Tonight we started the shift after the work performed to try the OMC lock.
The plan was to measure: 1 - arms losses, 2 - Gouy phase and 3 - Finesse
In order to keep the beam always centered on the B1p PDs, even when the North arm was misaligned, we opened the SDB1 B5 drift control and changed the offset in the following way:
LC_TX_set: from -155 to -100
LC_TY_set: from -950 to -870
After the measurements, the drift control has been closed, while we left the new offsets for the SDB1_LC.
1 - ARMS LOCK-UNLOCK
In the few hours left, we performed some lock-unlock measurements in order to measure the losses. As reported in #41555, we performed the measurements of the ITM reflectivity in three different conditions:
- cavity locked
- cavity unlocked (but still aligned)
- cavity unlocked and ETM misaligned (only ITM aligned)
GPS start West = 1221505594
GPS start North = 1221509643
(see figure 1)
2 - FREE SWINGING
North (6MHz with modulation depth at 5 dBm to allow the RFC to stay locked, 8MHz and 56MHz at -10 dBm)
GPS = 1221514610, 1221514818, 1221515051
8MHz modulation depth increased to -5 dBm to have one more reference in the FSR
GPS = 1221515254
West (same conditions as for the North)
GPS = 1221516268 and following
8MHz modulation depth increased to -5 dBm
GPS = 1221516863
3 - IMC FAST UNLOCK
We locked one arm at the time and unlocked the IMC in order to cut the input light.
The GPS can be extracted from figure 2.
The analysis of the data will follow.
We tried to engage the lock on LN1, but wehad 3/3 unlocks due to the usual SSFS ramp problem, therefore we leave the ITF in recombined.
Increasing the DARM offset did overcome some of the contamination and removed a second zero crossing. This allowed OMC1 locking (but only for very slow ramps).
OMC1 alignment should be checked to verify it is not the cause of the issue, otherwise the known slight astigmatism of the spare OMC installed last week is likely the reason.
A simple solution to be tested tomorrow is to use a double demodulation (first at 15kHz from the OMC length dither, then at 87.1Hz from the DARM line), which should remove the 56MHz order 1 mode from the OMC error signal. Hopefully the second OMC will filter out the leaking 56MHz order 1 mode from the DC readout signal.
This activity ended at ~ 17:40 UTC; after that Antonino and Annalisa started to perform some measurement for OptChar, activity still in progress...
VACUUM
Parallel activities carried at the terminal buildings in the afternoon.
Today we continued with the ITF recovery. Most of the times, unlocks were induced by engaging the SSFS boost.
Here's the summary:
- We've re-tuned the demodulation phases of B5_QD2_56MHz and enabled the PR loop;
- Changed the phase of the DIFFp error signal in such a way to flip the sign and enabled DIFFp loop
- Turned on the COMMp and BS angular loops. Before engaging this latter, Paolo twickled a bit the BS position by hand.
The interferometer is stable in LN1. The activity is currently focused on the OMC lock.
Today I found the ITF locked in RECOMBINED, the shift was dedicated to the recovery of the ITF (we were not able to reach the Dark fringe). People from LAPP worked from remote in order to solve these problems. 13.47 UTC we were able to reach LOW_NOISE 1.Recovery still in progress.
Vpm
6.21 UTC DET_MAIN restarted after a crash
According the Paolo entry, the trouble may come from the ramp used to perform the transition from the standard filter to the boost filter .
I put back the ramp of 10ms with a sine form ,
- at the first trial, the transistion to the boost succeeded
- at the second trial, the transition failed
Then I reduce the transition time to 5ms
- at the first trial, the transistion to the boost succeeded
- at the second trial, the transition succeeded too
The ramp was reduced and set to zero due to somes troubles observed and to some lack of knowledge .
According Paolo, the ramp is used to perform the switch between the 2 filters on the Sc_MC DSP too. As there is a delay between the SSFS rtpc and the Sc_MC DSP of 200us or 300us (2 or 3 10KHz cycles) the ramp must be slow compared to the delay.
According Paolo it could be useful to have the starting to the boost transition only when the error signal is close to zero
For the B7_PD2 the threshold was set to 10mA for a trial and we forgot to restored it to 50mA, is was done 2018-09-20-07h19m26.
Note that a reload Config is not enough to restore the threshold, today one need to reconfigure the mezzanine
For the B5_PD{1,2} and the B1_PD1 DC and Audio signal, there was some trouble with the Service mezzanine SN22 (SDB2_MezzPD1) while it was running with the firmware v3r1.
Restoring the previous firmware v3r0 allows to have the SDB2_MezzPD1 running correctly .We left the SDB2_MezzPD1 with this firmware (v3r0)
We repeated the test (moving the chiller's temperature by half degree), having a look at others signals: the effect (100Hz sidebands moving) is well visible in:
- CEB environmental magnetometers
- local magnetic probes (the one currently below SPRB)
- the electric ground noise probe (sensing the voltage drop between metallic towers body and racks in INJ EE_room)
- the UPS "sniffer" (a sample of the UPS voltage that supply the whole Virgo experimental area in CEB)
Worth noticing that:
- moving the temperature setpoint of both chillers affect the temperature in a different way (different time constant), allowing a better understand of who is who (which sidebands is related to which chiller)
- both chillers sidebands are visible in magnetic sensors, as well as in the ground electric noise probe
- only ONE of the two chillers sideband is visible in UPS sniffer
- nothing relevant is visible in IPS sniffer
Conclusions:
No definitive conclusion yet, we know where the problem begin (TCS chillers) and where it comes (DARM), but the path is unclear; could be a magnetic coupling, could be an "electrical" coupling (voltage drops un UPS line). For sure having this kind of signal (100Hz sidebands) well visible in a distant room (EE_room) as "voltage drop" between ground is not a good sign. We plan to repeat this test after all EE_room racks have been grounded (they are not, for the time being).
Many other unwanted signals seems present as sidebands around Mains harmonics, and these could have origin from other powerful devices (e.g. the beam-dump chiller) in EE_room, affecting DARM all togheter.
The fast shutter closing is due to the B1 PD2 ADC communication problem that mixes samples from B1 PD2 and B5 PD1/PD2.
This creates artificial flashes on B1 PD2 that trigger the fast shutter.
Figure 2:
B7_PD2 is systematically closing the shutter at 10mA of current draw. The configuration file correctly states that it should be closing at 50mA for B7_PD2 and at 10mA for B7_PD1. Probably the two have been mixed by PD firmware update.
The shift has been dedicated to the ITF recovery (entry #42737), but unfortunately we are not still able to reach the dark fringe. After the work on photodiode mezzanine firmware upgrade (entry #42733) we continued to collect unlocks during the acquisition.
VPM - Inframoni restarted in order to raise the upper treshold of "ENV..WAB_FiberStorage_TE" from 26 to 27 C. (following the request of the expert).
21:00 UTC - Commissioning Mode stopped, ITF left locked in Recombined Mode.
All the subsystems worked properly, no particular problems or events to report.
Today the work on the recovery of the ITF continued; the main points are:
- still no luck with the boost filter for the SSFS; a number of tests has been done (including reverting to an older, more tested configuration for the SSFS and a check on the MC DSP card) but we observed no change at all in the behaviour, so the origin of the problem is still not clear (entry 42736 may trigger some investigation on the ramps for the SSFS);
-
some problems occurred after the upgrade of the mezzanines' firmware:
- B1_PD2 is not usable at the moment, its configuration will be checked;
- most probably related to this, the logic of the slow shutter is not completely working as expected, but this is not a problem right now as the shutter is closed anyway, and it will be needed only after LOW_NOISE_1;
- later, it was not possible to get to dark fringe anymore: during the last transition, still well before the maximum circulating power we always reach, either the fast shutter closes (cutting the B1p beam, Figure 1) or the B7_PD2 one does (Figure 2).
The fast shutter closing is due to the B1 PD2 ADC communication problem that mixes samples from B1 PD2 and B5 PD1/PD2.
This creates artificial flashes on B1 PD2 that trigger the fast shutter.
Figure 2:
B7_PD2 is systematically closing the shutter at 10mA of current draw. The configuration file correctly states that it should be closing at 50mA for B7_PD2 and at 10mA for B7_PD1. Probably the two have been mixed by PD firmware update.
For the B7_PD2 the threshold was set to 10mA for a trial and we forgot to restored it to 50mA, is was done 2018-09-20-07h19m26.
Note that a reload Config is not enough to restore the threshold, today one need to reconfigure the mezzanine
For the B5_PD{1,2} and the B1_PD1 DC and Audio signal, there was some trouble with the Service mezzanine SN22 (SDB2_MezzPD1) while it was running with the firmware v3r1.
Restoring the previous firmware v3r0 allows to have the SDB2_MezzPD1 running correctly .We left the SDB2_MezzPD1 with this firmware (v3r0)
Today the ITF unlocked every time the SSFS boost was activated, except for one trial. Around the gps 1221332726 the ITF survived to the boost activation (fig. 1), then it unlocked after a few seconds due to different reasons; in the meanwhile, SSFS was behaving normally. Maybe some good adjustment was performed in that trial, but there is the possibility that the lock survived by chance. Locking at the error signal, one can notice that it was exactly crossing zero when the boost was activated (fig. 2). The activation of the boost on MC side happened when the error signal was already stabilized around zero (fig. 3).
Going back to the old successful lock acquisitions, one can see that the last one had the same lucky characteristic (fig 4). The previous ones look to me quite different in the way of activating the boost: sometimes I see a much longer ramp in the switch, sometimes the ramp is short but it seems that on SSFS side the boost is engaged well before the switch changes value.
Hopefully the magic adjustment will be applied again and the problem will disappear. If not, there is the possibility that the problem comes from the ramp.
According the Paolo entry, the trouble may come from the ramp used to perform the transition from the standard filter to the boost filter .
I put back the ramp of 10ms with a sine form ,
- at the first trial, the transistion to the boost succeeded
- at the second trial, the transition failed
Then I reduce the transition time to 5ms
- at the first trial, the transistion to the boost succeeded
- at the second trial, the transition succeeded too
The ramp was reduced and set to zero due to somes troubles observed and to some lack of knowledge .
According Paolo, the ramp is used to perform the switch between the 2 filters on the Sc_MC DSP too. As there is a delay between the SSFS rtpc and the Sc_MC DSP of 200us or 300us (2 or 3 10KHz cycles) the ramp must be slow compared to the delay.
According Paolo it could be useful to have the starting to the boost transition only when the error signal is close to zero
This version is now running for both SDB1_Rot and EIBAgilisRot. The VPM setting have been accordingly updated for EIBAgilisRot including the addition of the speed parameter on the MOVEREL command.
SDB1_Rot has been also added on FbsDet and the position data are now available on the DAQ (Fig. 1)
Today we have checked the delays between CAL_MIR/MAR_*_Z_CORR and Sc_MIR/MAR_*_Z_CORR for each mirror and marionette which should be consistent with the changes made in the DSPs :
- The marionettes have the same delay ~ 502 µs
- The mirrors have the same delay ~ 453 µs except for PR mirror which has ~ 666 µs
PR has been corrected today for aliasing but there is still an extra-delay of ~ 213 µs which is not seen in the other mirrors...
The firmware of the photodiode mezzanine of the DAQbox located in the suspended benches has been upgraded this afternoon around 4 pm LT. With this new version it is possible to send the command to close the shutter even if the open command is in progress.
The version of the DAQbox server has been upgraded to accept remote control of the VBias threshold via Cm message and to record the IBias current in mA instead of A.
-det
this morning I had to call the det on-call to realign the B4 beam.
-computing
at around 8:30utc we stared to have problems with users home directories preventing to use the linux workstation, promptly recovered by the computing team.
This morning around 9:00 UTC, I have restarted the process MdVim on olserver117.
Each command started by MdVim is a bash script which starts several root.exe in charge of doing the plots for the VIM pages.
The number of commands started by MdVim and running in parallel is now limited.
This should prevent the crashes of olserver117 that occured last week because of two many root.exe processes running at the same time because of a problem with the trend data access.
* B1_PD1 > 14mW
* B1_PD2 > 14mW
* B1p_PD1 > 10mW
* B1p_PD2 > 10.5mW (previously was 14mW, but the PD shutter closes at 13mA, so this was never triggering in July/August, for fast transients 13mA corresponds to 11.5mW)
* B1s2_PD1 > 40mW
* LSC_ENABLE going down to zero - at ~15:38:20 on Sep 18, tested that it triggers correctly the fast shutter, and that the fast shutter reopens after 10s even if LSC_ENABLE is still at zero
manual close
* INJ_IMC_LOCK going down to zero - it is able to for example very safely catch the IMC unlocks caused by the SSFS boost engaged at LOW NOISE 1, hopefully also will work for IMC fast unlocks
* manual switch
Figure 1 shows the INJ_IMC_LOCK is dropping 50ms before an unlock
Figure 2 shows that with INJ_IMC_LOCK added to the fast shutter trigger logic the unlock flash on B1p PD2 is not visible at all
There is work in progress to replace the trigger on PDs from the power in mW to current in mA, might be tested this afternoon. In that way the trigger quantity would be exactly the same for the PD shutter and the fast shutter, which would avoid the issue that fast transients break the linear relation between current and calibrated power. This will be more robust and allow a looser threshold on B1p_PD2.
The purpose of this afternoon's shift was to restore lock acquisition with the usual SDB2 sensors and work on the automation for the slow shutter; these are the results and the changes with respect to the past:
- the new slow shutter is handled by the DET_MAIN Metatron node; this is still operated manually until the procedure and the lock acquisition are validated;
- we restored the usual MICH fringe signal which uses B1p_DC with no issues;
- while in Recombined we did the first tests/adjustments for closing the shutter (state SHUTTER_CLOSED), and everything was working; since this shutter is closed, we do not close anymore the individual shutters for B1s1/B1s2/B1, so that they are ready to close later on if needed avoiding the risk for them to be stuck in-between transitions and not close quickly enough;
- nothing to report along lock acquisition before the realignment of the PRM but that the 7 Hz vertical mode seems to excite more easily than before;
- once the PRM is realigned, the vertical dampers for such mode are engaged as usual and their effectiveness is still good;
-
we had issues at first going to dark fringe, but in the end we succeeded and now the transition is very reliable, with just one unlock among many attempts, after the following operations:
- minor tunings of a couple of longitudinal gains (SSFS & PRCL);
- re-tuning of the B1p_56MHz demodulation phase;
- dis-engagement of the vertical dampers while going to dark fringe, as they were causing big oscillations on DARM; they are engaged again at the end of LOCKED_PRITF_DF;
- throughout the shift we managed to get to LOW_NOISE_1 just once (and for a brief time); the main issue seems related to the boost filter for the SSFS, which has some problem we did not understand so far, to be investigated;
- the engagement of the DIFFp loop seemed to work, as it lasted a few seconds, but probably the mixed driving does not when the COMMp is not closed, and I forgot to comment that out;
- both the B1p_PD1 shutter and (later) the slow shutter were opened in dark fringe before LOW_NOISE_1, even with the alignment loops still open; the logic seems to work fine and the thresholds have been tuned;
- the demodulation phase for B1p_QD2_56MHz was (hopefully) tuned, but the single test we managed to do is not conclusive; the B5_QD2_56MHz phase still needs to be tuned.
For these reasons ALL the alignment loops are commented in the Automation, as well as the change of the driving for the DIFFp; on the other hand, the new scheme for the vertical dampers is already in place.
Anything else to report.
The shift has been dedicated to the maintenance activity (started at 5:53 UTC), below the list of the activities communicated to control room:
- vacuum, large valves opening
- cryotraps refill;
- cleaning operation;
VPM
12:43 UTC - "RdsFbm_20" crashed/restarted
This morning we have checked and sligthly adjusted the polarization at the input of the pick off fiber.
We found that the buzzer used to inject some acoustic noise in the fiber was probably spoiling the measurement we did last time.
We aligned the output connector of the pick-off fiber so that the polarization in an optimal case should be s. Then we installed a polarizer
and minimized its transmission by adjusting the polarization at the input of the pick-off fiber.
We will ook at the data to check if the PC reference beam is now more stable in that condition.
Birefringence tuning:
Yesterday morning we checked the birefringence tuning of the OMC with the DET tower in vacuum. In the first part of Fig.1 the 2 OMC are locked on the direct bounce from NI (other mirrors misaligned). One can read 10.2 mW on B1_PD1 (B1_PD2 gives a different number at that time because it was not correctly calibrated), and 0,45 mW on B1s2P. Taking into account the fact that B1_PD1 reads only half of the B1 power, this gives a polarization mismatch between OMC1 and OMC2 equal to 0,45/(2*10,2) = 2.2%.
One can conclude from the previous result that the birefringence has been slightly detuned during the transition from air to vacuum (after the tuning performed in air last week, we had only 0,7% of polarization mismatch between OMC1 and OMC2). One plausible explanation could be that the tightening of the side screw (the one thightened with 4 cNm last week) has changed.
OMC1 alignment:
The alignment of the OMC with respect to the dark fringe beam has been finely tuned with the direct bounce from NI. A scan in temperature of OMC1 performed after this tuning is shown in Fig.2. One can read the following powers: 14.95 mW on TEM00, 0.116 mW on the mode of order 1 (with about 0.045 mW of piedestal due to the tail of the TEM00), and 0.5 mW on the mode of order 2.
This gives 0.5% of alignment defect (which corresponds to very good alignment), and 3.5% of mode mismatch (which can probably be improved in the future by a careful tuning of the mode matching telescope).
There is an imprecision in the conclusion of the entry: the effective RoC corresponding to a smaller Gouy phase is few meters SHORTER and not longer.
However, it should be noticed that the RoC LMA measurements average the radius of curvature over a region 150mm diameter, while the TEM00 and following 4 HOMs "see" a slightly smaller region of the mirror. In any case, more investigations will be performed.
Test has been repeated Monday 17th September, in order to have also a time whith FAN OFF and WATER and CHILLERS ON
Here the timetable (UTC):
13:18:00 +240S 13:22:00 ALL ON (reference time)
13:33:40 +290S 13:39:30 ONLY CHILLER and WATER ON (fan off)
13:39:50 +300S 13:44:50 ALL OFF
(subset 13:43:30 +100S 13:44:50 with no truks in FiPiLi)
13:45:45 +265S 13:50:10 ONLY FAN ON (water pumps, chillers, heaters are off)
Here a list of some relevant peaks (associated to the NEB HVAC system) disappearing in the acoustic or seismic sensors:
10 - 12Hz truks on FiPiLi ?
13.53Hz associated to the FAN system
20.60Hz associated to the FAN system
24.26Hz ALWAYS PRESENT
24.90Hz associated to the FAN system
33.70Hz associated to the FAN system
40.35Hz associated to the FAN system
41.70Hz ALWAYS PRESENT
43.90Hz ALWAYS PRESENT
44.60Hz ALWAYS PRESENT
45.45Hz ALWAYS PRESENT
46.15Hz ALWAYS PRESENT
47.35Hz associated to the WATER/CHILLER system
48.60Hz associated to the WATER/CHILLER system
Attached some spectrograms and some plots (Black is all OFF, Red is ALL ON, Blue is only FAN ON, Green is only WATER/CHILLER ON)
