We spend a lot of time talking about "spikes" and "transients" but seldom to we actually see or measure what we are discussing.
Had my oscilloscope in the train room for something else, so thought I would get some images of spikes. (For you electronic heavy hitters, these are just jpg screen shots of my scope, I can do better, but I'd need to haul a computer into the train room too!)
Anyway, here are some pictures of the voltage at the motor terminals of a 364 lumber loader with brand new brushes, springs and a clean commutator. This 364 has the new style motor with the slotted brushes. Transients clearly in evidence. Another photo shows the same setup with a #53 bulb also connected across the motor terminals; transients are greatly diminished but still in evidence. Transformer was not train related, just a large 12 volt transformer I had handy.
Did a little poking around, obviously the sparking connection at the brushes is the root cause, but I noticed that increasing teh brush pressure slightly made the motor speed up and reduced the transients quite a bit. I will run this motor in for a few hours to seat the brushes to the commutator and see what it looks like then.