jeffrey.kissel@LIGO.ORG - posted 17:50, Monday 23 August 2021 - last comment - 15:17, Friday 27 August 2021(59696)
MEDM Screen Updates for SUS in HAM 5, 6, and 7
J. Kissel Following last week's work in implementing the new h1susifoout and h1sussqzout front-end models (see LHO aLOGs 59651 and 59654), today I've done the following to make sure various MEDM screens and infrastructure have followed. - Commits and reverts and updates to clean up the *very* messy svn status in the local copy of the /opt/rtcds/userapps/release/sus/common/medm/ directory including - committed a *ton* of screens in which the only change was to change the start-up location on the screen to be one-monitor friendly. - Reverted the changes to following models because, all though they looked small an inconsequential, the differences didn't make sense to me: SUS_CUST_HSTS_OVERVIEW.mdl SUS_CUST_HSTS_OVERVIEW.mdl SUS_CUST_HSTS_OVERVIEW.mdl SUS_CUST_HXTS_M1_OPTICALIGN.adl pi/SUS_CUST_PI_PLL_OSC.adl pi/SUS_CUST_PI_QPD_CONVERSION.adl - updated SUS_CUST_TMTS_OVERVIEW.mdl to receive the new QUAD tracking as a part of ECR E2000186 that we'll install next Tuesday (which for now will be white and/or have a broken link to the screen, which is fine) - updated /opt/rtcds/userapps/release/cds/h1/medm/SITEMAP.adl to - gathered the new HAM567 suspensions into new drop down menus that match how they're collected in the front-end models. What was "SUS AS" and "SUS SQZ" is now "IFO OUT", "SQZ OUT", and "SQZ IN," since the fact that these are "SUS" menus is implicit in the color of the buttons. - add calls to the future FC1 and FC2 to the bottom of the HXTS dropdown menu - Updated / created new medm macro files for use with current / future MEDM screens: - edited the hard-coded variables in the following macro files in the /opt/rtcds/userapps/release/sus/common/medm/ directory to match new front-end model location, confirming that SUS_CUST_OFIS_OVERVIEW.adl accepts the new stuff, and reads correctly. - susofi_overview_macro.txt - susopo_overview_macro.txt - suszm1_overview_macro.txt - suszm2_overview_macro.txt - created new macro files for FC1, FC2, ZM3, ZM4, ZM5, and ZM6 in the same directory, - susfc1_overview_macro.txt, - susfc2_overview_macro.txt, - suszm4_overview_macro.txt, - suszm5_overview_macro.txt, - suszm6_overview_macro.txt - created new /opt/rtcds/userapps/release/sus/common/medm/hxds directory for all HXDS MEDM screens - filled in hxds directory with screens starting from the hsss directory - fixed a few bugs here and there - Filled out ZM6's infrastructure - filled out the OSEM2EUL and EUL2OSEM matrices with calculations from /ligo/svncommon/SusSVN/sus/trunk/HXDS/Common/MatlabToolsmake_sushxds_projections.m - pushed the matrices with /ligo/svncommon/SusSVN/sus/trunk/Common/MatlabTools/fill_matrix_values.m - Turned OFF all the off-diagonal elements of the new DRIVEALIGN matrix - Turned OFF the LKIN oscillator, and zeroed out the LKIN2OSEM matrix - Changed all the OFFSETS I could find from the RCG default of 1.0 to 0.0 - Set the *representative* BIO request to 2.0 such that the auto-compensation in the COILOUTF banks are set correctly. Not that this doesn't control the HAM-A driver's analog filter -- that's been jumpered to have the low-pass filter engaged always (and the text/coil enable bit also jumpered to "COIL" in, i.e. the DAC input rather than the front-panel "TEST" input). - Copied over the OSEMINF compensation and calibration filters from an HTTS - Copied over the COILOUTF compensation filters from an HTTS - Copied over the still-very-crude damping filters from an HTTS in the three DAMP banks - Copied over the WD_OSEMAC_BANDLIM filters from an HTTS, and the usual meaningless 25000 ct threshold. - Installed polarity flipping signs on LL and UR in the COILOUTF bank - Committed the filter file and the safe.snap, which have already been softlinked to the appropriate location in the userapps svn. - Tested that a ~10k offset each DOF of the COILOUTF bank creates a drive signal all the way out to the IOP model. No action from the coil's VOLTMONs though -- need to debug that later. More work will come in tomorrow as I draw up screens for the more sophisticated HXDS -- i.e. the HPDS, HTDS, and HDDS.
Images attached to this report
Comments related to this report
H1SUSOPO MEDM Screen / Foton / WD infrastructure Filled In J. Kissel Continuing the work of getting up-cycled and new SUS MEDM / FOTON infrastructure up and running (see start with ZM6 in LHO:59696), yesterday and today I built up the OPO infrastructure. This involved: - I actually *created* the MEDM macro /opt/rtcds/userapps/release/sus/common/medm/susopo_overview_macro.txt rather than what's misstated above that I've modified it. The macro file that supports the OPO when it lived in HAM6 is actually the (errantly named) /opt/rtcds/userapps/release/sus/common/medm/susvopo_overview_macro.txt. - After filling this macro out, I quickly found that the MEDM overview screen, /opt/rtcds/userapps/release/sus/common/medm/opos/SUS_CUST_OPOS_OVERVIEW.adl used a macro variable that resulted in pointing to the wrong user model GDS_TP screen. So, I updated the overview to use the (new for this model) macro variable "MASTERMODEL" and "mastermodel" set to SQZIN and sqzin, respectively, such that the right screen opens up. Also in the update to this screen, I removed all reference to the now-non-existent user DACKILL, since that's been removed in the transition of these suspensions from susopo.mdl to sussqnin.mdl front-end model. - I used /ligo/svncommon/SusSVN/sus/trunk/OPOS/Common/MatlabToolsmake_susopos_projections.m and /ligo/svncommon/SusSVN/sus/trunk/Common/MatlabTools/fill_matrix_values.m to populate the OSEM2EUL and EUL2OSEM matrices automatically. Since make_susopos_projections.m doesn't produce a SENSALIGN matrix, I just ran fill_matrix_values using the identical, 6x6 identity matrix from /opt/rtcds/userapps/release/sus/common/medm/make_sushsts_projections.m - In the DRIVEALIGN matrix of filterbanks, I did set these to be an identity mattrix by hand by turning off the inputs, outputs, and setting the gain to 0.0 on the off-diagonal elements. - Using the GUI copy-and-paste feature of foton, I copied over the OSEMINF, DAMP, COILOUTF, WD_OSEMAC_BANDLIM, and WD_OSEMAC_RMSLP "by hand" from the former /opt/rtcds/lho/h1/chans/H1SUSOPO.txt in to the new OPO banks space in /opt/rtcds/lho/h1/chans/H1SUSSQZ.txt and turned the appropriate FM modules on as I went. - Since "it all happened to fast," it took me a minute to find the damping loop settings, but after some digging and trending, I found that every DOF *only* used the "0:2:20" filters in FM7 (which is a zero at 0 Hz, and two poles at 2 and 20 Hz), with gains of [L,T,V,R,P,Y] = -1* [100, 100, 50, 70, 80, 20]. A crude filter, but good enough. L1 OPO is using this same set of damping filters and gains, but their Y gain is -30. - One more change was necessary to the MEDM screen collections: even though the HAM-A driver for the OPO continues to be "hard-coded" jumpered to the "low-pass ON" and "coil enabled" in the analog electornics, there remained misleading BIO bit state requests and readbacks in /opt/rtcds/userapps/release/sus/common/medm/opos/SUS_CUST_OPOS_BIO.adl As such, I've ripped all that out and repleced it with a message to the reader. Note: I did "set the M1_BIO_STATE_REQUEST" by-hand to 2.0 (using command line caput to H1:SUS-OPO_BIO_M1_STATEREQ), such that the COILOUTF compensation filters get accurately set to compensate the low-pass ON state. - Most excellently, the OPO's channel names have not changed at all. As such, the guardian infrastructure still works perfectly, and I was able to excercise it (though, nothing's plugged in yet, so I was merely confirming that it just turns on and off all the loops correctly). Attached are some screenshots of the important / changed screens mentioned in this aLOG above.
Images attached to this comment
H1SUSFC1 MEDM / Foton Infrastructure Filled In
J. Kissel
Here're the notes from me building up the FC1 control system system.
- updated medm macro file to match all the new underlying CDS GDS TP information:
/opt/rtcds/userapps/release/sus/common/medm/susfc1_overview_macro.txt
- Copied over all OSEMINF, DAMP, COILOUTF, and WD_OSEMAC_BANDLIM_T1 filters for an HSTS from SRM's foton file,
/opt/rtcds/userapps/release/sus/h1/filterfiles/H1SUSFC1.txt
- Used
/ligo/svncommon/SusSVN/sus/trunk/Common/MatlabTools/fill_matrix_values.m
to fill in the values for the OSEM2EUL, SENSALIGN, and EUL2OSEM for each stage that come from
/ligo/svncommon/SusSVN/sus/trunk/HSTS/Common/MatlabTools/make_sushsts_projections.m
However, two things interesting on the M3 stage:
- the M3 SENSALIGN MEDM screen has some white columns for a "B" DOF. Never heard of this, and it's not filled out in an HXTS in L1 or H1. I wonder what this "B" row / column is? My guess is that we can delete it. I'll check in with likely suspects.
- the M3 EUL2OSEM matrix is a ramping matrix, so fill_matrix_values.m doesn't work. You've gotta install that by hand.
- Standard HSTS calibration gains [P,Y] = [1.875,2.681] for the M1_OPTICALIGN alignment sliders and in the *misalignment* gains, aka the M1_TEST P and Y banks.
- Zeroed out all Pitch and Yaw OSEM lockin oscillators, and corresponding LKIN2OSEM matrices on each stage.
- Discovered a bug in the title of the M2 and M3 LKIN2OSEM matrix screens, where it had titled the screens "L1," i.e. a leftover from the copy-and-paste of these screens from the QUAD UIM stage. So, that's an update to
/opt/rtcds/userapps/release/sus/common/medm/hxts
SUS_CUST_HXTS_M2_LKIN2OSEM.adl
SUS_CUST_HXTS_M3_LKIN2OSEM.adl
- Set all the BIO to COIL, and all the filter responses to State 1 for now, toggling them through each state to make sure it appears functional. But, along the way, discovered the first 4 bits on the M2 BIO are stuck. This is on Fil's long list to debug next week.
Screenshots of my efforts are attached, with the overview screen first, and then all the relevant subscreens, in order of left to right across the control screens. Where there is three stages worth of information, I've grouped the screenshots together by function.
Images attached to this comment
