S. Dwyer, S. Karki, J. Kissel
We've gathered several sensing function measurements today in order to resume the continuing saga of the low frequency regime of sensing function.
The raw data has been committed to the CAL repo here:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Measurements/FullIFOSensingTFs/
2019-11-04_H1_NominalConfig_DARM_OLGTF_LF_SS_5to1100Hz_15min.xml
2019-11-04_H1_NominalConfig_PCALX2DARMTF_LF_SS_5t1100Hz_10min.xml
2019-11-04_H1_NominalConfig_PCALY2DARMTF_LF_SS_5t1100Hz_10min.xml
2019-11-04_H1_NominalConfig_PCALX2DARMTF_BB_3min.xml
2019-11-04_H1_NominalConfig_PCALY2DARMTF_BB_3min.xml
2019-11-04_H1_DHARDPx1p5_DARM_OLGTF_LF_SS_5to1100Hz_15min.xml
2019-11-04_H1_DHARDPx1p5_PCALY2DARMTF_LF_SS_5t1100Hz_10min.xml
Sudarshan will aLOG more about the impact of ASC (DHARD P), but I can conclude a few things based on the rest of the data:
(1) The new, 50 ct SRCL offset is not the perfect solution. Comparing 2019-10-31 data with today's data at the same SRCL offset shows quite a different response.
(2) Even though we've updated the PCAL calibrations, PCALX vs. PCALY still seems to be discrepant from each other by a similar small percentage.
(3) Though the optical gain is about the same as the 20190909 reference model, the cavity pole is lower, back down in the 410 Hz region we were back in the beginnings of O3A.
The > 20 Hz fit results of the PCALY Nominal Config data is
Optical gain, H_c (ct/m) | 3.156e+06 (+1111,-1106) or (+0.0352%,-0.03504%)
Optical gain, H_c (mA/pm) | 4.301 (+0.001514,-0.001507) or (+0.0352%,-0.03504%)
Cavity pole, f_cc (Hz) | 410 (+0.7372,-0.7409) or (+0.1798%,-0.1807%)
Detuned SRC spring frequency, f_s (Hz) | 0.2008 (+0.1228,-0.06052) or (+61.15%,-30.14%)
Detuned SRC spring quality factor, Q_s | 1.655 (+3.894,-4.288) or (+42.5%,-38.59%)
Residual time delay, tau_c (usec) | -0.02668 (+0.4573,-0.4558) or (+-1714%,--1708%)
kappa_c = 0.992245
f_cc = 410
f_s = 0.2008
Q = 1.655
Remember, our current model has no detuning, and since we fit the data only above 20 Hz, the above numbers for the spring frequency are meaningless. And as we continue to find out, there are many more effects going on at low frequency, so it doesn't make sense to fit the data as "just" an optical spring. But, until we understand what's going on better (and preferably, with that knowledge, just get rid of it), we don't both updating the analysis infrastructure and still "report" a "spring frequency and Q."
The first attachment shows today's nominal config against previous measurements.
The second attachment shows PCALX vs. PCALY measurements of today vs. 2019-08-08 when we last measured the two simultaneously.
The third and fourth attachments show the results of 20-5000 Hz MCMC fit of the data.
Things to do from here:
(a) Keep measuring the sensing function, to see if there's some SRCL offset that might be better suited.
(b) Understand the PCALX vs. PCALY data, perhaps by invoking the new PCALX vs. PCALY calibration line (see LHO aLOG 51915) to see if the information from that line is consistent with the sweep data.
(c) See Sudarshan's aLOG about the ASC plan.
The plots presented in this aLOG have been created by
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/FullIFOSensingTFs/
process_sensingmeas_collection_20191104_SRCLOffsetAgain.py
process_sensingmeas_collection_20191104_PCALXvsPCALY.py
process_sensingmeas_20191104.py
which have been committed to the CalSVN repo as well.
The results of the sensing function measurement with Nominal ASC loop configuration is compared with the sensing function measurement taken with ASC DHARD Gain set at 1.5 times the nominal gain of the loop. The measurement with ASC DHARD gain was interrupted by the Tonga earthqauke so we were able to get measurement upto only 9 Hz. The plots are attached, the first plot is normal calibration freq range and the second plot is zommed in from 3-50 Hz. For comparison the data from 2019-07-10, when similar measurmenets were made are also overlayed in the plot (LHO alog 50498). These measurements show that the sensing function below 10 Hz depends on some combination of ASC gain and SRCL offset. More thinking and measurements will require to understand this fully.
In future:
1. We would like to get better open loop gain measuremnt of DHARD loop to better understand its UGF and to see if what we are seeing is gain-peaking.
2. See if we can reduce angle to length coupling with frequency dependent A2L filter.
The script used to anlayze these latest measurements are here:
/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O3/H1/Scripts/FullIFOSensingTFs/process_sensingmeas_collection_20191104_ascgainchange.py