One of the concerns that I have with AFCIs and well as GFCIs is that an AFCI contains equipment ground fault protection. The amount of static electricity that the human body can carry is capable of nuisance tripping a GFCI - I am not sure if the 30 mA or 50 mA trip level in an AFCI would also be susceptible. On very cold and dry days during the winter, if I do not first ground my body and then the current carrying prongs of my electric shaver plug, the electric shaver will nuisance trip the GFCI in the bathroom. The mechanism is electrostatic induction. The static charge on my body induces and equal an opposite charge in the motor windings of the shaver. When I plug in the shaver ( without first discharging the static electricity ) the GFCI reads the static as a ground fault and trips. There is just not a way to make a GFCI that is smart enough to tell the difference between a real ground fault or a phantom ground fault from static electricity.

A similar thing can happen when a motor turns off or when the starting winding shuts off during the starting sequence. What happens is that the voltage spike from motor inductance forces a phantom ground fault pulse to flow through the winding to frame capacitance. Essentially, a small and transient version of how if you do not ground the frame of the hoist motor of a 480 volt 15 ton bridge crane the hoist ropes will have a nasty tingle voltage when the hoist is energized no matter hlw good the motor insulation is. Since 120 volt motors are built cheaply you can expect them to have more than their fair share of insulation capacitance and resistance problems.

What gets me is that I install a har-wired uninterruptible power supply in the smoke alarm circuit and that would completely defeat the AFCI.

I also consider 120 volt smoke alarms to be a bogus product. The 9 volt backup battery can be very difficult to replace for the mechanically inclined not to mention that there are a lot of people out there who are NOT mechanically inclined or who even own a ladder. The 9 volt battery could just as easily be located on the wall along with the test button. I have seen some smokes at trade shows that have an allegedly permanent battery that allegedly lives that same 10 years as the bit of radioactive material. There is also an alarm that speaks, "low battery," when the battery runs down.

In the case of one commercial customer in Cleveland, Ohio the owner, employees, and customers ignored the low battery chirp for like 3 years. One problem was that the ceiling was high enough ( 11 feet ) that the only ladder for changing the batteries required 2 people to handle it.

Sure, 120 volt smoke alarms are better than nothing, but my feeling is that a dwelling is better off to have a REAL fire alarm system. The backup battery is rechargeable and relatively easy to reach. A system that is moderately smart can periodically test the battery and if the battery is bad it can speak, "backup battery needs service or replacement."

My other feeling is that like other people have written is that an AFCI will not necessarily stop a fire from starting. In fact, I tell people that a GFCI will not stop you getting shocked, it will only stop the shock from killing you.

A further point is that General Electric and Siemens are the only 2 companies that have made a 2-pole AFCI for use with 3-wire circuits. These are common for higher load densities such as kitchens. There were some places such a Trumbull County, Ohio that required both 120 volt and 240 volt 15 ampere or 20 ampere receptacles in kitchens. Threfore, the kitchen circuits were essentially required to be 3-wire plus ground.

I have also experienced that even having 4 AFCIs in a SquareD 40 slot panelboard creates a lot of congestion because they stick so far into the gutter. SquareD would be better off to make a version that takes up an extra plle space ( like neutral switching breakers do ) in order to free up gutter space. Expect to need to use a lot of 16-slot subpanels when using SquareD.