Figure 1. When looking at a lock in dark fringe with the B2 QD2 galvo loop open there was to sharp steps in the beam position (mostly visible in B2 QD2 H norm at 58m01 and 58m15),

Figure 2, 3 and 4 show the B2 camera image before the first step, in between the two step and after the second step. The beam is not actually moving, but changing significantly in shape. In between the two steps it looks like an order 1 mode.

The 12MHz QD2 signal that is available is much less affected by this beam shape change. So in principle it could be a better centering error signal.

Figure 5 compares B2 quadrant centering signal with 300mW in transmission of the arms. Purple is centering with the DC signal and blue with the 12MHz signal. The 12MHz has clearly higher sensing noise (this is to be expected). But also show a much prominent motion peak at 7.5Hz /8.5Hz in H and V,  and 1.5Hz in H. I don't know what these are and if it is a bad or good thing that the 12MHz sees that motion.