Initial state: two BX (old metallic-sheathed cable) arrived at one outlet box on the second floor. They are sourced from a pair of 15 amp breakers tied together as an MWBC at the panel. Apparently, they show up at the panel with just one neutral, even though they arrived at this box with distinct neutrals. There is probably a junction somewhere in between. (Professionals did the panel for me last year.) It was a frighteningly crowded box, with an outlet.
Exiting the boxes was knob and tube up to junction boxes in attic. Out of the junction boxes came a mixture of knob and tube and modern NM, feeding a mixture of plugs and lights. Yea, welcome to 1920, as gradually modified since.
I got to see and change this when we gutted a bathroom on the second floor which exposed the back of this box.
(We got an clean NM circuit for the plugs in the refurbished bathroom. I was not prepared to rewire the whole house or get new home runs in place of the BX. It would take up too much room to explain why it was feasible to fix the outlets.)
I split the box into two boxes. I replaced the runs of K&T up to the attic with NM. Each box ended up with one BX coming in, one NM going out, and an AFCI outlet in the middle. Why two boxes and not one big one? I could avoid disturbing the intact lathe and plaster around the original box and open the new box into the bathroom vanity cabinet.
I neatened up the junction boxes in the attic so that the remaining K&T was handled appropriately. I did all this work while the walls were open in the bathroom. For 'reasons,' it took weeks for all the finish to get done. Nothing went wrong during this time, with power flowing through the AFCIs to the legacy wiring, including to one laser printer and one microwave oven.
Then, we finally were ready to turn on the new bathroom lights, which were wired downstream of one of the AFCIs. (In perfect 20-20 hindsight, I realize that for a little more wire I could have tied them in upstream. Since I knew that there was a mixture of lights and plugs downstream that I could not sort out, I didn't bother.)
We turned on the lights. All was well. Then the laser printer came on. The AFCI tripped. Note that it did not trip when the laser printer ran before we connected and turned on the lights. Remembering a prior saga with a laser printer and a GFCI, I move the printer to another part of the house that has modern wiring, and all was well. The next morning, someone turned on the microwave while the bathroom lights were on, and the AFCI tripped again.
For now I've taken out the AFCI on this branch.
By working down from above, I could still rework this to get the bathroom lights upstream of the AFCI. Still, I would still like to have some clue as to exactly what is going on here. The bathroom light fixtures are LED, they should not draw much current. Why does the combination of them with an inductive load trip the AFCI?
I could also prepare a big surprise for the next owner of this house by putting one or more AFCIs up in the attic so each one handles a smaller and better understood set of wiring. 'surprise' because, if one ever trips, no one is going to guess that there is a box-o-AFCI's in the attic.
Reading some other questions and answers, I suspect that the response here will be to go on a hunt for crossed neutrals downstream from here. I have corrected a fair number of bootlegged grounds, but I guess it's possible that I missed one, as well. Answerers are free to vote to close as a duplication failing any unique ideas about this particular story.
