Follow up to 83605.
I created a few templates to do a SR3 OSEM calibration. I called them
sus/trunk/HLTS/H1/SR3/Common/Data/
yyyy-mm-dd_hhmm_H1ISIHAM5_ST1_WhiteNoise_ISO_{X,Y,Z}_0p05to40Hz_calibration.xml
The measurement should be run with all gains on SR3_M1_DAMP_{L,T,V,R,P,Y} set to - 0.1 . Just to make sure the loop gain does not mess with the OSEM measurement of the cage above resonance.
The idea is that we ought to be able to only use the cartesian measurements of the ISI to calibrate all the OSEMs in the suspensions attached by using linear algebra. This is the first pilot for that idea. We want about 100 points between 5-10Hz to make sure we can make a "decent" fit.
___
When Oli runs these templates, the plan is to use them to calibrate the SR3 OSEMs for the estimator work (see 84171 for the latest news as of the making of this post).
We will use Oli's feedback to work out the kinks of the calibration procedure and to export it to other ISIs at a later time.
I've run these three transfer function measurements with settings as follows:
The measurements can be found in /ligo/svncommon/SusSVN/sus/trunk/HLTS/H1/SR3/Common/Data/as 2025-05-07_0000_H1ISIHAM5_ST1_WhiteNoise_ISO_{X,Y,Z}_0p05to40Hz_calibration.xml. They have been committed to the svn as r12295.
Checking the coherence between each excitation channel (H1:ISI-HAM5_ISO_{X,Y,Z}_EXCMON) and SR3's euler basis DOFs, the coherence between 5 and 10 Hz was good between the following channels:
ISO_X -> DAMP_T
ISO_Y -> DAMP_L
ISO_Z -> DAMP_V, but also pretty good coherence with DAMP_L and DAMP_T
I have run the calibration script to get the calibration factors for the SR3 OSEMs. The (multiplicative) factors are:
LF: 1.3206
RT: 1.3705
SD: 1.3806
T1: 2.4539
T2: 1.4819
T3: 1.4125
The factors are calculated by fitting a frequency-independent real factor to the measured uncalibrated transfer functions to fit the modeled one. The fit range is from 5 to 15 Hz. The attached plots show the before and after of the process.
A few observations:
1) As we had seen before, the Left and Right (LF and RT) OSEMs are pretty well matched to one another, which bodes well for our PR3 Yaw OSEM estimator plans.
2) The T1|T2|T3 OSEMs that measure the vertical direction have their orientation flipped respect to the model. I don't think there's a problem, but it is worth pointing out.
3) The T1 OSEM, which is used to distinguish roll is 70% off from T1 and T2. The large factor feels concerning, but it is consistent with other measurements of SR3. See for example 83963, where in the attached .pdf the Vertical to roll coupling is very clean (page 12). And the difference of T1, T2, and T3 can be seen very directly in page 24.
- I think it might be worth revisiting some of these transfer functions after recalibrating just to confirm that we're getting the numbers right.
If we wanted the OSEM measurements to match with the GS13 calibrations between 5-15 Hz, the factors that should be in the OSEM input filters should be:
T1: 3.627
T2: 1.396
T3: 1.345
LF: 1.719
RT: 1.490
SD: 1.781
Since this change will increase the effective gain of the damping loops by a factor of 30% at a minimum, we must compensate by changing the gains for the filter modules in the M1 DAMP filter bank.