Jonathan Lewis suggested that we cut down on the BMU readout time by cutting back on the occupancy by using a shorter time window: 1000 nsec. Bob looked at the time distribution for Z events and everything fits nicely in 1000 nsec.
The raw time distribution (for a run with a couple of oscillating
cells) is seen below. It looks like we can safely reduce the occupancy
by about a half.
Modified 31-October-2006 at 13:30
http://hep.physics.wisc.edu/~jnb/imu/31Oct2006