SInce the ETMs were available, I carefully tuned the PUM to TOP mas offloading gain for PIT and YAW. To do this, I drove the PUM at 0.1Hz with 20000cts on the L2_LOCK_[P/Y]_EXC, and set the gain on the TOP mass to exactly cancel the signal in the optical lever.
For YAW this can easily be done with sub-percent precision. The PIT TOP/PUM ratio features a lag filter with a complex pole zero douple-pair at 0.52Hz and 0.56Hz, which can be seen in the QUAD model. I designed a lead filter with those frequencies, but Qs low enough to match the phase at 0.1Hz exactly. The corresponding lead filter is in the M0_LOCK_P FM7 module:
zpk([0.325+i*0.405925;0.325-i*0.405925],[0.349999+i*0.437149;0.349999-i*0.437149],0.999999,"n")
The regquired gains in the M0_LOCK filter modules were:
ETMX PIT 898 *1e-6
ETMX YAW 1428.7 *1e-6
ETMY PIT 1294 *1e-6
ETMY YAW 1650 *1e-6
I put gain of 898, 1428.7, 1294 and 1650 into the M0_LOCK FM6 modules. By selecting FM4, FM5, FM6 (and FM7 for PIT) this implements a xross-over at exactly 0.1Hz. (pictues).
The same has to be done for the ITMs, and the guardian has to be updated to use those filters.
Also set the PUM to TOP cross-over to 0.1Hz for the ITMs, following the same procedure as for the ETMs. The gains were tweaked in the OFFLOADED configuration. The required gains were:
ITMX PIT 1147 *1e-6
ITMX YAW 1547 *1e-6
ITMY PIT 1155 *1e-6
ITMY YAW 1573 *1e-6
I also moved the gain filters to FM3 (removing the old +10dB filter), so it can be switched before the integrator. This was done for all 4 test masses, pit and yaw. With these filters, the settings shown in the attached screen shot result in a 0.1Hz x-over.
The guardian was updated to set this x-over just before engaging the soft loops. I havent switched the guardian to use the offloaded design for the ITMs though.