[nathan.chupp]

The brush holders are supported by a phenolic disc that is screwed to the aluminum end cover. The center of the disc has a hole that overlays the bearing boss. Clearance between the boss and the hole is tight, maybe one millimeter. Carbon dust (from the brush) is deposited liberally on the disc and the boss, especially on the insider diameter of the disc. When the brush moves outward, as it will when the commutator is inserted between the brushes, it tilts slightly toward the phenolic disc and contacts the carbon dust. Maybe this is the cause of the short to case -- not a very low impedance connection, but enough to (1) cause electrical noise on your robot's control system, and (2) get worse at higher electrical frequency.