J. Kissel
ECR E2500017
IIET Ticket IIET:33143
WP 12302

As a follow-on to LHO:82509, where I installed the infrastucture that down-sampled the new-ish 524 kHz, filtered, pick-off paths of the OMC DCPDs in the same way as the primary GW path, here is the explicit list of channels that are now available. 

Be forwarned -- unlike the primary path 16 kHz channels, the configuration of the ADC input matrix and filters of the 524 kHz test banks are *not* controlled by the SDF system, and are not a part of the DARM loop, so "we" can do whatever we want with them, even during observing without any impact on or change in the detector. In other words, you aren't guaranteed that these 16 kHz test channels will have the same signal processing forever. Please either check we me -- or preferably learn how to look at the configuration of EPICs records and filter modules yourself -- so you understand what to expect in the test pick-off channels when you compare them against the their primary path counter part.
I can at least promise when *I* make a change to these channels, it will be explicitly aLOGged, as I have been doing (something like LHO aLOGs 82384 or 82313). But -- I can't encourage this enough -- all EPICs records and filters are saved at all times always and forever, so being able to lookup and reconstruct the configuration from *that* is the end-all-be-all.

For now, the test paths are set up to produce results like in LHO:82512, and I outline the differences between the paths below.

All of these channels are calibrated into [mA] on the DCPDs. We can work with you if you need these in different units, but for line hunting and comparative analyses like ASD ratios and transfer functions the units really shouldn't mean much more to you than what you should label your y-axis.

It may be useful to reminder yourself of the analog side of these DCPD channels using the diagram from LHO:67644. Namely, below I mention "4CH average of analog voltage" because of the way the voltage from each DCPD is copied in analog and digitized on 4 separate ADC channels with the special "AA" chassis, D2300115, which is a normal AA chassis stripped of all filtering to be instead a pass-through-and-copy chassis.


H1:OMC-DCPD_A_OUT_DQ     
H1:OMC-DCPD_B_OUT_DQ      
 
    - primary GW path versions of DCPD A and B, 
    - 16 kHz down-sampled version of 524 kHz channels, because there's no filtering or gains applied in these 16 kHz banks
    - 4CH average of analog voltage, 
    - Stored in the frames at single-precision, but not limited by single precision noise below the ~7 kHz.
    - has NO 1 Hz 5th-order elliptic high-pass filter, or else the DARM loop would be terribly unstable and the IFO wouldn't work.
    - as of Friday 2025-01-24 has 2x 16kHz digital AA filters, and 1x digital 65kHz AA filters
    - BLUE traces in LHO:82512
          
H1:OMC-DCPD_16K_A1_OUT_DQ     
H1:OMC-DCPD_16K_B1_OUT_DQ   
 
    - first copy test path of DCPD A and B, 
    - 16 kHz down-sampled version of 524 kHz channels, because there's no filtering or gains applied in these 16 kHz banks
    - 4CH average of analog voltage, (by setting H1:OMC-ADC_INMTRX_1_9 and _2_10 element to 1.0, with all others in the _1 and _2 row set to 0.0)
    - has 1 Hz 5th-order elliptic high-pass filter ON to minimize impact of single-precision noise (in the 524 kHz version that's NOT stored in the frames).
    - as of Friday 2025-01-24 has NO digital AA filters
    - BROWN traces in LHO:82512
    
H1:OMC-DCPD_16K_A2_OUT_DQ   
H1:OMC-DCPD_16K_B2_OUT_DQ  

    - second copy test path of DCPD A and B
    - 16 kHz down-sampled version of 524 kHz channels4CH average of analog voltage, because there's no filtering or gains applied in this 16 kHz bank 
    - 4CH average of analog voltage, (by setting H1:OMC-ADC_INMTRX_3_9 and _4_10 element to 1.0, with all others in the _3 and _4 row set to 0.0)
    - has 1 Hz 5th-order elliptic high-pass filter ON to minimize impact of single-precision noise (in the 524 kHz version that's NOT stored in the frames).
    - as of Friday 2025-01-24 has 1x 16kHz digital AA filters, and 1x digital 65kHz AA filters, and thus 
    - a replica of how the primary path was set up PRIOR to 2025-01-24 (tho, the primary path never had the 1 Hz high-pass)
    - GREEN traces in LHO:82512


H1:OMC-DCPD_SUM_OUT_DQ
H1:OMC-DCPD_16K_SUM1_OUT_DQ     
H1:OMC-DCPD_16K_SUM2_OUT_DQ 
 
    - These are the primary path and two copies of the SUM of the DCPDs, "summed" with specific balancing matrix coefficients, i.e. the channels
        H1:OMC-DCPD_MATRIX_1_1 and _1_2
        H1:OMC-DCPD_MATRIX1_1_1 and _1_2
        H1:OMC-DCPD_MATRIX2_1_1 and _1_2
 
    - At the moment, the A and B inputs to the matrix have all the configuration notes of the corresponding A and B channels mentioned above, and there is no filtering or gain applied in the TEST SUM banks just like in the primary path.

    - These channels are directly proportional to DARM_ERR [counts], which is eventually calibrated into H1:GDS-CALIB_STRAIN (and all downstream "cleaned" products). Unfortunately, that coefficient of proportionality changes at the ~1% level every lock stretch, or else I'd quote it explicitly, but it's roughly 2.47e6 [mA/ct] or 4.05e-7 [ct/mA].
 
H1:OMC-DCPD_NULL_OUT_DQ       
H1:OMC-DCPD_16K_NULL1_OUT_DQ  
H1:OMC-DCPD_16K_NULL2_OUT_DQ    

    - Similar to the SUM channels, these are the primary path and two copies of the DIFFERENCE of the DCPDs, "subtracted" with specific balancing matrix coefficients, i.e. the channels
        H1:OMC-DCPD_MATRIX_2_1 and _2_2
        H1:OMC-DCPD_MATRIX1_2_1 and _2_2
        H1:OMC-DCPD_MATRIX2_2_1 and _2_2

    - Here as well, at the moment, the A and B inputs to the matrix have all the configuration notes of the corresponding A and B channels mentioned above, and there is no filtering or gain applied in the TEST SUM banks just like in the primary path.


I have not yet looked at the test-path SUM or NULL channels to understand anything about what's in there other than the obvious -- that they're replicas of the singular SUM and NULL channels that are *extremely* useful in commissioning the IFO. So, I anticipate these will be just as useful because we can use these test paths to configure two *other* versions of how the DCPD signals are processed.