(jim nic)
We've implemented the ISI CPS -> MICH feedforward (MICH freeze). It was a fairly simple process:
We installed the feedforward coefficients that Ryan suggests in his writeup. These could be re-measured to potentially improve the subtraction.
Then we compared the spectrum of the feedfoward signal and the MICH control signal. We saw a relative gain of 188, so this was added to a filter bank in the feedforward path.
We added a 50Hz low pass to the feedforward path so we aren't injecting too much noise.
When the feedforward path is enabled, we have a reduction in the MICH control signal RMS by a factor of 2-3. See attached figure (reference traces are with FF turned off).
The CPS has a 9Hz feature. Jim says it's from HEPI. This is injected into the MICH control signal, but the change to the RMS is insignificant.
Also attached is the settings of the MEDM screens. Several of these are 'hidden' because they have only been autogenerated and not added to the main screens.
Finally, the easy way to turn on and off the feedforward is by setting the channel H1:LSC-CPSFF_GAIN to 0 (off) or 1 (on). It has a 10 second ramp.
Dave, Daniel All Timing IRIG-B Modules at LHO are now currently showing the correct number of leap seconds: 17 The leap second was added automatically at 23:59:59 UTC on June 30th, 2015 using the leap second functionality on the IRIG-B.
Patched ldas-tools-1.19.32 with new leap second data, rebuilt and installed in /opt/rtapps. Built new versions of daqd for the frame writers, NDS, broadcaster, and data concentrator. Restarted data concentrator to get new daqd running on the DAQ. Frames are being written with the correct value for leap seconds.
Jun 30 16:59:59 h1conlog1-master kernel: [12118377.100031] Clock: inserting leap second 23:59:60 UTC Jun 30 17:00:11 h1conlog1-master conlog: ../conlog.cpp: 301: process_cac_messages: MySQL Exception: Error: Duplicate entry 'H1:ASC-ODC_CHANNEL_OUTMON-1435708800004681820-3' for key 'PRIMARY': Error code: 1062: SQLState: 23000: Exiting. Jun 30 17:00:12 h1conlog1-master conlog: ../conlog.cpp: 331: process_cas: Exception: boost: mutex lock failed in pthread_mutex_lock: Invalid argument Exiting.
Restarted
Kyle, Gerardo Connected LD to HAM6 annulus aux. cart (in parallel i.e. "split" flow) and sprayed audible flow of helium around entire periphery of East and South door outer O-rings pausing for 5 seconds at each leak check slot -> no LD response -> Began pumping HAM6 Also, backed X-end turbo with LD (LD only, scroll removed) and sprayed new, untested, conflats at X-end with 20 seconds of audible helium flow - OK -> resumed backing turbo with scroll pump
I repeated the charge measurements on both ETMX and ETMY. Plots are attached.
LVEA: Laser Hazard Observation Bit: Commissioning 07:30 Karen – Restocking garb at end stations 07:30 Jodi – In LVEA for 3IFO and property management work 07:55 Sprauge on site to check traps – Joe D will escort 08:00 S&K Electronics – Installing conduit in X-Arm near end station 08:20 Jodi – Out of the LVEA 08:22 Patrick – Setting up test equipment for Beckhoff vacuum testing 08:25 Gerardo – Vacuum survey in LVEA 08:32 Bubba – Going to Mid-X looking at the grout 08:35 Richard – Going into LVEA for Beckhoff vacuum testing 08:39 Karen – Back from the end stations 08:40 Christina & Karen – Going into the LVEA 08:41 Gerardo – Out of the LVEA 09:02 Katie – In the LVEA checking Magnetometers 09:05 Kyle – Going to End-X 09:22 Jim – Working on HAM3 09:26 Bubba – Back from Mid-X 09:27 Bubba – Going into fan room in the mechanical building 09:45 Jeff K. – Going into the CER 10:15 Richard & Patrick – Out of LVEA finished with test 10:23 Jordan – In LVEA installing PEM antennas 10:56 Dave & Jim – Installing 18 bit DAC cards at HAM5 & HAM6 11:00 Rick & Jason – Going into PSL enclosure to look for replacement optics 11:13 Filiberto – In CER to swap EtherCAT chassis 11:15 Richard – Power cycling MC2 (and other suspensions) chassis 11:26 Richard – Out of the LVEA 11:28 Kyle & Gerardo – Going to HAM6 to setup and start pumping 11:30 Jim – Going into LVEA to swap bad 18 bit DAC card 11:32 Rick & Jason – Out of PSL Enclosure 11:49 Filiberto – Out of the LVEA 12:04 Kyle & Gerardo – Going to End-X to recover leak checker 12:43 Kyle & Gerardo – Back from end station – Going to HAM6 12:48 Kiwamu – Restart LSC frontend model 13:18 Kyle & Gerardo – Out of LVEA – HAM6 is being pumped down 13:45 Sheila – Transition LVEA to laser hazard 14:06 Kyle – Going to HAM6 14:08 Jordan – Out of the LVEA 14:36 Greg & Nutsinee – Going into the LVEA to start TCS chillers and check tables 14:46 Patrick – Pause replication & backup on H1conlog1-replica 15:45 Greg & Nutsinee – Out of the LVEA 15:55 Dave – DAQ restart 15:59 Kyle & Gerardo – Going to HAM6
Completed logical backup of h1conlog database on h1conlog1-replica. Compressed and copied the file to /ligo/lho/data/conlog/h1/backups/h1conlog_dump_06_30_2015.sql.tgz. 6.6 GB WP 5318
Added 97 channels. Removed 318 channels.
J. Kissel, K. Izumi, J. Betzwieser Prior to ER7, Kiwamu and I had convinced ourselves that the delay installed in the front-end (i.e. what produces DELTAL_EXTERNAL) on the actuation path should be the difference in the known time delays on the actuation and sensing paths (see LHO aLOG 17951), and therefore entered in "1.0" to the number of clock cycles the actuation path should be delayed. Since then, Joe reminded us that what should be installed is the sum of the known delay. See pg 11 of G1500750 for clear pictorial explanation. This sum is closest to 4.0 clock cycles. I have now updated H1:CAL-CS_DARM_CTRL_DELAY_CYCLES (the EPICs record that dictates how much the actuation path should be delayed) to be 4.0 clock cycles, and accepted the new value in the SDF system. NOTE Because this front-end actuation delay has been moved upstream of the application of the actuation function calibration (see LHO aLOG 19382), only DELTAL_EXTERNAL receives this delay now. In other words, DELTAL_CTRL, and DELTAL_RESIDUAL must be delayed / advanced appropriately if you want the right answer. The intent is that GDS calibration pipeline will add the appropriate, as-precise-as-we-know-it, delays to each path (apply the other super-16[kHz]-Nyquist-frequency poles/zeros), and add them together to form the more accurate GDS-CALIB_STRAIN.
Note, I say "back" to 4 clock cycles, because we had 4.0 clock cycles installed prior to May 12 -- see LHO aLOG 18393, which cites that we got the number originally from copying the LLO front-end calibration setup (i.e. the number Joe and Den had already correctly calculated / installed as 4.0, LLO aLOG 16421) -- Kiwamu and I just hadn't grokked why it was 4.0 until a month or two later. Hooray for science!
Dave, Daniel, Jim Reprogrammed the MSR IRIG-B at 12:45 PDT with Rev. 113 code (provides compatibility with the display to be used in the Control Room). Last week, Jim and I used the test stand to confirm that the flywheels in the IRIG-B PCI cards could keep stable time over a few minute interruption in the IRIG-B signal. Reprogramming took 38 seconds; the whole process of reprogramming, rebooting, resyncing, and confirming the time zone/leap seconds took 102 seconds. I had to add the leap second scheduled for later today. MEDM screens for MSR IRIG-B are nominal. The rack-mount LED display no longer works, since Rev. 113 changes the display signal to be compatible with the display model to be used in the Control Room. Work permit 5317.
(nicolas kiwamu jeff)
We will try to implement the MICH freeze technique soon. In order to facilitate this, we made the necessary changes to the LSC frontend model. The ISI model changes were already done by Arnaud a few weeks ago.
We have compiled, installed, and restarted the new model.
The detailed changes are below:
We updated the lsc library part from the SVN which includes the new signal paths for MICH freeze feedforward.
The input terminals of the lsc library part had changed, this required shifting many signals to account for a new signal being fed into inputs 10 and 12.
We added the IPC receivers for the signals from the ISI models. These feed directly into inputs on the library model.
In addition to the update of the LSC frontend model, we also updated the h1omc model in order to make it to the latest (see LLO alog 18707 for detail). In order to do so, we downloaded the lateset common omc and lscomc models from the svn and compiled h1omc with the latest common models in place. We also moved the ETM suminng junctions from the inside of the lsc block to the top level as was done in LLO. For now the ITM outputs are terminated as we do not have IPC senders for them.
As Nic mentioned above, we had to change a few guardian codes such that they are adapted to the new LSC input matrix. I edited the following codes:
I basically replaced all the old matrix (LSC-ASPD_DOF) by the old style PD_DOF matrix in all the above three codes. We had a chance to test ALIGO_IFO and set it to the MICH_DARK state, which seems to be working fine so far. I did not get a chance to test any other states of ALIGN_IFO or the reset of the two codes yet.
J. Kissel, J. Betzwieser, A. Mullavey, D. Macleod Work Permit #5313 Two changes to the h1calcs front end model that I acquired from simply updating the library parts which had been modified by Joe and Duncan at LLO: (1) In the DARM / DELTAL calibration infrastructure (i.e. the CAL_CS_MASTER.mdl library part), moved the actuation path's integer cycle time delay to "after" the actuation function, such that the delay is only applied when DELTAL_RESIDUAL and DELTAL_CTRL are summed to form DELTAL_EXTERNAL. This way, the GDS pipeline can pick up DELTAL_RESIDUAL and DELTAL_CTRL, apply the necessary corrections (namely the super-nyquist frequency poles / zeros from AA, AI, ESD Driver, and OMC whitening), and sum them together with a more accurate (i.e. not limited by integer clock cycle) delay to form GDS-CALIB_STRAIN. See LLO aLOG 18845 (2) In the hardware injection infrastructure (i.e. the CAL_INJ_MASTER.mdl), modified the ODC bit comparator to determine whether a hardware injection is present from "is HARDWARE_OUT == 0" to "is HARDWARD_OUT <= 1e-200." Further, the status of each filter in the hardware injection path is compared against new EPICs values called "${BANKNAME}_CTRL_EQ" and "${BANKNAME}_GAIN_EQ." These values will be set (and accepted into the SDF system) by the ODC team, but this has not been done yet. See LLO aLOG 18913. These changes also come with corresponding updates to the MEDM screen overviews for each that reflect the new organization.
Swapped EtherCAT Corner Station Chassis 6 in CER. Old unit S1400076 was swapped with modified unit S1400077. Modified unit had Beckhoff terminals added for EOM drivers per E1300689-v5.
Daniel, Patrick We updated the system manager to reflect the change. We ran into trouble with PLC2, which must have gotten new code in the restart. Daniel had to change the code for the demodulator library to prevent taking the log of a negative number. I burtrestored to 6:10 this morning.
Leak tested RGA joints - OK Also, moved X-end crane to "Parking" position at ~1010 hrs. local
Hugh and I went to HAM3 this morning to see if we could find the cause of the 23 hz line. Finding nothing at the chamber, we decided to start turning stuff off to see if we could get it to go away. Taking the ISI and HEPI down did nothing, but turning the suspension off made the line go away. I then started turning the SUS back on and as soon as I took MC2 to aligned, the 23hz peak showed up. Jeff said that it was likely the sat amplifier needed power cycled, so Hugh is filing an FRS and this is being handed off to Richard. Attached image shows HAM3 GS-13s between 20 and 30 hz. Red is from last night, when everything was on, blue is everything off, green is MC2 aligned (everything else off). HEPI L4Cs and both SUS also saw the peak.
After all the restarts earlier and Richard power cycling MC2, the line is gone. See attached. Red trace is from 8 hours ago, blue is from a couple minutes.
Do we know what power cycling does? how often dows this kind of thing happen?
Richard--The thought/idea is and maybe there is some way we could verify it, that the satellite amplifier goes into oscillation.