jeffrey.kissel@LIGO.ORG - posted 14:24, Tuesday 11 June 2024 - last comment - 17:29, Monday 17 June 2024(78372)
Post OMC-Swap DCPD Transimpedance Amplifier Update :: Finally! A Good Measurement Data Set!
J. Kissel TIA D2000592: S/N S2100832_SN02 Whitening Chassis D2200215: S/N S2300003 Accessory Box D1900068: S/N S1900266 SR785: S/N 77429 I've finally got a high quality, trustworthy, no-nonsense measurement of the OMC DCPD transimpedance amplifiers frequency response. For those who haven't seen the saga leading up to today, see the 4 month long story in LHO:77735, LHO:78090, and LHO:78165. For those who want to move on with their lives, like me: I attach a collection plots showing the following for each DCPD: Page 1 (DCPDA) and Page 2 (DCPDB) - 2023-03-10: The original data set of the previous OMC DCPD's via the same transimpedance amplifier - 2024-05-28: The last, most recent data set before this, where I *thought* that is was good, even though the measurement setup was bonkers, - 2024-06-11: Today's data Page 3 (the Measurement Setup) - The ratio of the measurement setup from 2023-03-10 to 2024-06-11. With this good data set, we see that - there's NO change between the 2023-03-10 and 2024-06-11 data sets at high frequencies, which matches the conclusions from the remote DAC driven measurements (LHO:78112) and - there *is* a 0.3% level change in the frequency response at low frequency, which also matches the conclusions from the remote DAC driven measurements. Very refreshing to finally have agreement between these two methods. OK -- so -- what's next? Now we can return to the mission of fixing the front-end compensation and balance matrix such that we can - reduce the impact on the overall systematic error in the calibration, and - reduce the frequency dependent imbalance that were each discovered in Feb 2024 (see LHO:76232). Here's the step-by-step: - Send the data to Louis for fitting. - Create/install new V2A filters for A0 / B0 bank - Switch over to these filters and accept in SDF - Update pydarm parameter file with new super-Nyquist poles and zeros. - Measure compensation performance with remote DAC driven measurement of TIA*Wh*AntiWh*V2A confirm bitterness / flatness Once IFO is back up, running, (does it need to be thermalized?) - Measure balance matrix, Remember -- SQZ OFF confirm better-ness / flatness - Install new balance matrix - Accept Balance Matrix in SDF Once IFO is thermalized - grab a new sensing function. - push a new updated calibration
Non-image files attached to this report
Comments related to this report
The data gathered for this aLOG lives in:
/ligo/svncommon/CalSVN/aligocalibration/trunk/
Common/Electronics/H1/DCPDTransimpedanceAmp/OMCA/S2100832_SN02/20240611/Data/
# Primary measurements, with DCPD TIA included in the measurement setup (page 1 of the main entry's attachment measurement diagrams)
20240611_H1_DCPDTransimpedanceAmp_OMCA_DCPDA_mag.TXT
20240611_H1_DCPDTransimpedanceAmp_OMCA_DCPDA_pha.TXT
20240611_H1_DCPDTransimpedanceAmp_OMCA_DCPDB_mag.TXT
20240611_H1_DCPDTransimpedanceAmp_OMCA_DCPDB_pha.TXT
# DCPD TIA excluded, "measurement setup" along (page 2 of the main entry's attachment measurement diagrams)
20240611_H1_MeasSetup_ThruDB25_PreampDisconnected_OMCA_DCPDA_mag.TXT
20240611_H1_MeasSetup_ThruDB25_PreampDisconnected_OMCA_DCPDA_pha.TXT
20240611_H1_MeasSetup_ThruDB25_PreampDisconnected_OMCA_DCPDB_mag.TXT
20240611_H1_MeasSetup_ThruDB25_PreampDisconnected_OMCA_DCPDB_pha.TXT
Here are fit results for the TIA measurements
DCPD A:
Fit Zeros: [6.606 2.306 2.482] Hz
Fit Poles: [1.117e+04 -0.j 3.286e+01 -0.j 1.014e+04 -0.j 5.764e+00-22.229j 5.764e+00+22.229j] Hz
DCPD B:
Fit Zeros: [1.774 6.534 2.519] Hz
Fit Poles: [1.120e+04 -0.j 3.264e+01 -0.j 1.013e+04 -0.j 4.807e+00-19.822j 4.807e+00+19.822j] Hz
A PDF showing plots of the results is attached as 20240611_H1_DCPDTransimpedanceAmp_report.pdf. The DCPD A and B data and their fits (left column) next to their residuals (right column) are on pages 1 and 2, respectively. The third page is a ratio between DCPD A and DCPD B datasets. Again, they're just overlaid on the left for qualitative comparison and the residual is on the right.
I used iirrational. To reproduce activate the conda environment I set up specifically just to run iirrational.
activate /ligo/home/louis.dartez/.conda/envs/iirrational
Then run
python /ligo/groups/cal/common/scripts/electronics/omctransimpedanceamplifier/fits/fit_H1_OMC_TIA_20240617.py
A full transcript of my commands and the script's output is attached as output.txt.
On gitlab the code lives at https://git.ligo.org/Calibration/ifo/common/-/blob/main/scripts/electronics/omctransimpedanceamplifier/fits/fit_H1_OMC_TIA_20240617.py
Non-image files attached to this comment
Here's what I think comes next in four quick and easy steps: 1. Install new V2A filters (FM6 is free for both A0 and B0) but don't activate them. 2. Measure the new balance matrix element parameters (most recently done in LHO:76232. 3. Update L43 in the pyDARM parameter file template at /ligo/groups/cal/H1/ifo/pydarm_H1.ini (and push to git) N.B. doing this too soon without actually changing the IFO will mess up reports! Best to do this right before imposing the changes to the IFO to avoid confusion. 4. When there's IFO time, ideally with a fully locked and thermalized IFO: 4.a move all DARM control to DCPD channel B (double the DCPD_B gain and bring the DCPD_A gain to 0) 4.b activate the new V2A filter in DCPD_A0 FM6 and deactivate the current one 4.c populate the new balance matrix elements for DCPD A (we think it's the first column but this remains to be confirmed) 4.d move DARM control to DCPD channel A (bring both gains back to 1, then do the reverse of 4.a) 4.e repeat 4.b and 4.c for DCPD channel B then bring both gains back to 1 again 4.f run simulines (in NLN_CAL_MEAS) and a broadband measurement 4.g generate report, verify, and if all good then export it to the front end (make sure to do step 3. before generating the report!) 4.h restart GDS pipeline (only after marking report as valid and uploading it to the LHO ldas cluster) 4.i twiddle thumbs for about 12 minutes until GDS is back online 4.j take another simulines and broadband (good to look at gds/pcal) 4.k back to NLN and confirm TDCF's are good.