- for D1 : current 28.6 A, Temperature : 26.5°C
- for D2 : current 32.1 A, Temperature : 26.1 °C - the output power of this diode has clearly decreased and 4 additional Amps was needed to recover its former value.
Crystals' temperature are : 25.1°C and 25.3°C
The EOM injection was optimized independently of the slave laser. The beam propagates properly through irises and the EOM. At low power, the ML power at the fiber output is 50 mW and 39mW just before the SL.
The DDS RF output was adjusted to 58 by measuring the monitoring port of EOM driver to reach 0.740V pp on 50 Ohm. The error signal on the SL was 1.7V. Still a factor 2.5 below its nominal value.
The SL was easily locked for an output power of 17.8W (on Nice gentec calorimeter) which represents 3% below its value during Feb 2020 (based on the SL reflection detector).
45 mW were extracted from the ML right after the EOM to seed the high power fiber amplifier. This power was injected into a PM fiber with an efficiency of 80-85 %
Although not necessarily the main reason for the lack of effectiveness of the cooling circuit, it is worth noting that the main circuit was still in a "safe config" with the valves at the end of the main loop fully open so that the whole flux could in principle take this way instead of going through the appliances to be cooled down.
This morning the valves at the end of the main loop were almost completely closed, so that the effectiveness of the cooling should be at its maximum now.
The DC distributions and the RF chains have been switched back on today around 15h30 LT and set at their nominal values.
In view of the replacement of the mini racks housing the RH and HWS-HR power supplies in the two end buildings, CCD and SLED have been switched off (for the NE see the first figure in attachment, for the WE the second one).
The RHs power supplies have not been removed yet to avoid to switch off the RHs.
This morning we found that there is no more Timing signals (IRIGB,10MHz) delivered to the SNEB Daq boxes.
The following plots (trend)shows:
- the 10MHz amplitude extracted with a demodulation mezzanine located inside the SNEB tank: SNEB_Clock_10MHz_mag
- the timing error of each SNEB DAQ box: Daq_SNEB_DBOX_Left{Up,Down}_timing_error
- the 12V monitoring of the Timing distribution mezzanine located inside the tank: SNEB_PowerUnit17_Timing10MHz_p12v_volt . When the SNEB_PowerUnit17_Timing10MHz_p12v_volt is grey it means that the SNEB_PowerUnit17 is not remotely reachable through the IP network.
It appears that:
- the Timing signals are not anymore received by the SNEB Daq boxes the 2020-07-30 around 09h28UTC: the SNEB_Clock_10MHz_mag goes to zero
- the Daq boxes monitoring signals are not anymore in the DAQ 5 hours later, the 2020-07-30 around 14h30 UTC. The 2 Daq boxes remains reachable from their TOLM links
The PowerUnit17 device electric plug was found not well connected: re-plugged it correctly allowed to recover the Timing signals . The operation was done around 07h45UTC today, this allows
- to close successfully the local controls
- to recover the remote control via Ethernet of the SNEB_PowerUnit17 device
We made further tests without success. At the end, we went in the DET-EEroom and we discovered that the ethernet cable has been left unplugged at the end of Friday's activities. Rplugging the cable solved the problem. We made another test and the motors are now reachable.
This morning SL_TempController, TangoCmLaserAmp and the server bridge ModBus/PLC on olserverwin have been restarted. The Power supply for NeoVan Amplifier pumping diodes has been turned on. The currents of diodes have been left disabled for the moment. Also, the valves of the cooling circuit have been opened.
This morning at ~7h55 utc we could engage the angular and position loop controls of SNEB without any problem. The setpoints were similar to what they were in March at the end of the run (see comparison between Fig.1 and Fig.2).
What we did to improve the error signal:
- adjusting the master laser alignment before the EOM and at the slave input.
- Increasing the RF input on the EOM up to 45 (units of output on the DDS). The EOM_monit signal is now about 700 mV pk-pk on 50 Ohm (similar to logentry 37175).
Once the lock was done, the alignment on the reflection photodiode was optimized: PSL_SL_D1R_DC: 0.37 V --> 0.42 V in datadisplay.
As detailed in this ently:
https://logbook.virgo-gw.eu/virgo/?r=32859
the communication to the motor controller is realized via a Moxa NPort 5130A Ethernet/Serial RS485 bridge.
In case of problem like the one reported, the very first thing to check is the status of the bridge by simply tring to ping it.
Since the communication with the bridge is not possible:
fcarbogn@olserver111[~/experience]: ping ethsdb201
PING ethsdb201.virgo.infn.it (192.168.226.40) 56(84) bytes of data.
^C
--- ethsdb201.virgo.infn.it ping statistics ---
10 packets transmitted, 0 received, 100% packet loss, time 8999ms
I suspect that the motor electronic, including the bridge, hosted in the DET ELab is switched off
Although not necessarily the main reason for the lack of effectiveness of the cooling circuit, it is worth noting that the main circuit was still in a "safe config" with the valves at the end of the main loop fully open so that the whole flux could in principle take this way instead of going through the appliances to be cooled down.
This morning the valves at the end of the main loop were almost completely closed, so that the effectiveness of the cooling should be at its maximum now.
As detailed in this ently:
https://logbook.virgo-gw.eu/virgo/?r=32859
the communication to the motor controller is realized via a Moxa NPort 5130A Ethernet/Serial RS485 bridge.
In case of problem like the one reported, the very first thing to check is the status of the bridge by simply tring to ping it.
Since the communication with the bridge is not possible:
fcarbogn@olserver111[~/experience]: ping ethsdb201
PING ethsdb201.virgo.infn.it (192.168.226.40) 56(84) bytes of data.
^C
--- ethsdb201.virgo.infn.it ping statistics ---
10 packets transmitted, 0 received, 100% packet loss, time 8999ms
I suspect that the motor electronic, including the bridge, hosted in the DET ELab is switched off
We made further tests without success. At the end, we went in the DET-EEroom and we discovered that the ethernet cable has been left unplugged at the end of Friday's activities. Rplugging the cable solved the problem. We made another test and the motors are now reachable.
These two days we have been working on putting the different parts of the Second Harmonic Generation (SHG) into their dedicated boxes and putting them into the rack at north end.
For the moment we are seeding it with a fibered laser installed at NE.
We still have to fix a problem with the temperature controller of the crystal but we manage to get 40 mW of green light at the output of the fiber.
This power should be enough to do the alignement on the in-air bench of SNEB.
A new control strategy for the inertial vertical control of the top stage has been tested on SR. The main news is the use of the LVDT sensors measuring the elongation stage by stage below the top stage. There is some indication that the method can work and give some improvement, but further studies are needed before saying that it can really be used. More details can be found in VIR-0692A-20.pdf. Currently the loop is back in the standard configuration.
The old probes to monitor the temperatures of the main chiller are back in place.
They were left floating during the last upgrade of the cooling circuit, but the shelving of the threshold expired so this reminded me to get them back to work. The ENV_LL_MAIN_CHILLER_PIPE_TE is attached to the return of the cooling circuit (first picture) and is dominated by the temperature of the pipe, although the temperature of the environment does play a role. The ENV_LL_MAIN_CHILLER_TE monitors instead the temperature of the cooling liquid inside the tank of the main chiller (second picture).
The related flags in the DMS are still shelved, we will monitor the temperatures with the ISYS on and then adjust the alarm threshold accordingly (third picture).
In order to prepare the EE room and the Detection lab for the infrastructural works we:
- Remove all the SQZ cables from DET lab and sealed the holes (Figure 1)
- Covered the Spectrum Analyzer in EE room.(Figure 2 ).
- Moved the RF generator HP8656B and some cables to the Padova-Trento office.(Figure 3 )
- Prepared the cover for the racks in EEroom (Figure 4, 5, 6,7). Now all the racks are open and they will be closed only during the infrastructural works.
Time is Local, that is 09:41 UTC
At 11.41 the injection with the new coil and better amplifier was terminated.
The amplifier is decoupled from the power supply, so before new injections can be performed the amplifier should be powered again.
In retrospect this was thus not necessary, since the GAS blades will heat up in the course of the next two days. The steep change in the actuator voltage from UTC 8:15 till UTC 9:10 is due to actions with the motor. trend plots included.
Today at 10 local time we started testing a better amplifier. Normally it should be able to reach +-72V, but for an unknown reason it is only able to deliver roughly up to +-40V.
At 10.15 local time we started a broadband injection with the maximal capacity of this better amplifier.
The findings of all the last injections will be discussed below. We discuss the most relevant which is able to provide us all insights in the magnetic injections.
We will compare 3 different injections:
1. Old coil at ~ the same position as the new coil, with the old amplifier in voltage driven mode, with voltage at maximal capacity. We inject broadband noise from 1Hz-201Hz.
2. New coil with the old amplifier in voltage driven mode, with voltage at maximal capacity. We inject broadband noise from 1Hz-201Hz.
3. New coil with the new amplifier in voltage driven mode, with voltage at maximal capacity. We inject broadband noise from 1Hz-201Hz.
General conclusions: new coil new amplifier > new coil old amplifier > old coil old amplifier
Discussion:
In figure 1: we have the current and voltage in the coil as well as the signal in the three CEB magnetometers. Purple is scenario 1 (old coil old amplifier), blue is scenario 2 (new coil old amplifier). Since we drive both at the maximal possible voltage, the voltage in the coils is the same. Given the lower resistance for the smaller old coil, this has a significant larger current. Nevertheless the new coil is able to inject more successfully magnetic noise below 100Hz. Above they become comparable.
In figure 2 (and figure 3 = zoom): From bottom to top: yellow= no injection, dark green = scenario 1 (old coil old amplifier), olive green = scenario 2 (new coil old amplifier), blue = scenario 3 (new coil new amplifier).
Both scenario 1 and 2 are able to successfully inject noise up to 200Hz and even slightly higher ~250Hz since the injected broadband noise has not a very steep fall off. Below 100 Hz scenario 2 is significantly better compared to scenario 1, with a factor of 10 improvement at the lower frequency end and a factor 2-1.5 when nearing 100Hz.
Afterwards they have comparable effects.
Scenario 3 is clearly the best of all over the entire bandwidth. At the low frequency end it outperforms scenario 2 with a factor of 2. The effect becomes smaller at higher frequencies. Also above 100 Hz scenario 3 is (marginally) better compared to scenario's 1 and 2.
We can conclude that the new coil is clearly able to perform better magnetic injections compared to the old one. Nevertheless we have shown there is still a significant margin for further improvement if we use better amplifiers since we are limited by the voltage we can give.
Yesterday, at around 18:00 UTC +/-2 hours the temperature in the detection lab started dropping by almost 1 degree as visible on the thermometer on EDB. So the transient in the SDB2 suspension due to change in thermal load on the SDB2 bench cannot be studied further. I have switched back on the DAQbox in SDB2 this morning around 8:10 UTC.
Figure 2 shows a summary of the temperatures over the past few days, and the voltage on the vertical actuator of the SDB2 suspension (act3), with two adjustments of the stepper motors clearly visible. Before the external temperature change in total the voltage increased by 3+2 = 5 volts, and it was getting close to finding a new steady state. So probably the step of 52W in power consumption on SDB2 correspond to 6V on the actuators, so moving from saturation at -3V to +3V.
For the ~30W dumped on SDB2 during the CARM offset reduction during O4 lock acqusition commissiong this should not be a problem, as this should last a week or two, during which daily adjustments of the SDB2 mSAS stepper motors might be needed. But after the CARM offset reduction is commissioned the 30W power dump on SDB2 will last only a a few minutes, and won't have a large impact, as the time scale of the thermal response is ~1 day.
For the ~40W dumped on SIB2 when PR is misaligned, it will be a larger issue once SIB2 is suspended in vacuum. The power will be dumped when PR misaligned, but there will be no power dumped on the bench when the interferometer is locked. This means that the SIB2 suspension will need to be adjusted whenever the ITF is unlocked for more than 1 day, or whenever the ITF is locked for more than 1 day after being unlocked for more than 1 day. It might be cumbersome for operations. This could be reduced if the power reflected by the ITF is also dumped on SIB2 when the ITF is locked, instead of a beam dump in air outside SIB2. That beam dump in air is expect to receive 10W when the ITF is locked. So having these 10W dumped on SIB2 would reduce the power variation by 25%, from 40W to 30W. And as a bonus it could reduce acoustic coupling through scattered light of the beam dump that is presently in air attached to SIB2.
