J. Kissel

Chugging down the front-end calibration's time-dependent correction factor to-do list, I used todays locks to tune the uncertainty thresholds and then watch the performance of the gating, median-ing, and averaging (GMA) that's a replica of what the GDS pipeline had done during O2. The message: I'm not satisfied.

Check out the attached ndscope traces of the three actuator kappas during one of this evening's nominal low noise lock stretches.
On the left column, in three plots (one for each UIM, PUm, and TST kappa), I show the raw input to the GMA system, the output of the gating, the output of the median-ing, and the output of the averaging, which we hope to eventually send off to correct DELTAL_EXTERNAL and to create CFTD_DELTAL_EXTERNAL. On the right, I show the calculated uncertainty of both the respective SUS_LINE and PCAL LINE used to derive the actuator kappas.

Observations: 
- The uncertainty is low; in the 0.05 % level (we could probably reduce the amplitude of the calibration lines by a factor of ~2.)
- The raw signal wobbles around smoothly with a 10 second period, reflective of the 0.1 Hz low pass used on the output of the DEMODs.
- Given that at the time I show, the live uncertainty is passing the tuned threshold, the raw signal and the gated signal are identical. Good!
- Things start getting fishy once the signal is passed through the median function, stuff that looks like digital hair on what should otherwise be smooth portions, and there is an overall stair-step nature to it. I forget if we've studied whether (a) the O2, GDS/DCS produced, median-ed time series looked this step-like, and (b) if the application of step functions at the 0.1% level caused any high-frequency noise or glitch problems. There also appears to be a significant offset between the raw and gated value vs. the median-ed value.
- Finally, with the final, gated, median-ed, and averaged output, there's a lag between it and the input, only-gated-median-ed time series. At least the digital hair has been smoothed out...

Plenty of things to think about before I understand it enough to be happy about trying to apply these factors to DELTAL_EXTERNAL.