Reports until 17:11, Wednesday 19 January 2022
H1 SUS
jeffrey.kissel@LIGO.ORG - posted 17:11, Wednesday 19 January 2022 (61353)
Updated OFI OSEM2EUL and EUL2OSEM Matrices for Bug Fix in Yaw Lever Arm
J. Kissel

While perusing the Controls Design Description Summary table (G1100968), and trying to resolve issues we've been having with understanding HXDS actuation chain, I identified that the Output Faraday Isolator Suspension (OFIS) also had its lever arm listed wrong at 55.5 mm. 

To verify this, I went to the OFIS Assembly Drawing (D0900623), and used the e-Drawing to measure the distance between the LF and RT OSEM magnets (a screen of which now attached to D0900628 in the "other files," see D0900623-v16_aPLUS_OFI_LFtoRT_AOSEMFlagDistance.png). 

One can see from that measurement, that the distance between LF and RT is 222.25 [mm]. 

That means the yaw lever arm is half this distance, 222.25 / 2 = 111.125 [mm].

Because it was wrong in the summary table, I was worried it was wrong elsewhere, so I checked the script that uses, among other dimensions, this distance to compute the OSEM2EUL and EUL2OSEM basis projection matrices,
    ^/trunk/OFIS/Common/MatlabTools/make_susofis_projections.m


Indeed, the variable representing this distance in the code, M1.dLFtoRT, had been set as "111e-3," i.e. 111 [mm], a rounded version of 222.25 / 2. That means the OSEM2EUL projection matrix had been over representing how much LF and RT contributed to Yaw until now, and EUL2OSEM matrix had been over-driving LF and RT to create Yaw until now.

So, I've 
    - fixed the code, 
    - added more comments and references, 
    - committed to the repo, 
    - run the code on a workstation, 
    - installed the updated values into the OFIS EPICs records, and
    - accepted the new values into the h1susifoout front-end model's SDF system.

#FutureWork
We should re-run and compare transfer functions, and adjust damping loop gains as necessary.
The alignment offset slider calibration for Yaw has not been populated since the OFI made the front-end model transition from the (now defunct) h1susomc model to the new h1susifoout model. 
Nor has there been any requested alignment offset either.
I'm not even sure the OFI ever did have any alignment offsets applied. 
So, yeah -- the OFI control system needs some attention.
Images attached to this report