Skip to content

Confirm / update EPICS records for PCAL/SUS phase rotations

Given the new infrastructure (see ECR E2200291) for the PCAL, we need to revisit the correct phase rotations needed (or simply zero phase rotation) to make sure that the PCAL correction EPICS records used for the TDCFs are correct for both GDS and CALCS.

Quoting @jeffrey-kissel from issue #82 (closed)

In the era where the pcal_correction was correcting the channel stored at the end-station, a part of the "EPICS records" was to install the pcal_correction at each PCAL calibration line frequency in the corner station copy of PCAL such that "we" could do the same demodulation of the PCAL line that the GDS pipeline does in the front-end. However, those EPICs records did not account for the 16 kHz clock cycle jump from the end to the corner. That caused the front-end TDCFs to give a different answer that the GDS TDCFs. Once we-realized it, we fudged the front-end calculation by installing a clock-cycle delay phase rotation in the DEMOD parts to fix it, which are more EPICs records. See this helpful diagram: 2019-03-06_MissingDelay_PCAL_TDCF_Calculation.pdf from G1601472, and this alog from when we installed these last (when H1 moved its calibration lines): LHO:48534

Now, in the new era, where

  • you've provided the option for a pcal_correction to be for either the corner or the end station (see Issue 17 and the completed Merge Request 129), and also
  • we now have "the" reference PCAL channel stored in the front-end (see ECR E2200291 installed on July 19th 2022; LHO:64049)

then if the GDS pipeline switches to consuming "the" PCAL from the corner station then the same pcal_correction can be used for both GDS and front-end calculations of the demodulated PCAL lines, and we won't need a separate additional install of "the 16 kHz clock cycle advance" phase rotation, which would be the additional EPICs record that this bullet point was requesting.

Regardless, we just need to get @joseph-betzwieser / @aaron-viets on the same page about all this, and make sure all code generates the right thing. The boolean option for "end or corner?" helps, 'cause then that puts the honorous of understanding on the script that produces the EPICS records.