• we did not do an extensive campaign to damp everything, but just the main offenders (2, 4 and 6, see Figure); 4 and 6 damped quite quickly (at least to reasonable levels) with standard filters, low gains and actuating on TX; unfortunately we did not manage to damp mode 2, with either of the two filters, either sign of the gain and trying also to actuate on TX or Y degrees of freedom;
• after this, although the modes were still quite high, we managed to relock in LOW_NOISE_3 and started the work on TCS.

Tonight we worked at the optimization of the TCS working point, tuning accordingly the ITF alignment working point step by step. Hereafter a short summary of what we did (a longer and more detailed log will be posted tomorrow):

• started with the TCS OFF and locked the ITF in LN3. The BS loop was left open.
• switched on both the DASes at lower values with respect to the previous ones. The BS loop was closed after we found a good working point
• increased the power on the WI DAS 
• switched off the WI DAS
• switched back on the WI DAS and increased the power on the NI outer ring

We optimized the alignment working point by maximizing the sidebands on B4 and minimizing B1p. The new values of the offsets have been saved in metatron. The CMRF seemed to improve a bit when the NI outer ring was switched on, but it was not possible to find an alignment working point minimizing the CMRF at the same time.

The shift started with the ITF unlocked, and WE loops open; the closing of the loops took some time because one accelerometer was not working well. Valerio was able to fix the problem.

After that, at around 16:00, we started to lock with the aim to test the lock acquisition after the activity active common mode noise subtractor.

Unfortunately, we encountered two problems:

1 - lock of the cavity; fixed by Diego deactivating the IB automatic alignment.

2 - pass LOW_NOISE_2; after investigation, we discovered that the WE violin modes had been excited.

After some time we were able to dump them and to reach LOW_NOISE_3 at 19:52 UTC.

At 20:00 UTC started the commissioning activity ITF/TCS tuning; activity still in progress at the end of the shift.

INJECTION

IB left in drift control; IMC_AA red flag on DMS.

SBE

from 14:46 UTC to 15:56 the SWEB position loop has been open because the WE loops were open.

DOWNTIME EVENTS (time in UTC)

- from 15:00 to 16:00, ITF in troubleshooting, because one WE IP accelerometer was not working well; Valerio fixed the problem.

- from 17:00 to 19:37, ITF in troubleshooting (from 17:27), because the ITF was not able to pass LOW_NOISE_2 due to WE violin modes excited; Diego was able to dump them.

Today I noticed that the 8.65 Hz resonance was rising up, making the lock difficult; I disabled the full-bandwidth alignment in the INJ_MAIN node and the resonance disappeared. The IMC_AA flag is therefore red in the DMS.

The test (active common mode noise subtractor) has been done this morning, but had no visible effect in 50Hz and 150Hz lines.

We have then decided to roll-back everything (END mirrors as well INPUT mirrors); now the situation is the same as before Feb 22nd (see entry 44973)

This morning the ITF was locked in Science Mode, the shift was characterized by an high wind activity who, sometime, prevented the achieving of LowNoise 3. However the work scheduled (SUSP - Test 50 Hz noise eater) has been carried on without troubles.

Suspension - 10:02 UTC - during the Suspension work at the NE the guardian opened "F7_DC_ON", I properly restored the standard working condition.

12:40 UTC - ITF manually unlocked in order to undo the electronics work for the 50Hz done for the input and end mirrors.

We tuned the threshold of the squeezer DMS Line in order to avoid red flags when the subsystem is not required. 

PLL

• Internal PLL: can be red in each ITF state to notify problem in the Squeezer rest position
• External PLL: the flag can be red from ITF in LOCKED RECOMBINED

Squeezer

The 4 sub-flags can be red from LOW NOISE 2 

SQZ_AA

• the first 5 subflags: SQZ_AA_SLOW__AA_LOCK_SWITCH, SQZ_AA_SLOW__CamFarX_ERR, SQZ_AA_SLOW__CamFarY_ERR, SQZ_AA_SLOW__CamNearX_ERR, SQZ_AA_SLOW__CamNearY_ERR can be red from LOW_NOISE_1
• the last 4 subflags: mean_SQZ_MIRROR1_X_cmd_1kHz_50Hz_60, mean_SQZ_MIRROR1_Y_cmd_1kHz_50Hz_60, mean_SQZ_MIRROR2_X_cmd_1kHz_50Hz_60, mean_SQZ_MIRROR2_Y_cmd_1kHz_50Hz_60 can be read in each ITF state

Shutter and SQZ_INJ

the flags can be red from LOW_NOISE_3

A remark, when the CMRF of the ITF is not well tuned (a step above 3.5 kHz in hrec) the subflag brms_LSC_DARM_20_2_2600_2800 of SQZ_INJ can be red even if we are correctly injecting vaccum squeezed state with the right CC phase

Figure 1 shows the ITF optical gain before installation of the two B1 photodiode, and Figure 2 after installation of the two new HQE B1 photodiode.

There is a factor 2 difference, which I guess is due to some mistake in the ITF gain computation that didn't take into account properly the change from 1 to 2 photodiodes. Correcting for that the optical gain seems to have changed from 2010e6 to 2170e6, or an increase of 8%. Given that the optical gain scales as the square root of the detection efficiency, this correspond to a 16% increase in the detection efficiency. So the old photdiodes had a quantum efficiency of ~83%, and the improvement with then new HQE photodiodes is slightly larger than expected. Of course, looking at these data the optical gain can change by 5%, probably due to misalignment and other working point issues. But it is highly unlikely that this magically improved while changing the photodiodes, so the improvement should be due to the new PDs.

I switched back the PCals on around 5h31 UTC.

After the ITF working point tuning i switched on the squeezer and I tuned it:

I increased the AA dither line by a factor 5 and I found the AA working point. New set points

• CamNearX = -260
• CamNearY = 70
• CamFarX = -80
• CamFarY = 322

I performed a phase scan from 0 to 3.14 rad with a ramp of 600s and I found that the maximum squeezing level correspond to the minimum of the magnitude subtracted by 0.25 rad (attachment 1).

I retuned the CC dither demodulation phase. The new CC parameters are

• sqz_b1_phase = 0.667

• phase_dither_phi0 = 5.3

• phase_diter_set = 0.16e-3

I measured about -2.6 dB of squeezing in HREC both in the 1350-1750 Hz and 2600-2800 Hz bandwiths (attachment 2 normalized plot)

I left the squeezer switched on with the slow AA closed on dither lines and SLOW CC closed with the PyALP based servo loop

Attachment 3 hrec fft with (blue) and without (magenta) squeezing injection

Looking at the Cm connections on the DAQ servers I found an unexpected server connected to the FbmFE and asking to receive the frames. It was the server ingv_plg launched by giuliob and running on VTF host

•   -text CmName ingv_plg, port 29002,  Host cmplvtf2.virgo.infn.it, owner giuliob, socket_id 25 (16384,425984),  local 0, alive 1

Only DAQ experts can connect on the frame mergers ie FbmFE , so I  killed the ingv_plg Cm connections .

There is a VirgoOnline Test VPM server to perform these  kinds of tests

Today at 17h34 UTC, I swapped the permanent lines frequency on the PCal WE and NE 60.5 Hz <--> 63.5 Hz just for a few minutes, in order to cross calibrate WE Pcal with NE Pcal.

At 17h40 UTC I swapped them back to standard configuration.