J. Kissel
WP 10553
ECR E2200291
IIET 12638

After fixing the issues with the CAL-CS front-end model infrastructure (see LHO:64000) I'm now able to extend it to add a new capture of the channel used as the "reference" PCAL in the corner station, as per ECR E2200291 / IIET Ticket 12638. In short, the change is so that out-of-loop, non-real-time down-stream consumers (like the GDS-CALIB_STRAIN pipeline) only need to know one channel name for the PCAL we use as "the" reference channel, and that channel will always get corrected with the exact same factors whether it's processed by the front-end to produce time-dependent correction factors, or used in front-end subtraction algorithms, or the down-stream consumers. In the past, front-end stuff consumed the reference PCAL channel in the corner, where as down stream consumers grabbed the channel from the end station, resulting in a one 16 kHz clock-cycle difference between the two versions. Small enough to get forgotten, but large enough to matter to a "2%, 2 deg" level of systematic error in the calibration.

Attached are the "before" vs. "after" screenshots. The "before" are the first three, from the top level drilling down to where the start of the change is, and the "after" are the second three, from the start of the change "up and out."

List of changes: 
    (1) the CAL_CS_MASTER library part, where I 
        (a) inside the "TDEP" block inside the main "CS" block of the library, added a pick-off downstream of the PCAL_REF_END_SW and PCAL_REF_TR_SW which chose which EX or EY, then RX or TX PD is used as "the" reference PCAL channel,
        (b) piped this pick-off "up and out" to the "CS" level as the first output of the TDEP block (which shifts all the existing outputs down by one)
        (c) At the "CS" level, connected the REF_PCAL_Out output of the TDEP block to a new "DELTAL_PCAL_OUT" tag, and connected that tag to another "up and out" output, importantly choosing for this output to be output 9. This number places this PCAL output at the end of the "production" channels, segregating it from the R&D projects like the DCPD A and DCPD B calibrated out, and the "Compensated For Time Dependence" or CFTD pipeline.

    (2) then, at the h1calcs top-level model, I
        (a) shifted all channels below output 9 down, 
        (b) created a new EPICs channel and fast channel test point, DELTAL_REF_PCAL_SLOW and DELTAL_REF_PCAL, respectively, and
        (c) added DELTAL_REF_PCAL to the DAQ channel list, such that this channel will appear in the frames as 
    H1:CAL-DELTAL_REF_PCAL_DQ.

The changes have been built on the build machine, and are ready to go for install, restart, and a DAQ restart tomorrow.

In for LLO to receive these changes, they must:
    - svn up 
        /opt/rtcds/userapps/release/cal/common/models/CAL_CS_MASTER.mdl
        
to revision r23296
    - For visual comparison, svn up 
        /opt/rtcds/userapps/release/cal/h1/models/h1calcs.mdl
        
to revision r23297
    - Shift the all the top-level output connections in l1calcs.mdl starting at DELTAL_A_SLOW + DELTAL_A output connections down by one. The last existing connection should be from "TDEP_CORR_DELTAL_CTRL_L3_WHITENED" to the channels "CFTD_DELTAL_CTRL_TST_SLOW+CFTD_DELTAL_CTRL_TST".
    - Add terminators to the last three "new" outputs called "CAL_PRCL_NO_WHITEN," "CAL_SRCL_NO_WHITEN," "CAL_MICH_NO_WHITEN," as per the last sentence in LHO:62102 and discussion in LHO:64000.