AFCI breakers tripping every 7 days
We have a new addition to our house with its own subpanel. Since it is a bedroom the outlets and lights are on their own two 20Amp AFCI breakers. We now narrowed it down to these two AFCI breakers tripping every 7 days at roughly the same time (from the time they were last reset). In the beginning we did not pay close attention but it always seemed roughly a week. Now it tripped again this evening, 7 days (within the hour) since we last had to reset it. It has been doing this even when the new room was not occupied and there was practically no load. Yesterday I disconnected the Smoke Detector because I thought it might do a self test every 7 days. But it still went out today. We have three dimmers on there and they control recessed lighting as well as one ceiling lamp with 3 or 4 light bulbs. Initially while no one was living there, there was no other load (yet a week later they had tripped). Now there is a small refrigerator in the room, plus a TV, a cable box, a micro wave, toaster and a lamp (there is also a Gas Fire Place with low voltage ignition). One would think that it has to be a device that keeps track of time and does a certain function every 7 days that will trip it. Have you ever heard of such a thing? Are recessed lights that touch the insulation known to cause any trouble? I just don't know of any device in that room that has a circuit that does things on a weekly basis (on the 7th day) after it was turned on. And we replaced both breakers once.
Ask and ye shall receive... something
I asked Square D customer service if a fault on the line side could trip an AFCI breaker. They said no. I didn't like that answer, so I did a little more research, and found the following block diagram of an AFCI (from http://www.iaei.org/subscriber/magaz...mblin_fig5.gif), which seems like it will trip the CB, regardless of which side the fault occurs on:
I then sent the diagram and the same question to the IEEE, which has a special arc-fault advice line set up. They said yes, based on that diagram. There is, of course, no guarantee that the diagram accurately describes Square D's breaker -- they could have incorporated a filter to prevent detecting fault signatures on the line side, for example. Depending on the sensitivity of the sensor, you may also need some load on the load side to complete the circuit through the sensor.
Shall we try for 2 out of 3? Anybody want to do the obvious experiment?