Alexa, Rana, Keita, Kiwamu, Sheila, Jeff
Today we worked on more robust prmi sideband locking. After the coil driver swap, alog 13956, and the required model changes alog 13958 we changed our PRC loop. We are now acquiring with just M3 on PRM, and it is easier. We also reworked our locking filter for PRCL, to give us more gain at low frequencies and more phase margin at 80 Hz. A comparision of the old filter (FM7) and the new one (FM9) is shown on the right hand side of the first attached screen shot.
Also shown are the open loop gains of both MICH and PRCL.
The second attached screen shot shows the configuration that has been locking pretty robustly, with the exception of the power normalization, which we have off when we are acquiring lock. The guardian has been updated to fit this configuration.
After a looooong struggle with the filters, I finally managed to have some sensor correction on HAM2-ISI.
I've installed a FIR filter and its companions (LP&AA, Blend with IIR) in the FIR path, and a High Pass Blend in the IIR path of X and Y (for more information on those filters, see DCC #T1200285).
The following plots show the performance with sensor correction on X only. For some reason, I had to tune the gain to -10 to obtain some noticeable improvements.
- The red curve shows the nominal configuration. This config is wanted nominal because it provides good performance above the microseism without reinjecting noise at lower frequencies.
- When we turn on the sensor correction (orange-ish curve), we can see some improvement at the microseism and in the integrated RMS of the OPLEV (factor of ~3). BUT, as a tradeoff, we are reinjecting some noise below (the coherence between the GS13s and the ground STS is dropping below 100mHz).
- Switching to a higher blend doesn't seem to help the sensor correction (blue-ish curve)
- Even if the 100mHz blend is helping at the 0.7Hz resonance, it reinjects a lot of noise at low frequency (green-ish curve). We might not want to use that configuration.
By looking at the optical lever, those preliminary results are encouraging. But we have to be careful: even if the absolute motion is reduce, it doesn't say anything about the relative motion in the cavity. The next step will be to look into that.
-> The idea will be to implement some sensor correction on HAM2 and HAM3 using the same ground sensor. Driving the chamber with the same input signal should limit the relative motion in the cavity.
Commissioners were playing with PRCL today, I'll have a look into that ASAP.
Starting at ~01:15 local
All is automatic except watchdog trips. Please leave SUS etc damped.
This morning Patrick covered the shift from 9am-noon (THANKS!). Below are today's primary activties:
Hugh, Dave
We added the hepi pump controller alarms to the operator alarm station. This is a new HEPI alarm handler, started and stopped in the same was as the other alarms.
Today Dave noticed on the CDS Overview medm that the Slow Controls computer froze at EX (h1ecatx1); basically, on the CDS Overview the Slow Controls channels (bottom left of window) were WHITE/INVALID for EX.
Sheila showed me how to remotely access the Desktop to this computer via a shortcut on the Ops Workstation (basically go to Applications icon/CDS/Remote Desktop/h1ecatx1). Click the mouse on the desktop and then a remote desktop will open. On here, the Windows PowerShell window was displaying Errors. I typed in "exit" and waited many minutes but it did not exit. I couldn't find "Task Manager" on the PC, so Sheila showed me how to search for it and bring that up. We selected the Windows PowerShell task & clicked End Task (closing the PowerShell window). Then the "Restart" icon was clicked to restart the computer (this closes the Remote Desktop session).
At this point, I closed the CDS Overview window (since the EX Slow Controls section was WHITE-box-ed). Upon re-opening, we were back to having live & GREEN channels for EX.
J. Kissel, T. Sadecki Travis has plugged in OSEMs to QUAD06 mounted to Mechanical Test Stand 1, read out by QUADTST, in the West Bay of the LVEA for the first time in many moons. After having re-compiled, re-installed, restarted the front end process a few days ago (see LHO aLOG 13833), the EPICs records ha not been restored. In addition, the OSEMINF and DAMP filter banks were sorely out of date. So, I BURT-restored the presumably very-old ${userapps}/release/sus/h1/burtfiles/h1susquadtst_safe.snap which restored all matrix elements correctly. I then installed the just-measured open light current values (recorded below for QUAD06). However, since the foton filters were all wrong, I took some time to copy the OSEMINF and DAMP filters from H1SUSITMY, turn them on correctly, checked that the matrices and COILOUTF gains are correct, and grabbed a new safe.snap. and committed both filter file and fase.snap. Travis will gather the OSEM serial numbers shortly and match them to there location installed below. OSEM Open Light Compensation Compensation Current [ct] Gain [ ] Offset [ct] M0F1 29723 1.009 -14862 M0F2 29212 1.027 -14606 M0F3 28998 1.035 -14499 M0LF 27100 1.107 -13550 M0RT 29422 1.020 -14711 M0SD 29032 1.033 -14516 R0F1 26949 1.113 -13474 R0F2 28436 1.055 -14218 R0F3 27950 1.073 -13975 R0LF 28311 1.060 -14156 R0RT 27786 1.080 -13893 R0SD 26947 1.113 -13474
Due to some shorting issue that occurred while swapping the R0 LF and RT BOSEMs, I had to swap out the R0 LF BOSEM. We recalculated and reloaded the offset and gain for this channel. The new values are:
OSEM Open Light Compensation Compensation Current [ct] Gain [ ] Offset [ct] R0LF 29804 1.007 -14902
And the corresponding serial numbers for all BOSEMs are:
M0F1 026
M0F2 195
M0F3 015
M0RT 102
M0LF 302
M0SD 294
R0F1 310
R0F2 292
R0F3 085
R0RT 029
R0LF 574
R0SD 437
Per Work permit 4849 Kiwamu and I replaced the bottom stage coil driver for the PRM. M3 now has the increased range coil driver as modified per ECR 1200931. We swapped Triple Acquisition Coil Driver S1100022 with S1100045
I have restarted the DAQ using a hand modified H1EDCU_GRDLITE.ini file, which is the guardian generated INI minus the GRD systems not currently operational: ISC_LOCK, SEI_HAM6, ISI_HAM5, HPI_HAM6. No Beckhoff INI changes were found.
Shut down h1oaf0, powered off I/O chassis, and replaced the 16 bit DAC card (on a carrier card) with a 16 bit PCIe DAC card. Powered up I/O chassis and computer, measured voltages to verify we didn't have a repeat of last week's attempt. The DAC card we tried last week was tested on the test stand and verified to be bad. The DAC card used this week was tested beforehand and found to be good.
Got the full TF this morning on HAM4. As expected and seen on the other platforms, the HEPI features seen in the ISI TF from 10-40hz are lower in Freq & Q--likely from the floating HEPI itself. And now with the HEPI loops at 2hz ugf rather than 5, that nasty zero near 6hz becomes much less nasty--see attached. This shows the local to local TF of the ISI Actuators to the CPS. The H3 sensor on the HAM ISI aligns with the long axis of the HEPI crossbeam. In the Cartesian basis, this is the Y axis for HAMs 2 & 3 and the X axis for HAMs 4--6.
Shutdown of H1OAF0 to replace TCS DAC card. This caused TCS chiller and laser shutdown. The laser chiller temperature set point is controled by the DAC, and a zero volt output puts the chiller setpoint outside the safe operating range causing the chiller to shutdown. This then causes the laser to shutdown.
Peter and Dave.
We rebuilt all h1psl0 models against RCG2.8.5 (h1ioppsl0, h1psliss, h1pslfss, h1pslpmc, h1psldbb). We restarted the models, using the latest safe.snap burt restore files created on Sept 5th.
After the 14th August 2014 power outage, I have downgraded the PSL to RCG2.8.3 as part of our attemp to open the closed shutter.
We were unable to reproduce the problem of not being able to open the Beckhoff controlled shutter in the diode room. When the PMC model was started the shutter closed. On the LASER MEDM screen, we RESET the Flow Sensor External Shutter Request (required confirmation) and then pressed the Control Room RESET (and confirmed), the shutter then opened.
The temperature sensor interface box (D980400, S/N BL3001, S1400577) was installed on the PSL table. All the temperature sensors were plugged in. The ambient temperature sensor was taped to the side of the PSL table. The DB25 cable for the interface box was pulled from the outside. This cable has a gender changer on it temporarily as a DB25 male to DB25 male cable was not available. Whilst there I tweaked the alignment into the reference cavity. Transmission improved from ~1.1V to ~1.8V. We should look at this when there is a period of time greater than the 4 hour maintenance interval. Also whilst in the enclosure, all of a sudden the air conditioning and the HEPA fans switched off. The AC control units inside the enclosure were off, ie nothing on the LCD display was displayed. The control unit outside had a message "TEMP SWITCH OFF" and was indicating that the fans were still running even though they were not. I tried restarting the fans, using the procedure near the display unit, and nothing happened even though it indicated the various fans were running. After waiting about 5 minutes, I tried restarting the fans again. This time it worked. The AC control units now indicated: CHECK E4 01 00 ERROR CODE 10 I cycled the AC using the control unit, everything seemed fine afterwards.
J. Kissel I've resolved issues identified with the ETMY GND STS storage in the frames, originally identified in LHO aLOG 12818 -- the channels H1:ISI-GND_STS_ETMY_X_DQ H1:ISI-GND_STS_ETMY_Y_DQ H1:ISI-GND_STS_ETMY_Z_DQ are now properly recording the ground STS2 as the rest of the STS2s on site. Details: At the end stations, the STS2s are piped directly into an AA chassis, as opposed to the corner where there is an analog distribution box in which there enough spigots to swap cables around such that the three STSs enter the ADC in an arbitrary order (see D0901301). However, internal wiring of the STS2 Interface Chassis, spits out the X, Y, and Z, channels out on its DB15 such that when piped directly into the AA chassis, skipping the STS Distribution Chassis, it comes in on ADC3 channels 23, 24 and 25. In the the corner, ADC 3 channels 23-25, 24-26, and 27-29 are piped into the common front-end model library part for the BSC-ISI, isi2stagemaster.mdl, as STSA, STSB, and STSC, respectively in that order. At EX, in order to get the GND STS which comes in on ADC3 23-25 to map to STSB, the top-level model has been re-arranged to get ADC 23-25 into STSB. This had not been done at EY. That's now fixed.
Thanks for implementing this! It's very useful to have the science frame version of these channels reading data.
I did a quick comparison of one DOF (Z) to the rest of the LHO STS2s for a time a couple hours later and it looks like we're off by a gain of 10 - see attached png.
J. Kissel, P. Thomas As per work permit 4840, and in order to resolve issues reported in CDS Bugzilla #721, I've added the "accum_overflow = 1" flag to the parameters block at the top level of the H1SUSSRM and H1SUSTMSX front end models this morning. With this addition, these SUS will match every other SUS's overflow behavior in that it will accumulate overflows for each individual channel in their respective DAC monitors, as opposed to displaying the overflow rate. For reference, this was added to every other suspension during the changes described in ECR E1300578, and on pg 18 of G1301192; see aLOGs from November 2013 (e.g. 8226, 8325, etc.), and these were just accidentally overlooked in all the fervor of activity. Detailed Notes SRM captured safe.snap before restart, but just in case alignment offsets are p 0 y 2000 safe did *not* restore alignment offsets... why? moving on. restored alignment to values prior to restart. IPC errors popped up on SR3 and OMC because SRM is the master model, distributing Binary I/O channels to the other two. Upon restart the connection was lost, but it's a one-time error. A diag reset on the GDS_TP screen cleared the error. TMSX captured safe.snap before restart, but just in case alignment offsets are p -24 y -315 safe restored alignment correctly. Tested cumulative overflow on TMSX by adding large offset to M1 F3 COILOUTF. DAC saturates, and DAC monitor accumulates as expected. Tested cumulative overflow on SRM by adding large offset to M2 UL COILOUTF. DAC saturates, and DAC monitor accumulates as expected. safe.snaps committed to userapps repo.
Swapped the laser on the BS optical lever to hopefully resolve the small glitch reported by Kiwamu here and here. The SUM out of the QPD has dropped to just under 5300 counts, down from over 11000.
I have NOT swapped the fiber yet, as this will disturb the alignment of the optical lever. If the glitch is still there I will swap the fiber next.
AS Jason, Keita and I disscussed this morning, I left the BS oplev damping off for several hours today, from 9/9/14 17:00 UTC to 9/9/14 23:00