As an exercise to test out the OMC --> LSC signal path, I locked the simple Michelson using the OMC-DCPD_SUM signal.  With MICH locked on a dark fringe with a 35 count offset, I measured the DCPD-to-MICH transfer function.  The DCPD sum was a factor of ~4 smaller and off by 180deg at low frequency, so I loaded "-4" into the LSC input matrix (OMC DC --> MICH) and zeroed the ASAIR_RF45 element.  The lock was very smooth, with about 2x more gain than the vanilla MICH loop.  (I found the guardian-set gain the  MICH_DARK_LOCKED state was about 3x smaller than maybe it should have been, so I increased it from -500 to -1400 before the handoff.  After the handoff I reduced the gain to -700. The UGF with these settings was ~7.5Hz, roughly aligned with the peak of the phase bubble.)

Comparisons of the OLTF in both states (with the gain settings described above) are attached, so are noise spectra for the error and control signals.  In both plots the references are the RF lock, current signals are the DC lock.