Ibrahim ran a calibration measurement last night, report attached to this alog (again, the weird bug happened where it pulled the wrong report reference so I had to regenerate). Just from the first page, we can see that the PCAL/GDS transfer function, as measured by both PCAL X and Y, show the uncertainty to be around 1% again. We can see that the modeled actuation strength matches up very well with the current model that has been running since 8/28, indicating that I corrected the drivealign gain properly (ignoring all the mistakes I made in the middle).
The difference in the sensing function is mostly due to the loss of optical gain, this came with the power outage. There might be some small change in the spring, but it is very minimal, and that lines up with the fact that trying to adjust the SRCL offset yesterday only made things worse.
Comparing the GDS/PCAL broadband with previous measurements, we can see that this broadband (in red, Oct 10) is almost exactly the same at the broadband taken on Aug 28 (gray reference) when we pushed this particular calibration. There are small differences at low frequency, perhaps due to some change in the spring behavior of the sensing function, and at high frequency, which I don't have a good immediate reason for.
Given that correcting the TST actuation strength brought the calibration almost exactly back to the model, I spoke with Joe to understand then what the "mysterious" large deviation in the calibration was on Oct 4 (green trace). In the process, I also made a plot comparing the same measurement times, but instead looking at CAL DELTA L/PCAL, so examining what our calibration would look like without TDCFs (I matched the colors with teh GDS/PCAL plot). We determined that by comparing the CAL DELTAL measurement to the GDS measurement, indeed the TDCFs are doing something correctly; without them on Oct 4, our uncertainty woyld be over 6%. However, they are clearly not doing a perfect job of correcting us back to model. Joe thinks that this is because the TCDF correction process may not completely account for the effect of the DARM OLG between each of the actuation TDCF measurement points. We trended what happened to kappa UIM and kappa PUM when I adjusted the L3 drivealign gain to bring kappa TST back to 1 and we see that kappa UIM dropped by 0.9% and kappa PUM by 0.4%. This is happening because the DARM OLG is changing as we adjust kappa TST, and kappa UIM and kappa PUM are seeing some frequency dependent change, but their calculation is assuming that the DARM OLG is not changing.
In summary, the increase calibration uncertainty on Oct 4 is exactly due to the increase in kappa TST, even though the TDCFs were correcting GDS. We can can see by comparing the broadbands from Sept 27 (purple trace) and Oct 4 that at both times the TCDFs were doing an imperfect job of correcting GDS, and that scales with how large the deviation from 1 is; on Sept 27 kappa TST was about 1.02. Now that the actuation strength has been corrected, there is no need to update the calibration at this time.
