evan.goetz@LIGO.ORG - posted 19:09, Thursday 18 August 2016 - last comment - 13:59, Friday 19 August 2016(29189)
Measured OMCDCPD AA transfer functions
Evan G., Jeff K. Summary: We measured the transfer functions of the OMC DCPD anti-aliasing (AA) module paths of ch13-16 for chassis S1102788. We measure this because the filter board was modified for these channels (LHO aLOG 28010). This is similar to the measurement Kiwamu made in LHO aLOG 21123. The OMC DCPD AA channels are 13 and 14 for DCPD A and B, respectively. The OMC PI AA channels are 15 and 16. We find that the notch behaviour of channel 13 matches what Kiwamu found in LHO aLOG 21123, but the notch of channel 14 is distorted (broken?). Channels 15 and 16 do not have the same notch behaviour (as to be expected for the PI paths, matching LHO aLOG 28085). These differences--compared to the calibration model reference AA filter--are below 1% in magnitude and less than 1 degree in phase below 7 kHz. While the calibration group is unaffected by the broken channel 14, noise from 65 kHz will be aliased down into the detection band. We should consider fixing this. Details: The setup for the measurements is shown in the first attachment, and the reference measurement (to remove the gain of the single ended to differential box) is shown in attachment 2. Each transfer function measurement is normalized by the reference measurement transfer function. Data is saved in /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O2/H1/Measurements/OMCAAChassis Analysis script is saved as /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/O2/H1/Measurements/OMCAAChassis/process_aachassis_data_20160818.m Plots are shown in the third attachment. Of particular note is to compare the new measurements to the AA filter model. Channel 14 (OMC DCPD B) now has a different behaviour near the notch. The difference is below 1% in magnitude and 1 degree in phase below 7 kHz, and does not force a revision in the calibration model reference AA filter.
Non-image files attached to this report
Comments related to this report
J. Kissel, D. Sigg, R. McCarthy We haven't really quantified what sort of super-Nyquist frequency junk lies around this notch that might cause aliased noise in the detection band. Thus we don't really know whether this abnormal notch is "good enough," -- but we also don't know so for any functional/normally behaving notches either (e.g. the CAL model reference or DCPD A / CH13's response). The best I've seen is Carl's study with an SR785, seen in LHO aLOG 28611, but the frequency axis of his plot leaves one with desire. In either case, we think it prudent to just fix the notch, so as to not leave this hanging chad lying around, in case the abnormal response is indicative of a component failure (gradual or otherwise). For bookkeeping purposed, I've opened FRS Ticket 6071. We think it sufficient to wait until next Tuesday to fix it; no emergency here.