J. Kissel Over the weekend, Evan, Stefan and Kiwamu has explored running the IFO at 24 [W] request PSL input power (see e.g. LHO aLOG 18923) instead of 17 [W] for which the interferometer had been calibrated (see LHO aLOGs 18769 and 18813). Because we do not yet have automatic optical gain scaling built into the IFO's control system, the calibration will be incorrect for the following science segments: Seg Num GPS Start GPS End Duration [s] UTC Start UTC End 1 1117601338 1117665437 64099 Jun 06 2015 04:48:42 Jun 06 2015 22:37:01 2 1117667359 1117702258 34899 Jun 06 2015 23:09:03 Jun 07 2015 08:50:42 3 1117709888 1117743019 33131 Jun 07 2015 10:57:52 Jun 07 2015 20:10:03 4 1117748054 1117773188 15134 Jun 07 2015 21:33:58 Jun 08 2015 04:32:52 5 1117803358 1117814235 10877 Jun 08 2015 12:55:42 Jun 08 2015 15:56:59 6 1117814313 1117815464 1151 Jun 08 2015 15:58:17 Jun 08 2015 16:17:28 7 1117829201 1117833161 3960 Jun 08 2015 20:06:25 Jun 08 2015 21:12:25 --- Total Time Uptime@23[W] Single IFO Duty Factor During These Segments 64.40 hr 45.27 hr 70.3% Offline optical gain calculations are being made by the GDS calibration pipeline, but they are not being applied directly since this is the first time they've been calculated. However, evidence still suggests that LHO's DARM coupled cavity pole frequency (i.e. the single-pole approximation to the interferometer's response to gravitational waves / displacement noise) is still a moving target, so the calibration error (not uncertainty, but actual error) may not only just be a scale factor, but a frequency-dependent error. We should* have enough information from PCAL and DARM calibration lines to make an estimate of how the frequency dependence is changing over time. *"should" is still "in principle;" we have not yet finished the commissioning of processing PCAL / DARM calibration lines to a point where we can determine at what precision the 6 lines will be able to determine the optical transfer function. This work is on-going.
"Incorrect" sounds too strong to me.
I would say it was incorrect only in the sense that the cavity pole frequency was uncertain which is the case not only for the 24 W configuration but also for the 17 W. Otherwise we believe that the calibration had remained valid both in GDS and CAL-CS at 24 W (though, rememeber CAL-CS has not fully updated to the equivalent of GDS, alog 1880 and hence the descrepancy between them). The OMC has a power-scaling functionality (alog 18470) and therefore, ideally it does not change the optical gain as we change the PSL power. As for the cavity pole frequency, the Pcal lines should be able to tell us how stable it has been.
As reported in alog 18293, the optical gain seemed to have dropped by 4% in this particualr lock according to the Pcal line at 540 Hz. Sudartian is currently analyzing the Pcal trend, but it seems that the optical gain typically changes by 4-5 % in every lock stretch probably due to different OMC error gain (which is computed in every RF->OMC transition) and perhaps different alignment somewhere. We compensated it by increasing OMC-READOUT_ERR_GAIN by 4 % at the beggining of this particular lock and therefore we thought the calibration was good assuming the cavity pole stayed at the same frequency, 355 Hz.
I suspect there is a lingering source of error in the gain of the OMC-DCPD --> DARM_IN1 path. This may be due to the initial gain-matching calculation between DCPD_SUM and RF-DARM, but it could also be due a scaling error as we adjust the overall gain during the power-up step. We initially set the gain at a DARM offset of ~40pm, but as we power up to 23W we reduce the offset to ~15pm. The current gain-scaling calculation that Kiwamu links to does not account for the small static DARM offset that we have observed (it's a fraction of a picometer, see here). I will post a note about this today -- the overall effect should be very small, but may account for the ~4% change that we have observed. (If this is the source of the gain error it will be proportional to DARM offset, which is the same as power level since we change both at the same time.)