jeffrey.kissel@LIGO.ORG - posted 13:58, Wednesday 26 April 2023 - last comment - 12:13, Friday 28 April 2023(69047)
CAL-CS and GDS Calibration Updated
J. Betzwieser, L. Dartez, E. Goetz, J. Kissel, J. Rollins
The details are happening all too fast in order to write them all down here but at the 10000 foot level, as of 2024-04-26 19:54 UTC, we've updated
- the primary CAL-CS variables that create CAL-DELTAL_EXTERNAL_DQ, which correspond to the interferometrically-measured free parameters (optical gain, cavity pole, actuation gains for TST, PUM, and UIM changes)
:: Values are informed by last Thursday's 2023-04-20 IFO measurement. No major change here, just pushing these to get them consistent with the rest of the pipeline. Not informed yet by the current state of the IFO, especially the TCS settings and SRCL offsets
- The "DARM model transfer function function values at calibration line frequencies" EPICs records
:: Importantly, this comes with changes that are the fall out of LHO:68880 and the solution pyDARM Merge Request 234
- The GDS-CALIB_STRAIN pipeline has new filters, and importantly its got a configuration change that now using the corner station copy of the PCAL, H1:CAL-DELTAL_REF_PCAL_DQ, as it's measure of the PCAL system rather than the end-station.
:: Again, this is the fall out of LHO:68880
Many details to come in the comments.
Here's a blow-by-blow of what happened: - Had been locked and powered up to 76W PSL input power for ~3:30 hours - Louis walked us through the calibration report for what would be pushed upon update from /ligo/groups/cal/data/H1/reports/20230420T014029Z/H1_calibration_report_20230420T014029Z.pdf - Took ISC_LOCK to NLN_CAL_MEAS - Took a "before change" PCAL broadband injection, /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/FullIFOSensingTFs/ 2023-04-26_1925UTC_H1_PCALY2DARMTF_BB_3min.xml - While the broadband was going, took inventory of what filter modules were available for us to place the new IFO-measured free parameters in the H1:CAL-CS_DARM_ERR H1:CAL-CS_DARM_ANALOG_ETMX_L1 H1:CAL-CS_DARM_ANALOG_ETMX_L2 H1:CAL-CS_DARM_ANALOG_ETMX_L3 filter banks. Found that FM1 and FM2 were available in DARM_ERR, and FM4s were available in the L1, L2, and L3 banks. - Edited the pydarm command line interface configuration / parameter file, /ligo/groups/cal/ifo/H1/pydarm_cmd_H1.yaml to chose these filter banks, and a new name for the filter. For the sensing function we chose "ER15_Gain" and "ER15_NoD2N" for the optical gain and cavity pole frequency in FM1 and FM2 of DARM_ERR. For the actuation strengths for the L1, L2, and L3 stages, we chose "Npct_ER15" in FM4. - Ran the command line interface for exporting the calibration model to the front end (currently using a curated path and pydarm branch) :: First, without actually pushing, aka a dry run, which spits out everything it's *going* to push as it stands. The first couple of lines of that output are the IFO-measured free parameters, copied here for future searchability Hc: 3.1796e+06 :: 1/Hc: 3.1450e-07 Fcc: 4.3371e+02 Hz Hau: 1.6072e+00 N/A :: 7.5454e-08 N/ct Hap: 3.0579e-02 N/A :: 6.2634e-10 N/ct Hat: 2.5464e-11 N/A :: 9.6474e-13 N/ct These are not that different from what was installed on 2023-04-20 01:15 UTC, see LHO:68856, so we don't expect that much change here. :: Then, we actually pushed those values, PYTHONPATH=/ligo/home/louis.dartez/repos/pydarm pydarm export --push which spit out to the commanding what it was changing, when it was changing it (but lacked any "done!" message) - Went CAL-CS model's GDS_TP screen, waited ~5 seconds until the "Modified Filter File Detected" message to appear, then hit "LOAD COEFFICIENTS" (i.e. pressed the H1:FEC-117_LOAD_NEW_COEFF button), and waited another ~10 seconds for it to complete - Went to each of the above mentioned filter banks and switched over to each of the new filters by hand; FM1 & FM2 in the DARM_ERR, and FM4s in ANALOG_ETMX_L1, ANALOG_ETMX_L2, and ANALOG_ETMX_L3. - Took another post-calibration update broad-band injection in order to see how much better / different the calibration is/was, /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/FullIFOSensingTFs/ 2023-04-26_1950UTC_H1_PCALY2DARMTF_BB_3min.xml - Louis, on the side, created / updated new PCAL and DELTAL calibrations for the DTT template, and put them in /ligo/home/louis.dartez/projects/20230426/ pcal_calib_dtt.txt deltal_external_calib_dtt.txt so, we loaded these in to the 1950UTC DTT template. Hopefully this gives us a more accurate answer for what the systematic error is without all the extra caveats we've had in recent history about "well, the DTT calibration is some old crap, that has an outdated version of the super-Nyquist corrections for DELTAL, so you can't trust the answer *exactly*." While we're not yet confident that you can trust the answer "exactly," we are confident it's at least better. To load in these files, :: Hit "Calibration" button to open calibration pop-up window :: Select H1:CAL-PCALY_RX_PD_OUT_DQ :: Hit "New" to open the labeling pop-up. . Enter in a name for the Teference (can be anything, "asdfsdf" for eg) . Enter in a Unit of "m" . Enter in a Time that's much earlier that now (hold down CTRL and click the down arrow once to change the year to 2022) . Hit OK to return to the calibration pop-up :: Unselect the check box in every other tab besides "Trans. Func." :: Under the "Trans. Func." tab, . Check the box . Hit Edit... to open up the transfer function editing window . From the toolbar, select File > Open... . Navigate to pcal_calib_dtt.txt, and double click it to select, close the pop-up, and have it populate in the TF editing window. . Hit "OK" to accept the new TF and close the TF editing window :: Hit "Set" to make sure this new calibration sticks. :: Repeat the process for H1:CAL-DELTAL_EXTERNAL_DQ using deltal_external_calib_dtt.txt See 2023-04-26_1950UTC_CalibrationUpdate_BBInjection.png which shows the change between the BB taken after the last calibration update on 2023-04-23, just before this update on 2023-04-26 19:25 UTC, and just after at 2023-04-23 19:50 UTC. This is after we've updated the calibration within the DTT template, so for the first time in a long time, the low-frequency end of phase of this transfer function makes sense "starting" at 0 deg. - We then went back to nominal low noise, turning the calibration lines back on so as to start measuring the TDCFs again. Even though we've moved around some delays and got rid of some DEMOD phase rotations, the TDCF answers remain about the same before and after the calibration change, as expected. This gives us confidence in what changes we made to pydarm as a result of last week's solutions to long standing mysteries about the EPICs records. Satisfied with this "first brush" answer, - Went to the SDF screen for the CALCS model, and accepted all the new settings (including all the changes to the "DARM model transfer function values at calibration line frequencies") -- see 2023-04-26_1950UTC_CalibrationUpdate_SDFAccept.png. - Jamie then used the FIR filters generated in that 2023-04-20 report, /ligo/groups/cal/data/H1/reports/20230420T014029Z/ gstlal_compute_strain_C00_filters_H1.npz to inform and restart the GST-LAL calibration pipeline, in some way that hopefully he'll document in further comments below.