Installed 2 years ago, 12/2 NM on a 9 pack bank of high bay t5's at 216 watts each. Started tripping breaker (after it melted) and found wire melted in attic under 2 ft of insulation. (This was new construction) Between rafters, 2 ft from any other wire etc, it was three way switch and this wire was the switch leg. After looking around also found that traveler was burned but not shorted yet. Replaced all of this and turned on, my meter reads 18 amps which is a little higher then it should be but well in the realm of safety factor. This is the only thing on this circuit, 20 amp breaker. Can't believe it happened and also can't believe it didn't burn the place down. Any theories? Thanks.

  • 1
    Uh oh. Was this a continuous stretch of wire that had this meltdown? Can you provide photos of the damaged cable? Also, can you get a thermocouple or three on the cable while pulling full load through it, or an IR shot of the ceiling from below even? (Trying to get an idea of the temps the cable's hitting here.) Aug 21, 2018 at 2:50
  • 1
    Was the cable buried in insulation? Aug 21, 2018 at 3:02
  • If I were you I'd buy an AFCI breaker for this circuit pronto. If it happens again, the AFCI can hopefully prevent a fire.
    – Machavity
    Aug 21, 2018 at 13:08
  • cable was buried in 2 ft of blown fiberglass insulation, gfi is a great idea, not sure how I'd do thermocouples or ir shot. Don't have that kind of equipment
    – dean
    Aug 21, 2018 at 13:53
  • 1
    He didn't say gfi. He said AFCI. Totally different thing. Aug 21, 2018 at 21:16

2 Answers 2


They sell "850 watt" power supplies for gaming PCs. When you look at the power supply sticker (we're talking the responsibly built, UL listed ones), they indicate 10A / 1200VA. How much power do you need to provision to such a machine? 1200VA, not 850VA. The difference is conversion losses inside the power supply (which is why power supplies have fans), including power factor.

So I'm a little skeptical of where this "216 watt" figure came from. It is precisely a multiple of both 6 and 4, and "36 watt" and "54 watt" are standard ratings for fluorescent tubes. However, extrapolating from nominal tube spec totally disregards:

  • over-unity ballast factor (meaning the ballast drives the lamp harder than its nominal spec)
  • internal ballast inefficiencies and conversion losses
  • power factor of the ballast (VA vs watts; you must provision VA).

I also don't put a lot of weight on your 18A (2A per fixture? Exactly?) number. I would be happier if you obtained a Kill-a-Watt meter and plugged one of the fixtures into it, and see how it actually performs at operating temperature. Look at amps, and VA. (Watts is useless).

I think you will find you have significantly oversubscribed this circuit.

The cable being under a carpet of insulation isn't helping either, but I'm not an expert on Romex installation rules, so I leave that to others.

Continuous loads must be derated to 125%

Lighting is a continuous load. So it must be derated: you must provision 125% of the VA that the lighting actually requires: NEC 215.2a1.

Another way of looking at this is, if all the loads on a circuit are continuous loads, a circuit can only support 80% of breaker rating. So a 20A circuit is good for 16A only. (16A * 125% = 20A).

Even with your perfect 216W number, your 9 lights would draw 16.2 amps (they don't) and that's a tick over limits. In actuality I think you will find your lights are even more than 2.00 amps each, so over 18A requiring you provision over 22.5 amps of power to the circuit. Upsizing to 10AWG wire would reduce your wire overheat, but you would not be allowed to upsize the breaker to 30A, because you are not allowed to put NEMA 5-15 or 5-20 receptacles on a 30A circuit (table 210.21B3).

Break it into 2 circuits

The most straightforward solution here is peel off 3-4 of these lights onto a separate circuit. And, you know, if you want to be able to switch some of the lights separately from the others, this is a great time to do that.

Or... switch to 240V

Pull down a light and read the labeling carefully. Given these T5 lights are super popular in Europe, and given trends in the lighting industry to make multi-voltage ballasts that work on 100, 120, 208, 240 and 277 volts, chances are these luminaires are already capable of running at 240V.

So convert the circuit to 240V. The switches won't care. The lamps won't care.

  • The plugs and sockets must be changed from NEMA 5-15 to NEMA 6-15 or 6-20. (they look practically the same and use the same cover plates).
  • The lighting circuit must be punched down onto a 240V breaker
  • White wires in conduit must be replaced with colored wires.
  • White wires in cables must be taped black or any color at both ends.

What will this do to wire heating? It will cut current in half, which will cut voltage drop in half, which will reduce wire heating by 3/4. it's a very good deal that should solve your overheat problem. As long as you don't add even more lighting or other loads.

  • While the suggestion to move the circuit to 240V is technically sound provided the OP's ballasts are multi-voltage -- how would you address the AHJ invoking 210.6(A) to put the kibosh on it? (Not that the rule makes a lick of sense for fluorescent luminaires anyway -- the actual luminaire terminals are buried in the wiring housing where you have to take the fixture apart to get to them, and the quasi-pokable bits on the tombstones are going to be at well over 120VAC RMS between them no matter what voltage is going into the luminaire, no?) Aug 22, 2018 at 4:33
  • @ThreePhaseEel My argument is OP is not in a dwelling unit or hotel room. 210.6(C)(1) applies (fluorescent=discharge). I do not see any conceivable use for 194,400 lumens in a dwelling unit, except for cases where 210.6A would not be high on your list of problems if the gub'mint came round :) Aug 22, 2018 at 4:48

This sounds like a very close call!

Did you inspect and test the insulation of the other wiring in the attic? The wires up to the switch loop were under the same load in the same attic... I'd probably be replacing it all.

Converting to LEDs would reduce the load on the circuit, reducing the heating of the wire, and also may reduce the heat from below raising the ambient temperature. This would get you down to a 15A breaker.

You probably have to derate very carefully here. Let's say the ambient temperature around the cable gets as high as 80*C / 176*F. The derating multiplier is .41 for 90*C wire. You'd actually need 10AWG NM for a 15A breaker. But after this incident I'd be wary and might run MC just in case.

(To really know how you have to derate, you'd have to meausre the ambient temperature carefully on a worst case hot day, but realistically it's probably better to just aim high to be safe.)

However if the temperatures are not that high, maybe there was some other cause for the problem.

Did you know there are heat detectors you can buy for attics? They go off at 200*F or something like that.

A thermostatically controlled attic fan might be a good idea if the temperatures up there are that high, not just for the wiring. Run oversized MC for it :)

(edited as I reconsidered a few points...)

  • Changing to LED won't be that helpful. Those T5s are highly efficient already. Aug 21, 2018 at 21:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.