From the troubles you're having with these things I'm glad they haven't appeared over here. It's probably only a metter of time though.
When the manufacturers got their "current signatures" of typical fault arcing, I wonder if they bothered to compare these with those obtained from a motor which is sparking slightly at the brushes?
it might be interesting to compare the signatures and see how close they are. If sparking commutators is going to cause problems with AFCIs, I see you having a lot of calls if/when they start to be used on circuits feeding power tools etc.
