J. Kissel

I've processed the DARM loop sensing function measurement suite that Tony took this morning (LHO:63425), and the new is good; still confusing, but good.

See attached plot of the processed data compared against an MCMC fit of the data, and compared against the 2021 model (the last model from which we got good calibration, and the parameters of which are what's installed right now calibrating the detector).

One can immediately see that compared with the 2022-05-25 data from LHO:63405, that the data look nice and simple above 20 Hz -- well modeled with "just" an optical gain and a cavity pole. There's no extra odd frequency dependence up to 80 Hz, and below 20 Hz, we're "back" to the "known" mysterious behavior where we see (what we keep claiming we think is) L2A2L cross-coupling with the HARD ASC loops and some sort of optical spring from the detuning of the signal recycling cavity.

That means that a lot of the "favorite suspects" from LHO:63405 that I waved my hands around to explain all the frequency dependence  are actually ruled out:
    [At least above 20 Hz]
    - The OMC DCPD compensation appears to be actually quite fine
    - The OMC DCPD balance between the A and B PDs are actually quite fine
    - We can and should trust the new pyDARM infrastructure, Evan's done a great job.

Here're the results of the fit:
          Optical gain, H_c (ct/m)                 | 3.138e+06 (+2229,-2199) or (+0.07104%,-0.07006%)
          Optical gain, H_c (mA/pm)                | 4.999 (+0.003551,-0.003503) or (+0.07104%,-0.07006%)
          Cavity pole, f_cc (Hz)                   | 457.4 (+1.052,-1.044) or (+0.23%,-0.2282%)
          Detuned SRC spring frequency, f_s (Hz)   | 0.2301 (+0.2078,-0.09986) or (+90.31%,-43.39%)
          Detuned SRC spring quality factor, Q_s   | 2.191 (+3.034,-4.038) or (+72.21%,-54.27%)
          Residual time delay, tau_c (usec)        | 3.253 (+0.4981,-0.4966) or (+15.31%,-15.27%)

Now that the data looks "clean" above 20 Hz, I've only thrown out the data below 20 Hz. As before we must throw out any guesses the MCMC has at an optical spring, because the model we give the MCMC to use is entirely incomplete, since it doesn't have any of the L2A2L complexity.

But -- ya! good news! Let's stay on the all EX actuator (and fix the EY actuator in due time).
Note: The recommedations for what to do about the calibration pipeline were relavant only for if/when we stay on the split EY actuator. On the All EX actuator, the optical gain and cavity pole are almost the same as what they were in 2021, so we'd just be "refining" the sensing function calibration at the ~few % level if we updated to the above MCMC results. And I don't think we need to.

The other exciting news is that the residual time delay is only 3 [usec]. This is a catch-all term that we throw in that would be large if, say, we'd not gotten the five ~10 kHz poles from each OMC DCPD A and B right. But with Dripta's quality fits, indeed, we appear to have gotten things right! (And also, the new pyDARM infrastructure is still dividing out all the other usual known high frequency stuff like the analog AAs and digital AAs). So -- good news there too.

Now it's on to characterizing the actuators from LHO:63393.