Before taking the beam jitter measurements last Friday (aLOG post coming soon), I adjusted the input filter offsets for each individual MC2trans and IM4trans DCQPD quadrant such that even with the shutter closed they always read a (small) positive number of counts. This should ensure that the SUM channels never swing through zero. The purpose of this was to avoid the "divide by zero" errors in the pitch and yaw channels, which are normalized by the SUM values.
After adjusting these offsets, the DCQPD SUM dark levels were:
MC2trans DC SUM dark level: ~1.5 counts
IM4trans DC SUM dark level: ~1.3 counts