J. Kissel, S. Dwyer
WP: 8239

As Jenne "concisely" explained in pgs 8-21 of G1900965, we've been exploring the implications of large IFO beam spot positions on the DARM and ASC control system. As you know, H1's currently operating with spot positions on test masses rather far away from geometric center of rotation (in the few 10s of mm range). With these spot positions, the problems of length-and-angle coupling comes to a head at the L2 (PUM) stage of all the QUADs, especially that which are used for DARM control (i.e. for us, in O3, ETMX)  One of the ideas to combat this multi-layer cross-coupling is to 
    (a) have the option of feeding the output of the L2A filters feed back in to the A2L filters, and
    (b) split the A2L filters into two parallel paths -- one for scalar spot position control, and two for mechanical, frequency-dependent A2L decoupling.

I've implemented this infrastructure change today to help facilitate the prototyping of such schemes. 
No new scheme has been *implemented* yet, the new infrastructure was built to be backward compatible, and is now running in a functionally identical way as before.

Attached are screenshots of the simulink model changes (1st is before, 2nd is after), and of the MEDM screen changes (3rd is the overview change, and 4th is the new screen).

The change only required modification of 
    /opt/rtcds/userapps/release/sus/common/models/
        FOUROSEM_DAMPED_STAGE_MASTER_WITH_DAMP_MODE.mdl
-- the obnoxiously named library part for all QUAD's L2 stages (used by both the QUAD_MASTER.mdl and QUAD_ITM_MASTER.mdl larger library parts for the two types of QUADs [frustratingly only differing by their ESD driver control]).

However, the impact on screens was a little bit larger: I created a few new screens,
    /opt/rtcds/userapps/release/sus/common/medm/quad/
        SUS_CUST_QUAD_L2_DRIVEALIGN_L2A2L.adl
        SUS_CUST_QUAD_L2_DRIVEALIGN_L2A2L_MTRX.adl

and modified the overview screens,
    /opt/rtcds/userapps/release/sus/common/medm/quad/
        SUS_CUST_QUAD_OVERVIEW.adl
        SUS_CUST_QUAD_ITM_OVERVIEW.adl

If LLO wants these changes to begin prototyping this sort of decoupling as well, I can draft the ECR. 
Install instructions:
(1) save your spot position A2L gains ahead of time, and any filters you've had in the L2_DRIVALIGN_${A}2L banks, since these are now split in to the two new L2_DRIVALIGN_${A}2L_MECH and L2_DRIVALIGN_${A}2L_SPOT banks.
(2) svn up the above folders, 
(3) recompile, install, and restart all QUAD models. 
(4) turn OFF the inputs to the new $(A)2L_SPOT and $(A)2L_MECH filter banks
(5) load a 1.0 in to the new $(A)2L MTRX _2_1 elements
(6) restore the spot position gains in the ${A}2L_SPOT banks, and if you did have any frequency dependent A2L filters, load them in to the ${A}2L_MECH bank. 
(7) turn ON the inputs to the $(A)2L_SPOT banks.

The 4th attached MEDM screenshot shows the configuration in which LHO is currently running -- i.e. with NO L2$(A) gains or filters, so there was no need to install any non-zero elements in the new $(A)2L matrices.