Today we had some problem in relocking the ITF due to some SSFS fault. This was caused by the LNFS, after fixed this issue ( 16.04 UTC just put LNFS in external mode), we menage to relock the ITF at LOW_NOISE 3 and the calibration shift finally started (see the dedicated entry for details). ~18.00 UTC Calibration activities ended so I stopped the Commissioning mode from ITF status and as agreed with the crew I relocked the ITF in LOW_NOISE 3 (18.48 UTC). 20.14 UTC ITF unlocked and relocked at 20.28 UTC in LOW_NOISE.
ITF left in relocking mode LOW_NOISE 3 as requested by the crew
Vpm
18.53 UTC FbmStDy restarted after a crash
Modification of the calibration lines around 60 Hz and 355 Hz + addition of calibration lines with PCal
This is done in order to have frequencies multiple of 0.125 Hz to test Hrec with lower latency (test will be done with shorter FFTs of 8 s)
13h55 UTC, restart of CALnoise and update of the Hrec configuration
| BS | WE | NE | PR | WE_PCAL | NE_PCAL |
| 35.0 | 36.5 | ||||
| 61.0 | 61.5 | 62.5 | 63.0 | 60.5 | 63.5 |
| 356.0 | 356.5 | 357.5 | 358.0 | 355.5 | 359.0 |
A small line at 1012.5 Hz has been added to NE and WE calibration signals. Its amplitude have been set to be seen in SC_MIR_Z_CORR channel, but not in DARM. The goal is to monitor the injection path from Acl the to DSPs (in particular the delay) without being seen in DARM. This line will be monitored by the process TFMoni.
Injections with the PCals
PCal at NE and WE switched on at 14h10 UTC, with their three calibration lines. Note that their frequencies will be updated in the coming months for h(t) monitoring purpose.
On the NE and WE cameras, the small PCal beam spot is visible close to the center of the mirror when the main beam is switched off. The bright spot seen on top-right is an image of the PCal injection bench due to diffuse PCal light (1047 nm).
- 14h25 start some injections of lines with PCal and then with end EM actuators (this will be analysed to calibrate the EM actuators with respect to the PCals) (10 Hz to 200 Hz)
- 14h35 start injections from 10 Hz to 500 Hz
- 14h40 start injection of a single line around 80 Hz to compare results with previous measurements where a series of lines were injected
Note: NE PCal photodiode was not switched on during the measurements. Only the WE injections can be analyzed.
More usual calibration injections
- 14h45 to 15h measurement for calibration transfer from mirrors to marionettes.
- 15h manual unlock.
- 15h02 measurement of CAL to Sc delays.
- 15h05 start to relock with NI, WI relays kept closed in HP mode. We had some issues with "SSFS synchro", which was probably resulted from the LNFS being free. It worked once Julia had relocked the LNFS on the timing system. ITF relocked around 16h25 UTC.
- 16h35 to 17h measurements for calibration transfer from input to end mirrors.
- Manual unlock
- 17h20 : Tune B1p phases at 6, 8 and 56 MHz and start free Michelson measurements.
Remarks
- Suspension DSPs: the delays from the CAL channels generated by Acl to the Sc channels sent back by the DSPs are the same for all the mirrors (714 µs) as expected. But they are all different for the NE, WE and BS marionettes (between 980 and 1200 µs) while there where all the same at 915 µs in the past. Something wrong is probably on-going on the injection path to the marionettes.
- There is still aliasing of the CAL channels into the Sc channels (see Sc_NI,WI,BS_MIR_CORR around 17h27 UTC)
- Around 15h30 UTC, we found that the Vbias of PCAL_NE_PD2 had the wrong sign and fixed it. This is related to the replacement of the photodiode made during last week (6_9 August) to investigate the presence of a lot of 50 Hz lines in this photodiode. It is the same with the new one, so we will have to continue investigation. The offset of this channel must be retuned.
A good part of this morning shift was focused on finding a good working point for the alignment loops on LN3, including BS which is a bit more tricky. We have tuned the Diff+ and the PR by minimizing the power on B1p and maximizing the power on the sidebands senn by B4. INstead, for finding the good working point for the BS we have used as a figure of merit the CMRF. The scope was to find the offset that minimizes the CMRF, the region explored is shown in FIgures 1 and 2 (for TX and TY), together with different relevant probes.
It can be seen that for the TY degree of freedom the minimum of MCRF agrees with good values of sideband gain and a "dark" dark fringe. However, it is not the case for the TX. It seems that a good alignment in terms of sidebands/dark fringe i not compatible with a low CMRF. For the moment we have chosen a good CMRF, since the impact on the sensitivity is quite evident. Figure 3 shows the sentivity curve at the beginning of the BS loop tuning (yellow) and at the end (blue). The high frequency improves significantly, due to the improvement of the CMRF.
However, the fact that a low CMRF does not agree with a high value of the sidebands for vertical offsets of the BS suggests that we are missing something. One hypothesis was that we are trying to compensate with the BS an asymmetry between the arms so we are misaligning (bad behaviou rof the sidebands) to improve the symmetry of the cavities (good CMRF). One parameter to explore is the TCS impact on the CMRF. In order to check it we have plotted the CMRF for different TCS configurations.
We have used the data of tuesday, since the working point of the alignment was not changed during the whole shift. The steps indicated no the x-axis corresponds to the configurations of the TCS shown in the entry 42442. The point 0 is the one at the beginning of the shift. Figure 4 shows the CMRF, the sidebands power, the power on B1p and the cavity power for each TCS configuration. It can be seen that in this case the CMRF changes for each configuration, but at each point good CMRF is consistent with good recycling gain and good dark fringe. It can be seen that for different TCS configuration the CMRF changes up to a factor 2.
We could try to explore the dependency of the CMRF with respect to the BS and the TCS, trying to make these paraemters to converge.
ITF found in LOW_NOISE_1, the lock lasted since yesterday night. For the planned activity of the day ( ITF tuning ) we tried to reach LOW_NOISE_3, yet the OMC1 weren't able close the loop. I called M. Was (DET Oncall) and thanks to his help we were able to reach the locking state. From 7:40 UTC M. Mantovani and J. Casanueva took care of the tuning, then from around 10:30 UTC L. Aiello worked on improving the sidebands.
After an unlock at 11:54 UTC we faced, like yesterday afternoon, some issues during the lock acquisition related to the SSFS. Also, we had again problems with the OMC1 and we needed M.Was intervention.
ITF left unlocked.
Injection
Under the request of G. Pillant at 10:35 UTC we switched the LNFS to Internal PLL via VPM. The procedure was executed in LOW_NOISE_3 without unlock.
The attached figure shows the bahaviour of B1p, 56MHz sideband gain and B7 B8 during the nigth (up to t=11000s TCS tests were ongoing).
The sideband gain was stable around 8 to be compared with G=6 with TCS OFF (see entry 42425).
At 17.42 LT the ITF has been locked in low Noise 1 with TCS ON (Pin/out=[450-900]mW on the NI; Pin/out=[200-400]mW on the WI). We waited for the steady state of the YAG thermal lensing before starting the DAS tests (19.31 LT).
We applied the same powers of the STEP 6 during the shift of 14th August (Pin/out=[350-700]mW on the NI; Pin/out=[200-400]mW on the WI; see entry 42442) since we were not sure of the increase of 56 MHz because the signal was affected by large fluctuations.
The measurement performed tonight confirmed the increase of the 56 MHz sideband gain already observed in the previous shift and also an improvement of the dark fringe (see figure 1).
We were forced to stop the measurement for 3 hr because of the earthquakes in Molise and Japan (see operator report entry).
At 22.22LT a new stable lock has been achieved. We started a new measurement, STEP 7, by applying Pin/out =[100- 200]mW on the WI and Pin/out=[350-700] mW on the NI.
A slightly increase of the gain has also been observed.
Thus we leave the ITF with these DAS powers.
The attached figure shows the bahaviour of B1p, 56MHz sideband gain and B7 B8 during the nigth (up to t=11000s TCS tests were ongoing).
The sideband gain was stable around 8 to be compared with G=6 with TCS OFF (see entry 42425).
We noticed that the 2 photodiodes on B4, PD1 and PD2, show a different behavior when the TCS is turned on (see figure 1).
At the DAS thermal lensing steady state the 112MHz measured by PD1 is higher with respect to the value with DAS off, while PD2 measures the same value when no DAS applied. PD1 is the photodiode that we usually consider to evaluate the gain.
We could also notice a decrease of the 56 MHz sidebands amplitude probably due to some tweaking of the alignment (see figure 1).
M 5.3 - 5km SE of Palata, Italy
Time 2018-08-16 18:19:05 (UTC)
the itf had been easily relocked unfortunately these others two big earthquake unlocked it again:
M 6.0 - 263km SE of Iwo Jima, Japan
Time 2018-08-16 18:21:29 (UTC)
M 6.4 - 252km SE of Iwo Jima, Japan
Time 2018-08-16 18:22:52 (UTC)
in this case the effect lasted about ~40 minutes, the earthquakes and the others difficulties in the lock acquisition, most of them linked to the SSFS, preventing to relock for a while; finally we achieved a stable LOW_NOISE_1 lock at 22:20utc.
I left the itf locked in LOW_NOISE_1 state, the TCS measurements will go on during the night...
ITF found in LOW_NOISE_1. The morning shift was dedicated to the OMC lock acquisition speed optimization, carried out by M. Was, and remote PLL enabling/disabling of the LNFS, carried out by F. Carbognani. From 13:55 UTC G. Pillant, M. Mantovani, and J. Casanueva started working on the fmoderr setpoint tuning, activity still in progress.
VPM
InjMoni restarted at 10:11 UTC under the request of G. Pillant to add a new flag under RFC, mean(BsX_ML_TH_CORR_50Hz,10), used to check the state of the thermal setpoint of the laser.
* restored a good speed for the OMC lock acquistion,
5 minutes in total = ~2 minutes to lock OMC1, ~1 minute for DARM hand off to DC signal, ~2 minutes to lock OMC2
The part of full ITF lock acquistion due to OMCs should drop from 50% during latest commissioning run to ~25%.
Increased the temperature ramp speed to what was before the July intervention (0.3 degree amplitude, 0.0005 Hz frequency). So a factor 2 faster for OMC1 and 2.5 for OMC2 than in the past few weeks.
Adjusted the Peltier locking filter gain to be a factor ~2 below oscillation.
Reduced in ITF_LOCK.ini omc1/2_therm_ramp to 5 seconds (from 15 seconds for OMC1 and 10 seconds for OMC2). The automation of the OMC lock should be detecting properly any thermal transient in the OMC, so this shouldn't be needed anymore.
Switched on the SDB1 picomotors. Used the two folding mirrors in front of OMC1 to realign the beam while looking at OMC1 locked on the order 2 carrier mode. This improved the alignment from order 1 and 0 TEM modes of same height, to order 1 mode 5 times lower than order 0 mode.
Switched the SDB1 picomotors off.
Relaxed the condition on OMC temperature stabilization precision before starting a scan in the automation from +/-3mK to +/-5mK.
09:24 UTC locked in LOW_NOISE_3, the OMC2 lock switched to quickly to full bandwidth control with PZT
09:36:40 UTC unlocking ITF on purpose, to check OMC lock again.
Figure 1, Relocking OMCs at 09:51:30 UTC, OMC1 locked at 09:53:30 UTC, OMC2 lock starts at 09:54:25 UTC, OMC2 locked at 09:56:10 UTC
relaxed the OMC saturation check to ensure PZT correction is in +/-7 V instead of +/-5 V
Figure 2, 10:14:40 UTC OMC2 lock ramp starts, 10:16:45 UTC OMC2 locked
Observations:
Before the installation of the monolithic suspensions, the phase camera images suffered from a second beam both on B4 and on B1p during the initial locking stages. After the installation these spurious beams vanished. Figure recombined.png shows the situation of PC B4 and PC B1p for the interferometer locked in recombined, before and after the installation of the monolithic suspensions. Additionaly the beam shape on the PC B4 during dark fringe has changed from round to astigmatic. No drastic change in astigmatism are observed on B1p (see Fig. df.png). The alignment of the PC B4 telescope has been changed since the installation of the monolithic suspensions, while the telescope on B1p is not touched (see entry 41414).
Hypothesis:
The second beam on PC B4 during recombined (in the past) might have been caused by the BS just after the view port, that splits the beam between SFP and PC (the mount was broken and Matteo found during the realignment that it was clipping, at least after the replacement of the suspensions). As the beam leaving the view port is astigmatic, the ellipticity we observe now could be explained by a change in the distance between the two PC telescope lenses after the last alignment. The cause of the second beam we did not investigate. To check whether the PC telescope has changed, we remeasured the beam profile on 15.08.2018.
Beam profile:
Indeed the beam profile of B4 at the level of EPRB (outside the SPRB vacuum) has changed. Most likely cause is the PC telescope itself, as the beam on the SFP path looks the same as on the SPRB camera and both have not changed after the installation of the suspensions. Changing the distance between the two lenses by 9 mm explains the observed horizontal profile (see Fig. b4beamprofile.png for comparison). We decided not touch the telescope for now, as the interferometer was very unstable with a person on the BS tower (which is needed for the realignment due to the location of EPRB) and a change in telescope corresponds to a change in image point, but not in HOM content (which is the quantitiy we are currently most interested in).
Future action:
We will discuss a redesign of the telescope.
The amplitudes and powers of the PC signals are not stable on time scales of 10 min, and most definitely not on time scales of hours. As the effects are the same for all sidebands and the carrier, the DC of the corresponding main diode has been used in the past to correct for the drifts in the _PWR channels, but this situation is not satisfactory. To debug the problem we added new channels to dataDisplay:
1.
2.
In the attached Figure a comparison of the PC and the main diode signals is shown, observations so far:
1. RF channels containing the reference beam do not follow the main diodes
2. RF channels not containing the reference beam do follow the main diodes
3. DC channels do follow the main diodes
4. The electrical reference is stable
Point 1 is the problem to be solved.
Point 2 seems to indicate that it is not an electronic problem, or interferometer beam problem.
Point 3 shows that it is not the reference beam amplitude.
Point 4 shows that it is not an unstable reference beam frequency (reference beam is at 80 MHz).
Remaining suspects are:
- Reference beam polarization
- something else we did not think of?
We will contact Eric to help with further investigation of the reference beam.
The new version (v2r0) has been made default in VPM and the server restarted from there.
To be tested at the first occasion.
The shift was dedicated to the Phase Camera Alignment and signals check carried out by Schaaf and Beuzekom.
Upon arrival it was impossible to close the loop of the RFC. Under the request of the INJ Oncall I switched the LNFS from Internal to External PLL and after this change we were able to proceed with the lock.
At 7:20 UTC the ITF reached PRITF _MICH_OFFSET_0_1 and the activity on EDB started. Later the work moved on the EPRB1 in LOW_NOISE_1: in this phase the ITF was unstable and we had several unlock probably related to the human presence near the towers. At 14:25 UTC the activity was concluded and Majorana started working on the Local Controls. Activity in progress.
VPM
ITF_LOCK crashed at 6:58 UTC, properly restarted.
Today activity has been devoted to the tuning of the DAS powers with the ITF locked in Low Noise 1.
In the first part of the afternoon Alessio and Viviana performed a new calibration of the DAS powers, because the ADC offset changed after the timing upgrade.
After that, Julia relocked the ITF in Low Noise 1 with the TCS on (P in/out=456/912 mW on NI and P in/out=233/466 mW on WI). After the ITF reached the thermal steady state, we started to change the DAS power applied in order to maximize the 56 MHz sideband gain.
The different annuli powers applied are summarized in the following table.
| Step | P NI (IN/OUT) | P WI (IN/OUT) | Time (LT) |
| 1 | 450/900 | 450/900 | 20:43 |
| 2 | 900/1800 | 450/900 | 20:55 |
| 3 | 600/1200 | 450/900 | 21:20 |
| 4 | 600/1200 | 200/400 | 21:45 |
| 5 | 450/900 | 200/400 | 22:25 |
| 6 | 350/700 | 200/400 | 22:54 |
The behaviour of the 56 MHz gain, B7, B8 and B1p are shown in Figure 1 and 2.
At 23 LT Julia checked the alignment and improved the position of the BS and of the PR, but few minutes after the intervention the ITF unlocked due to the earthquake in Molise.
We left TCS on with the DAS powers of step 5, that seems to maximize the 56 MHz gain.
The fine tuning will be finalized in the next shifts.
We could also notice a decrease of the 56 MHz sidebands amplitude probably due to some tweaking of the alignment (see figure 1).
The firts part of shift has been dedicated to the ITF recovery. At about 18:30 UTC the ITF has been locked back in Low_Noise_1, therefore the TCS activities started as planned.
21:00 UTC - Commissioning Mode stopped, ITF left locked in LowNoise1.
All the subsystems worked properly, no particular problems or events to report.
This morning, I found the ITF locked in LOW NOISE 1.
At 06:30 UTC started the weekly maintenance:
- standard vacuum maintenance and cryotraps refill;
- cleaning of the CB;
- standard check refill of the TCS chillers.
Maintenace finished at 10:30UTC.
In parallel, R.Gouaty worked on SIB2 and SDB2 offset quadrants, A.Masserot on the Minitowers local controls and G.Pillant on IMC.
Also the TCS started with the atcivities planned for the afternoon, and still in progress.
VPM
12:51UTC, INJ_MAIN and ITF_LOCK processes restarted after a crash.
This morning the residual offsets on the SDB2 quadrants DC channels were checked:
B5_QD1 H -4.27046e-05 V 4.80133e-05 Sum -3.26598e-04
B5_QD2 H -1.64740e-04 V 1.71002e-04 Sum -7.35306e-06
B1p_QD1 H -1.77353e-02 V 7.63090e-06 Sum 7.06499e-05
B1p_QD2 H 4.57601e-05 V 7.29158e-05 Sum 6.02074e-05
We tried to correct all offsets exceeding 1.0e-4. This was successful on B5. On B1p_QD1 the Horizontal channel still has a residual offset of 6.1e-04 which cannot be corrected due to the large offset fluctuations on this channel.
The LNFS was set to internal PLL instead of external PLL (10MHz ) today at 10h31m48sUTC by Gabriel.
This plots shows the transition, there is a jump on all deltaFreq channels
- blue: 20180814-11h50mUTC
- brown: 20180813-05h00mUTC
- purple: 20180807-05h00mUTC
The last plots shows the reduction of noise and on the correlation when the LNFS is not slaved
This morning we recentered the B2 beam on the SIB2 quadrants.
First the residual offsets on the B2 quadrants DC channels were measured:
B2_QD1 H -6.20809e-05 V 1.70932e-04 Sum -1.25085e-04
B2_QD2 H -8.14457e-05 V 2.51887e-05 Sum -1.04111e-04
After correcting these offsets, one could measure:
B2_QD1 H 7.41747e-06 V -1.95561e-07 Sum 2.28872e-06
B2_QD2 H 8.06733e-07 V -1.48503e-05 Sum -1.55439e-05
Then the PR mirror was aligned in order to have the B2 beam properly aligned on SIB2. It was then possible to adjust the beam centering on the B2 quadrants by acting on the picomotors B2_M7 (+150 steps in H and +500 steps in V) and B2_M9 (-10 steps in H and +110 steps in V).
The LNFS was set to internal PLL instead of external PLL (10MHz ) today at 10h31m48sUTC by Gabriel.
This plots shows the transition, there is a jump on all deltaFreq channels
The last plots shows the reduction of noise and on the correlation when the LNFS is not slaved
This morning, the offsets of the SDB1 B5 quadrants were checked and updated.
From Fig.1 one can observe:
- before 06h39 utc the IMC is stil locked and one can observe some signal on the SDB1 B5 quadrants (B5_QD1_sum fluctuates around 11-12, and B5_QD2_sum fluctuates around 13).
- at about 06h39 utc the IMC is unlocked, and one can observe residual offsets on the quadrant channels.
- at 06h42m46 utc the old offset corrections are removed from the SDB1_Quadrants config file and one can measure the total offsets by reading the B5 quadrants channels (see zoomed plots on Fig.2 for a better view).
- at 06h53m38 utc, the configuration of the process SDB_EDB_dbox_rack is reloaded after correcting these offset values. All channels of the quadrants are then equal to 0 (as expected from the offset subtraction), including the sum channels.
- at about 07h11 utc, the IMC is locked again. One can see that the offsets on B5_QD2 had an almost negligible impact on the sum channel. However the DR, UR, UL channels have changed by about a factor 2, compared to the situation before offset correction.
It was then possible to reengage the bench drift control without problem.
Checking the spectras with only the test beam using raw ADC data and no scanning:
- Spectra are flat at frequencies above 20 MHz with exception of the expected peaks at 56 MHz and 112 MHz.
- The Noise levels agree with expectation.
- Below 20 MHz there are structures between 7 MHz and 9 MHz, as well as between 1 MHz and 5 MHz. Similar structures are visible both on the B4 and the B1p PC. They are visible with blocked beam but do not show up if the input to the ADC is disconnected, origin is under investigation. Note that these peaks are well below the lowest demodulation frequency for the phase and intensity images (as well as derived quantites).
- On B4 a 20 MHz peak is visible, we do not know whether this peak is expected?
(- side remark: the 6 MHz peak does not show up in the PC data on B1p because during DF it did not fall on the diode)
Figure showing the mentioned observations is attached as rawadcdata.png.
Remarks on the raw demodulated signals:
The signals obtained with the PC follow the trend of the main diodes. However, there are discrepancies which could be caused by a moving beam spot (the diode of the PC is much smaller than the beam spot) and are therefore of limited use (unless integrated over a scan which we did not do). Figure showing the mentioned trend is rawdemodulated.png.
