jeffrey.kissel@LIGO.ORG - posted 15:56, Friday 16 April 2021 - last comment - 18:15, Friday 16 April 2021(58626)
2021-04-16 PCAL Estimate of Current Systematic Error in Calibration
C. Compton, J. Driggers, S. Dwyer, J. Kissel, G. Mansell As I'm dusting off the ol' calibration gears, but I can give one answer immediately -- the systematic error in CAL_DELTAL_EXTERNAL is large. See attached plot which compares a broad-band PCALY excitation injected in DARM now vs. in Sept 2020, when we believe we had a "no different than O3" IFO. Thanks be to Camilla and Jenne who were able to grab the following actuator measurements this morning, but we've been suffering from some lock losses in the afternoon, with the current suspect being not enough range in the ETMX L2 coil drivers after the afternoon winds have picked up. That'll help the debugging of *why* there's more systematic error than before, but it'll take some time. However -- that investigation leads me to suspect at least one thing that we know is different that would impact (create large) systematic error in calibration -- the ETMX L2 coil drivers were in a different state in this measurement (State 1) than we ran in O3 (State 3). I'm not sure how different the frequency response is of the leftovers of digital compensation of the coil driver response between State 1 (what we've been running on today) and State 3. Remember we're still using the O3B model in the front-end for everything to produce DELTAL_EXTERNAL (and GDS-CALIB_STRAIN) (same suspension actuator strengths, same assumption of no frequency dependent systematic error, same optical gain, and same cavity pole as O3 -- at least a few of these things are wrong).
Images attached to this report
Comments related to this report
We're at 25W input to the IFO, but the equivalent of 38W into the IFO in "O3 power units" because of the better power recycling gain we've seen thus far.
I was able to grab several more measurements this afternoon which explains lots of things immediately:
(1) We were able to *stay* locked when we left the PUM (L2) coil drivers in State 2 instead of State 1 (as had been the morning's configuration). So that's in what DARM actuator configuration we were in for these later measurements:
UIM/L1: State 1 (highest range),
PUM/L2: State 2 (highest range), and <<< here's what was different from this morning -- we were in State 1 (medium range).
TST/L3: State 2 (lowest noise).
I remeasured the broad-band DELTAL / PCAL transfer function in this different state and the reported systematic error in the response function is a *little* bit different. So, The PUM coil driver state does impact the level of systematic error in DARM calibration right now, but not that much, and mostly a few % / few deg below 100 Hz.
See Blue vs. Red in first screenshot attachment.
(2) I took the two standard measurements that are needed to measure the sensing function, and they reveal the cavity pole is now at 451.5 Hz, and the optical gain is ~10% higher than before at 3.473e6 [DARMIN1 ct / DARM m]. Previous O3 values were 410.9 Hz and 3.162e6 [ct/m], hence (3.473e6 - 3.162e6)/3.162e6 = 0.09835 = ~10%.
WOW!! This is likely a huge (if not *the*) contributor to the systematic error in the response function.
A fit to this evening's measurement is the .pdf attachment.
So -- easy fix to reduce the systematic error in the calibration right now:
(a) Update the "H1CAL-CS_DARM_ERR" filter bank to create a new pair of filters, in say, FM4 and FM5:
FM4: "PostO3_NoDe2N": zpk([451.5],[7000],1,"n")
FM5: "PostO3_Gain": gain(2.879e-7) == 1/3.473e6
(b) Then turn off FM2 and FM3, and turn on FM4 and FM5.
Note, this will break the TDCF calculation / compensation, and GDS-CALIB_STRAIN will still have some error as well, but it'll at least improve the accuracy of DELTAL_EXTERNAL in to the "pretty dang good!" range again.
Images attached to this comment
Non-image files attached to this comment
