A. Urban, on behalf of the calibration group

The C00 vs. C01 comparison pages have been updated and include all C01 data from 2016-11-30 through 2017-08-01.

As part of this analysis, I've got updated PCAL-to-DARM ratio trends at 36.7, 331.9 and 1083.7 Hz (attached below).

Plots of both magntidue and phase residual timeseries trends are attached. These trends are computed by demodulating 300-second segments at each PCAL line frequency, then averaging to get a magnitude and phase out of each channel. In practice, I break up each 300-second segment into overlapping 100-second chunks, apply a Kaiser window, and then average over chunks to get a less noisy measurement. (It's basically Welch's method, except I'm demodulating rather than computing a PSD.) Finally, I remove most outliers that are due to locklosses and loud transient glitches. The same analysis was done on Livingston data; see L1 aLOG 35501.

In each plot below I show PCAL/strain trends for C00 (blue) and C01 (red) data. The trend shown in aquamarine applies a correction for f_cc to C01 data at the appropriate PCAL line frequency. Gray shaded regions correspond to times when we had a break in O2, once during the holiday season and once for commissioning work.

At mid and high frequencies, everything continues to be in order at the PCAL lines. You can see that in the bucket (331.9 Hz) all systematics are accounted for by f_cc and only very small (~1%) statistical fluctuations remain in C01. At high frequency (1083.7 Hz) there's a slight systematic offset at the level of 1% or so in magnitude and 1-2 degree in phase, which is consistent with Craig's error budget. However, in the C01 low frequency line there is a mounting phase offset prior to the May vent that becomes worse after the break: the offset is 2-3% in magnitude and about 2 degrees in phase from June all the way through August. I suspect this has to do with spring detuning in the signal recycling cavity, and will follow up on it soon.