aaron.sevigny@LIGO.ORG - posted 11:12, Thursday 21 August 2014 (13537)
SR3 OPLEV laser DC regulator added
SR3 Oplev laser dc S/N 194 DC regulator D1200461 added
SR3 Oplev laser dc S/N 194 DC regulator D1200461 added
I physically realigned the MC1 and MC3 GigE cameras (h1cam11 and h1cam12 respectively) because they had been misaligned spontaneously. The attached are the snapshots of their views to show how they look like after the realignment.
Purge air is on while Jim works there.
Based on the matlab tf measurements taken after coil balancing on SRM SR2 and MC2, and in order to have a physical idea of what's happening, I calculated the DC coupling between longitudinal motion and test mass angular motion for the three stages of the three HSTS suspensions. The results are summarized in the attached spreadsheed, and the numbers in the table are from the transfer function measurements at 0.01Hz (7 averages using schroeder phase drive), and assuming the coherence was good. The transfer functions results are summarized in the second attachment.
To note from the results :
*Length to pitch coupling is roughly ~2rad/m and length to yaw is ~1rad/m.
*M2 length to M3 yaw coupling of SR2 is higher because of the broken actuator.
*M1 length to M3 yaw as well as M3 length to M3 yaw has a different phase than the other sus tested.
The plots can be generated by using the script living in trunk/HSTS/Common/Matlabtools/plothsts_matlabtfs_cross_levels.m
The matlab tf results can be found under /ligo/svncommon/SusSVN/sus/trunk/HSTS/{optic}/{level}/Data
K. Venkateswara
This morning I noticed that the BRS was rung up to large amplitudes (500 microrad), so I looked at the data from yesterday and noticed two main incidents/causes: Borja and Sheila were near the BRS at ~ 3:30 PM but that disturbance was small and did not drive it up the resonance much. The ASD and time series of the first disturbance is shown in the first attached file.
The second disturbance at ~5 PM was much bigger and really drove up the resonane of the bar from 50 nrad to 500 nrad as shown in the second file (the y-axis scale in the time series should be in nrad not rad). I'm not sure who this was yet. The gravity gradient from a person standing next to the balance is very large and can easily drive up (or down) the resonance amplitude to microrads. As the decay time is ~two days due to the high Q, it will take a long time for the bar to return to normal amplitudes. Furthermore, due to the non-linearities of the autocollimator, the harmonics of the fundamental become visible and destroy coherence with the seismometer as visible in the third file.
This is a known problem due to the open loop operation of the BRS and we hope to address it very soon with some simple feedback/damping loop. In the meantime, I would like to ask that people stay away from the BRS unless necessary. Gravity gradient torque falls off as distance^3, so even an extra distance of 2-3 meters is sufficient to keep the influence small. Grvaity gradient torques are also filtered by the resonance of the balance, so if you move by it fast, that is generally okay.
When it does get rung up, it is kind of fun to drive it back down using your gravity :)
While the laser is shuttered this morning, I adjusted dark offsets for MC2 trans QPD, IM4 trans QPD (I changed the whitening gain on these yesterday to match the gain settings paul used for the calibration (alog 9716 ). I also adjusted IMC trans and refl PD dark offsets.
The LVEA is now LASER SAFE
model restarts logged for Wed 20/Aug/2014
2014_08_20 15:09 h1broadcast0
2014_08_20 15:09 h1dc0
2014_08_20 15:09 h1fw0
2014_08_20 15:09 h1fw1
2014_08_20 15:09 h1nds0
2014_08_20 15:09 h1nds1
2014_08_20 15:10 h1dc0
2014_08_20 15:12 h1broadcast0
2014_08_20 15:12 h1fw0
2014_08_20 15:12 h1fw1
2014_08_20 15:12 h1nds0
2014_08_20 15:12 h1nds1
no unexpected restarts, DAQ reconfiguration.
Alexa Dan Fil Keita Koji
This afternoon we brought the second of Koji's three OMC children under control.
After working in the morning to understand the quadrant layout of the OMC QPDs, we were able to close simple diagonalized versions of the OMC QPD alignment loops to stabilize the beam in the chamber. With Fil's help we enabled the high voltage to the PZT driver (in the process finding a cable that was plugged into the wrong slot on the AI chassis). Before fiddling with the high voltage we checked that the HV readbacks were working as advertised. We were able to drive the high-voltage PZT in a triangle wave across its entire range (0 -> 100V) and see flashes in the OMC. We tried to improve the alignment by watching the flashes and searching for a clear 00 mode, but this turned out to be fruitless, probably because the beam is moving too much.
We proceeded to align the rest of the beam paths on the HAM6 table: OMC REFL QPDs, AS WFS, and the OMC REFL beam through the viewport. For a while we were confused by a mirror on a too-short pedestal; Keita showed up at a lucky moment and fixed it. The beam has been aligned onto all of the QPDs in the vacuum.
We should be able to lock the OMC tomorrow and study the alignment (and, possibly, take a first stab at the mode matching). At the moment, the beam is moving a lot in pitch; this motion is due to something upstream of HAM6. We suspect SR3 and the ISI table in HAM5, but it could be something else. This motion really needs to be fixed before any robust OMC locking can happen, either by stabilizing the source or constructing a more robust OMC QPD servo. While we flashed the OMC Koji was able to visualize the modes inside the cavity, and the 00 mode was there.
Before leaving the chamber we relieved the yaw and pitch bias on the OM tip-tilts and centered the OSEMs, and then clamped the tip-tilts to the table. The TTs are in good shape. One of the OM3 OSEM cables can potentially slip and block the beam to the AS WFS, we should find a way to clamp this cable in a safe place.
Tomorrow we need to install the kapton washers on the pico drivers. We still need to align the OMC TRANS path, but we probably have to lock the OMC on a fairly strong 00 mode in order to see the beam. In principle OMC TRANS can be aligned after the north door is installed.
Now that we have a beam we can do RF balancing of the AS WFS, if that's necessary.
We'd like to try to put the north door on HAM6 by the end of the day tomorrow. Two other things need to happen before we can attach the door: cable dressing and table balancing by the SEI crew, and a check for ground loops. (These steps may need some iteration if the ground loop situation is bad.)
For now the purge air has been turned off to give the mirrors a rest.
SR3 appears to be moving around a lot more than the other large triples (judging from the BLRMS channels that are used by the watchdogs). The gains for the roll, pitch, and yaw damping filters on the M1 stage are lower than longitudinal, transverse, and vertical gains. It doesn't look like these gains can be increased indefinitely, but I increased pitch by a factor of ten (-0.002 to -0.02) and things got a little better. These loops need some love.
Here are the current gain settings (only pitch has been changed):
L: -3.0
T: -5.0
V: -1.0
R: -0.02
P: -0.02
Y: -0.02
Note: Although T1300611 suggested that the HV power supply was to be 200V and the nominal output voltage was 100V,
we uses 100V for the power supply and thus the nominal output is 50V. This is due to decision after the failure of
one of the PZTs in L1 OMC.
From the point of view of the beam the OMC QPDs segments are arranged as follows:
3 2
4 1
It appears that the end X FMCS EPICS readbacks went invalid at around 08/19/2014 23:35 UTC (from a trend of H0:FMC-EX_VEA_202A_DEGF). I checked the FMCS Windows computer and they appear to be down there as well. I'm not sure how to fix this, so I will email John.
J. Kissel, J. Worden John called and suggested we reboot the FMCS machine in the back of the control room. After 20 minutes of windows updates, I was able to log back in, but the status of the X-end did not change. We're both unsure if there's any program to be restarted, but John thinks not. The in-vacuum components (SEI / SUS), nor to the ground instruments (T240 / BRS / PEM) and other PEM instruments show no sign of bad news, and the vacuum pressure looks reasonable, so we assume this is just a failure of the readback system. John will look into the problem when he's back in on Friday.
This is a known issue with the network hubs that are used in the mechanical room to connect the FMCS controller to the network (along with a couple other systems); they are flaky after power outages, and just in general. In this case, I changed out the hardware since it's been flaky over the past few days and the usual power on/power off thing didn't clear it up today. The long term fix is to pull more cable so that these things can be directly connected to a real switch. P.S. Only the vacuum system shares any common infrastructure with FMCS - the operation (or non-operation) of any 'fast' front end EPICS system is not a useful indicator for problems limited to either of these (vacuum/FMCS).
Borja, Sheila, Gerardo
As per original plan, and after taking another set of charge measurements in X and Y, today at about UTC 2014-08-20 23:00:00 we turned on the green laser light with a measured power of 57.5mW going into the periscope that injects the beam into the ETMX chamber. In parallel Gerardo turned on the turbo pumps in ETMY and closed the gate valve of the ion pump.
model restarts logged for Tue 19/Aug/2014
2014_08_19 08:56 h1isietmx
2014_08_19 12:00 h1susetmy
2014_08_19 12:00 h1sustmsy
2014_08_19 16:33 h1hpibs
2014_08_19 16:33 h1iopseib2
2014_08_19 16:33 h1isibs
2014_08_19 16:35 h1hpiham4
2014_08_19 16:35 h1hpiham5
2014_08_19 16:35 h1iopseih45
2014_08_19 16:35 h1isiham4
2014_08_19 16:35 h1isiham5
2014_08_19 21:37 h1dc0
2014_08_19 21:38 h1dc0
2014_08_19 21:42 h1dc0
2014_08_19 21:43 h1broadcast0
2014_08_19 21:43 h1fw0
2014_08_19 21:43 h1fw1
2014_08_19 21:43 h1nds0
2014_08_19 21:43 h1nds1
2014_08_19 21:46 h1dc0
2014_08_19 21:47 h1dc0
2014_08_19 21:48 h1broadcast0
2014_08_19 21:48 h1fw0
2014_08_19 21:48 h1fw1
2014_08_19 21:48 h1nds0
2014_08_19 21:48 h1nds1
maintenance work on ISI ETMX and SUS ETMY. Unexpected restarts of SEI B2 and H45. Restart of DAQ following maintenance.
Borja, Sheila
We turned on the laser at end X, so that Borja can do a charging measurement after having green light on for 1 day. To do this Rai would like 50mW of 532nm light on the optic. When we turned on the laser the setttings were 1.843 Amps 31.8 C and for the SHG crystal 33.83 C (set value 33.81C). We measured 54 mW of green out of the laser and 34 mW going into the chamber.
Borja adjusted the doubling crystal temperature to 33.53C actual value (33.51C set value) and the power going into the chamber increased to 38.5mW.
Borja then turned the laser current up to 1.912Amps and the power going into the chamber stabilized at 57.5mW.
We plan to reset the current to the old value after the charging measurement is done, but leave the doubling crystal temperature at this setting.
In order to figure out what is going on with the IMC cavity pole, I made a comparison between a recent and past IMC open-loop transfer functions.
If the cavity pole really dropped to 7.7 kHz from 8.8 kHz, this would cause a measurable change in the open-loop transfer function. According to the comparison, I did not see a large change. This result supports the hypothesis that the IMC cavity pole stayed (almost) the same.
The plot below shows the IMC open-loop transfer functions: one is taken yesterday shown in in blue, and the other was measured on February 7th with the same servo board configuration but with a higher laser power.
As shown in the plot, they look similar to each other except for the gains. The difference in the gains must be due a combination of the different laser power and non-carefully adjusted gain on the IMC servo board. In addition, there is a large discrepancy at high frequencies above 80 kHz, but this is pretty much out of our interested frequency band. So we don't care about it.
* * * *
If the cavity pole really dropped from 8.8 to 7.7 kHz, this would decrease the response above 7.7 kHz by about 12 % in amplitude and also give an extra retardation in the phase by 3-4 degrees at around 8 kHz. In order to check if there is such change between the two transfer functions, I took a ratio of the two TFs -- the plot below shows a transfer function of (recent TF) / (past TF).
Even though the data does not look super smooth, we can already see that there is no 12% drop above 8 kHz in the amplitude or no 3-4 degrees retardation in its phase at around 8 kHz. It rather looks as if the IMC stayed almost the same since February. Also I am attaching a matalb fig file of the same plot in case somebody wants to play with it.
Here are the raw data.
