I've had this debate before
The AFCIs have to detect the difference between "good" arcs and "bad" arcs. I doubt they can do that with 100% certainty.
The "test" button only gooses a µP input; not much of a test.
You have some chance of testing GFIs for valid/invalid trips with commonly available electrical stuff; this gadget cannot be checked without a lab full of equipment.
BTW, motors draw big arcs when the switch controlling them opens, so these are probably candidates for the cause of AFCIs tripping. Usually the motor's internal switch or relay contacts have arc suppression networks (MOVs, capacitors, cap/resistor networks) across them. Maybe this network wears out with time.
You could also try
http://www.google.com/search?hl=en&q="square+d"+afci+patent&aq=f&oq=
These new devices might solve some of the problems that the current crop of AFCIs "can't possibly have."
Nobody is going to sue over a nuisance tripping AFCI and nobody will be killed by one either, so without the threat of a wrongful death lawsuit, all the manufacturers of all these safety devices, appliances, connecting cable, "poor installation", whatever, can play hot potato indefinitely.
It must be, therefore, the fault of the buyer!
BTW, if a manufacturer tells you something, and you want to be hung up on, ask for what they just said, in writing, over the signature of a company principal.
One manu. guy could hear that I was typing what he was saying, while he was saying it, and he actually told me to stop typing. Note that he wasn't asking, he was telling!
This guy worked for one of the companies that is now asking for bailout money. No, I won't say which one.
Is this a great country, or what???