guard tours(UTC):
5:15 --> 5:50
7:05 --> 7:35
9:15 --> 9:40
11:05 --> 11:40
ITF kept the lock for the whole shift. BNS Range ~46 MPC. The weather condition were worsening therefore during the night I put the GIPC in the wind configuration.
Enviroment / windactivity
from 02:51UTC to 04:23UTC - GIPC in "wind cfg", I activated the procedure for "moderate wind", from the PR Gains Client I put the SW_SW Gain value from 1 to 0 (Adjusting mode engaged during the transition).
Guard Tour (UTC)
22:28 - 22:59
00:24 - 00:54
02:18 - 02:49
03:56 - 04:27
ITF found in Science mode.
It unlocked at 14:09 UTC because of an earthquake of 5.8M in North Coast of Albania occurred at 14:04 UTC; no possibility to switch in "earthquake configuration" because any alert notification was received (earthquake close to Virgo site and published on the usual WEB sites some minutes after the unlock).
ITF relocked without major trubles, Science mode set at 14:52 UTC; still in Science mode at the end of the shift.
Guard tours (times in UTC)
- from 13:10 to 13:40
- from 15:15 to 15:45
- from 17:15 to 17:45
- from 19:15 to 19:45
SUSP
yellow flag for Sa_MC_F0_COIL_H1 indicating coil close to saturation, Valerio has been informed by mail,
By means of the NonNA line tracker tool, I reconstructed the frequency evolution of the wandering line between September 17 and September 18: refer to the attached median normalized spectrogram and the red dashed line on it. Then I correleted the corresponding frequencies time series with all the ENVironmental channels. A very high value (~98%) of the Pearson correlation coefficient has been found with some accelerometers of the PR and SPRBLINK.
Here the NonNA report with the plots of the 10 most correlated channels. Take also a look at the log file with a longer list of correlated channels.
-guard tours(UTC)
5:25 --> 5:55
7:10 --> ?
11:05 --> 11:40
SUSP
At 8:25UTC the WE local control opened by the guardian following the ITF unlock, properly closed.
ISYS
(21-09-2019 07:00 - 21-09-2019 09:30) PMC realignment.
Yesterday night a drift of the power transmitted by the PMC had occurred. This has made the IMC input power much noisier and in turn the strain signal very glitchy. The recover of the power injected to the ITF via tapping in the reservoir of IPC1 (see 46980) was only effective in easing the life for the locking configuration, but of course it did not change the dynamical behavior - nor was it expected.
As reported, this drift was a due to a worsening of the throughput of the PMC cavity, so after discussion with Frederic Cleva, it was agreed with the commissioning coordinator to try and realign the beam sent to the PMC "ASAP".
So this morning I came on site, discussed with the operator on duty and, after the green light by both th the run coordinator and the commissioning coordinator, I went to the laser lab to perform the operation.
What I did [just for the record, you can skip itif you want the conclusios]: I
TF node down, INJ node down, metatron paused, BPC with DC ON mode, scanned the PMC to block the beam on the laser bench, once the beam blocked via the flip mirror, changed the lock trigger of the PMC with the trasmission of the PMC, then locked again the PMC, then obesrved on the beam dump the shape of the transmitted beam to spot some singature of misalignment. I realized the TEM00 is too intense for tthe TEM01 to be visible, so I switched to scan mode again and selected a scanning ramp so to have only the misalignment mode in the range. Tried to minimize it by looking at the peak intensity, but found out that looking at it on the beam dump and trying to make it as close to a laguerre as possible wa much more effective. The transmitted power of the PMC was larger and much more stable then before.
Then I restored the nominal operation for the laser bench and the BPC and had a look to the trasnmitted power of the IMC. It was larger than yesterday as expected, so I adjusted it to the value we set yesterday.
In the attached plots the effect of the operation:
1) PMC trasnmitted power before and after the realignment (more stable and larger value than before)
2) Range over time before and after the operation: too early to assess the glitchiness has gone, but seems better
PS: Now I do not remember anymore if I restored the trigger of the PMC to the quadrant on the EIB2, it could be still the transmission of the PMC. If this is the case, the hardware safety against the EIB going crazy is OFF. If oscillations occur until monday morning, please call immediately INJ on call. I will check on monday morning and, if it is the case, unlock and restore the safety.
During the troubleshooting, an oscillation of DARM apparently due to WE roll resonance at 9.7 Hz has been observed. It is an old problem of the related damper, but it was fixed sometimes ago. The damper has been disabled in order to avoid major problem, and for that reason there is a little more noise at 10-12 Hz. On monday we will try to put it back in operation.
Upon arrival, ITF was relocking. Easily relocked. Science Mode set at 21:21 UTC with BNS Range 46 Mpc.
Many drops in the BNS Range. Probably it's due to the microseismic of the sea activity.
ITF unlocked at 03:07 UTC. After several attempts ITF back in Science Mode at 4:47 UTC.
Guard Tour
21:35 UTC - 22:15 UTC
00:38 UTC - 01:08 UTC
02:30 UTC - 03:02 UTC
03:58 UTC - 04:25 UTC
SUSP
The unlock of 03:40 UTC opened WE LC. Properly closed.
The ITF unlocked twice during the shift (13:53 UTC and 20:42 UTC). During the afternoon it relocked at the first attempt (Science Mode engaged at 14:26 UTC), BNS Range ~44 MPC affected by many drops.
21:00 UTC - lock acquisition in progress.
Guard Tour (UTC)
17:28 - 18:04
18:52 - 19:17
An activity started a couple of week ago, has come to a conclusion: the arm angular control filters currently running seems to meet expectations. The new controls are supposed to have about the same accuracy, and it is like that (fig 1, fig 2). The main difference visible in the plots regards a big peak at 200 mHz in COMMp_TY, and partially in NAm WAm (but it is just a coupling with COMMp. It is not due to a worsening of the control, but to an external problem, which will be expained later.
The new design has been developed in order to reduce the control noise above 10 Hz, and a certain result has been achieved, as one can see in the control corrections (fig 3, fig 4). More in detail, the loops which do not need a big accuracy have been improved a lot, in terms of noise reduction. On the contrary, for DIFFp, which requires the best possible accuracy, the noise has been reduced just a little, and now it is dominant on the budget of the corrections on each mirror (fig 5, fig 6).
The current projection of the total Arm noise on h is shown in fig 7.
The problem of COMMp_TY is in fact a problem of SR longitudinal control. In fig 8 the motion of SR suspension, measured at the level of F7, is compared to a the same dof of NI. SR motion produces a steering of the beam out of the ITF, visible on COMMp_TY sensor. Fig 9 shows the coherence between the two signals.
SR problem is not always there: sometimes the suspension is quiet as the others (fig 10). It is going like that since the beginning of the commissining, but a deep investigation has never been done.
ITF found in Troubleshooting mode since the previous shift.
At 5:30 UTC I contacted the ISYS on call to inform him that transmitted input power had jumped to 17.7 W.
We decided to recover first the transmitted input power and then face the locking troubles.
At 7:00 UTC Antonino started to work in Laser Lab and at 9:00 UTC the transmitted input power was back to its standard value (~19.5 W), see entry ITF input power tuning.
At 9:00 UTC we started the investigation about the locking troubles; in some way we reached LOW_NOISE_3_SQZ at 11:53 UTC, Science mode set at 12:02 UTC.
SUSP
WE LC open by the guardian at 6:19 UTC, 9:26 UTC, 11:03 UTC after ITF unlocks; properly closed.
Other
(19-09-2019 21:00 - 20-09-2019 12:00) The ITF unlocked at 21:00 UTC yesterday evening and it started to be unstable unlocking at various states; at the same time the transmitted input power jumped to 17.7 W.
ISC
(19-09-2019 22:00 - 20-09-2019 02:00) From remote
Status: Ended
Description: Locking troubles
(20-09-2019 06:00 - 20-09-2019 12:00) On site
Status: Ended
Description: locking troubles
ISYS
(20-09-2019 07:00 - 20-09-2019 09:00) On site
Status: Ended
Description: The transmitted input power jumped to 17.7 W.
Actions undertaken: Standard power recovered by Antonino
During the troubleshooting, an oscillation of DARM apparently due to WE roll resonance at 9.7 Hz has been observed. It is an old problem of the related damper, but it was fixed sometimes ago. The damper has been disabled in order to avoid major problem, and for that reason there is a little more noise at 10-12 Hz. On monday we will try to put it back in operation.
During last night, around 22:00 LT, a slow drop of power brought the IMC transmitted power below the DMS threshold (see operator report) so this morning (around 10h00 LT) we anticipated the tuning of the input power foreseen to happen next monday. We used the EIB IPC1, axis 1 channel 1 with the manual box driver to slightly adjust the input power with a target value of 19.4W. Current value seems to oscillate a bit around 19.5W.
About the origin of this further drop, it seems related to a worsening of PMC throughput (see attached plot), while the Neovan amplifier does not seem to be blamed.
To be noticed that the power injected to the IMC seems now noisier than it was before the event of power drop.
Just an example of a 'low noise low gain' design: with an UGF ~ 5Hz it is possible to have about the same accuracy of the old filter, and a significantly lower correction (noise). The risk for a design like that is the lack of gain in the region 1-2Hz, which in certain environmental condition could become a problem. An adaptive strategy, at a very basic level, could allow a safe use of a low gain filter.
Sorry, I forgot I was on-call, my phone was in silent mode for the night...
Added the quantum noise of 120uW of contrast defect light measured on B1 to the noise budget.
Figure 1, shows that it is a factro 4.5 below the sensitivity curve.
ITF unlocked at the beginning of the shift (21:00 UTC).
This night, with Julia, after retuning a bit ITF working point and adjust the working point of the SSFS boost, we reached only the lock of LN1 but we have still problems to reach LN3 due to a saturation of OMC2 Corr.
Det OnCall not available, Det referent not available. Experts informed by mail. Run Coordinator contacted by phone.
Troubleshooting shift.
Guard Tour
22:28 UTC - 23:03 UTC
00:33 UTC - 01:00 UTC
02:30 UTC - 03:01 UTC
03:55 UTC - 04:26 UTC
ISYS
Mode Cleaner output power out of range. Red flag on DMS (plot1)
SUSP
The unlock of 02:10, 02:39, 02:58 UTC opened WE LC. Properly closed.
ISC
(19-09-2019 21:00 - ) Locking troubles
ISC
(19-09-2019 22:00 - ) From remote
Status: On-going
Description: Locking troubles
Sorry, I forgot I was on-call, my phone was in silent mode for the night...
ITF kept the lock for the whole shift. The planned activity on ISC stopped at 14:08 UTC (Science Mode activated). BNS Range ~44 MPC.
Guard Tour (UTC)
17:29 - 18:09
18:55 - 19:27
Fig 1: Model of the old and new filters (without the notches).
Fig 2: Expected change of the in loop error signal, calculated applying the model to data with the old filter running.
Fig 3: Expected change of the loop correction, calculated applying the model to data with the old filter running.
Fig 4: Actual change of the in loop error signal.
Fig 5: Actual change of the loop correction.
This is not yet an optimal design, even because it is not clear what is the required accuracy. If we realize that the current accuracy is plentiful, we can study a way to lower significantly the 'noisy' correction.
The second example of seismic glitch looks associated to a sudden and pretty loud seismic transient covering 5Hz to 20hz detected by ENV seismic sensors almost everywere in CEB, Figure 1. No sound detected.
Figure 2 shows a map of sensor indicating the two sensors shown by florent in Omicron plots, they are not co-located.
Interesting to know the rate they occur and if there is some periodicity.
Monday morning during the commissioning shift, the NI and WI heating belts were turned on at voltages of 10 V and 6 V resepctively (log entry 46932). One objective is to measure the time constant of the mirror temperature rise due to the heating belt (actuator of etalon loop). The time constant is important to the etalon control loop as this will dictate the choice of the UGF (unity gain freqency). It was immediately clear that the time constant would be on the order of days. The time constant is calculated as the coefficient on the exponent in the equation
where is temperature,
is time in seconds,
is the time constant, and
is the steady state temperature value.
During the measurement, the interferometer went down several times. During the down period the mirror temperature dropped because the laser was no longer heating the mirror. We determined that the days long data collection was not compromised during these down periods, granted that the data is removed when used for the fit of the time constant. The temperature of the mirror has two main heat transfer contributors, laser radiation and tower radiation (due to heating band). The time constant of the laser was assumed to be much faster than the heating belt. Therefore, after the ITF went down, the temperature would return to the laser steady state value quickly.
The time constant of the laser heating was measured in the same method of the heating belt by finding a period in which the ITF was down and measuring the heat rise due to the laser. This period was on GPS = 1251694861 (2019-09-05 05:00:43 UTC) after maintenance. Figure 1 shows the NI and WI mirror temperatures during the unlock and relock. The grey boxes show the periods when the ITF was down. When the ITF relocked, the rise time was fit to the exponential equation to find the time constant. Figure 2 shows these data with the associated fit. The time constant is on the order of 1e5 seconds.
These fitted functions were used in order to remove data from the etalon time constant measurement which included thermal transients due to laser heating. During a down period the mirror temperature was assumed to have linear growth from the points prior to the unlock. The time of the laser tranisent upon relock was determined as the time it would take to reach 99.9% of the linear temperature value. The data points within this time period were removed from the fitted data. Figure 3 shows the timespan of the etalon time constant measurement. Again grey boxes are when the ITF was unlocked. The red points show the data that is removed from the fit due to the laser transient, unlocked periods, and a brief period at the beginning in which the tower was in a transient heating period. The calculated etalon time constant is on the order of 1e6 seconds, one order of magnitude larger than the laser time constant.
The ratio of the time constants is approximately the ratio of the voltages squared. This is expected as the temperature should go with the power of the heating belt. The steady state temperature based on the applied voltages may also be determined as the offset (C value in the exponential formula) in these fits.
Another objective of this commissioning activity is to measure the calibration of the applied voltage of the heating belts to the mirror temperature. Currently there are two data points for each mirror: the applied 10 or 6 V for the NI and WI mirrors and 0 V. However, since the temperature is a function of the power () a third data point is necessary to calculate the calibration factor. Therefore, this morning, both actuators (NI and WI) were set to 3 V in order to gain the last data point. This is obviously a low number of samples. Three points will be used as an initial implementation of the loop and the calibration will be evaluated again once the loop is in operation.
Today we have worked to improve the alpha substraction (MICH to DARM) in the low frequency region (10-20Hz) where the substraction was not properly working. I have modified the script to compute the filter (from MICH to DARM) in order to have a proper fitting @ 20Hz.
The alpha filter is really more simple than before (essentially 1 pole and 1 zero instead of 7 poles and 7 zeros for the old alpha). The improvement was visible at low frequency but there was a spoiling of the performance above 60 Hz. Thus the MICH corrector has been modified, the new filter TF will be shown in a comment to this entry, to have a better roll-off.
In Figure 1 the various alpha performance are visible:
- Purple : no alpha
- Green : old alpha
- Yellow : new alpha
- Blue : new alpha and new MICH filter
A noise injection has been done in order to further improve the current alpha
we will observe the loop in the following days
Fig 1: Model of the old and new filters (without the notches).
Fig 2: Expected change of the in loop error signal, calculated applying the model to data with the old filter running.
Fig 3: Expected change of the loop correction, calculated applying the model to data with the old filter running.
Fig 4: Actual change of the in loop error signal.
Fig 5: Actual change of the loop correction.
This is not yet an optimal design, even because it is not clear what is the required accuracy. If we realize that the current accuracy is plentiful, we can study a way to lower significantly the 'noisy' correction.
Just an example of a 'low noise low gain' design: with an UGF ~ 5Hz it is possible to have about the same accuracy of the old filter, and a significantly lower correction (noise). The risk for a design like that is the lack of gain in the region 1-2Hz, which in certain environmental condition could become a problem. An adaptive strategy, at a very basic level, could allow a safe use of a low gain filter.
ITF found in Science mode.
From 7:32 UTC to 7:33 UTC the commissioning team changed the heating belts working point without stopping the Science mode.
At 9:14 UTC ITF in Commissioning mode for the planned activity on ISC, still in progress at the end of shift.
SUSP
At 11:14 UTC the WE LC were open by the guardian after an unlock; properly closed.
One possible explanation of why Omicron did not trigger on the HI done at 1667 Hz and 811 Hz in the set of detchar HI of 2019-09-11-15:04 UTC is that loud low-frequency noise was generated during the injections, as can be seen on plots 1 and 2. And Omicron triggered on this noise. Plot 3 shows a zoom on this noise.
In all these plots, upper row shows the spectrograms of Hrec_hoft_raw_20000Hz (no noise+calib lines+HI subtraction) and Hrec_hoft_clean_20000Hz (only noise+calib lines subtracted) and the lower row shows the TF and Coherence plots between the injected signal (CAL_HI_DetChar_hoft) and Hrec_hoft_clean_20000Hz, iin the frequency band of the injected sine-gaussian signal.
The low frequency noise is not present for the other injections at 407, 211, 129, 73, 47 and 31 Hz. We will investigate the origin of it in the case of 1667 and 811 Hz injections.
For completeness, plots 4,5,6,7 show the same information for the injections at 407, 211, 129 and 31 Hz, where the TF module is around 1 and the TF phase is around 0, as expected.
For seismic noise hunters, let me post 2 glitch examples:
- The first one looks like an excited line @10.8 Hz. These glitches seem to be visible in VIM spectrograms: https://vim-online.virgo-gw.eu/resources/2019-09-18/resources/20190918_spectro_spectromedian_Hrec_hoft_20000Hz_0_30_604800.jpg
- The second example seem to be a different category of glitch. The disturbance is visible in both ENV_WI_ACC_X and ENV_IB_CT_ACC_X (are these probes co-located?). Moreover there is clearly some upconversion to h(t)
Another ENV channel identified by UPV targeting only higher frequency component of the seismic glitches: V1:ENV_IB_CT_ACC_X.
--> a veto channel will be generated (dead-time=0.005 %)
