A scary discovery, what's going on?

[Rollie Meyers]

You have to account for the ballast loss in addition to the lamps.

[roger wiegand]

OK, yes, I forgot about that. Unfortunately I don't have any way to measure the actual amperage.

[Michael Burnside]

There is no way that timer switch is rated for 15A continuous current or even 80%. I don't know if the specs are written on it or not, but even if it were 3-7 stranded cores you'd be looking at TOPS 10A and most likely 7 or 8 looking at the pictures and assuming > 10 cores. Maybe it would spec 10-12A, but given the lights you have and assuming +/- variability, current the timer draws itself, temperature, etc., no way it can handle the load safely.

[Dan Friedrichs]

There is no way that timer switch is rated for 15A continuous current or even 80%. I don't know if the specs are written on it or not, but even if it were 3-7 stranded cores you'd be looking at TOPS 10A and most likely 7 or 8 looking at the pictures and assuming > 10 cores. Maybe it would spec 10-12A, but given the lights you have and assuming +/- variability, current the timer draws itself, temperature, etc., no way it can handle the load safely.

Looks like it's Honeywell rpls740b, and the manual for it does say 15A @ 120V...

[Dan Friedrichs]

1) The stranded wire is melted along its length

It's hard to tell from the pictures, but it sure looks like the stranded wire insulation is melted all along the length (from the nut to the switch), but that is the "line" wire, and the output ("load") wire from the timer looks to be in fine condition. If it were an overcurrent condition, you'd expect both of those to heat equally. So it seems safe to assume it's not related to excessive load (or the load wire would be melted, too).

The wire nut may be suspect, but it sure looks like it's well-twisted. The risk, here, is that poor contact between the wires causes high resistance and heating. This would affect both the solid (input) wire and stranded (output) wire - are both exhibiting melting? (Understandably, the solid wire may be less heated due to larger gauge or higher thermal mass). But I'm suspicious of this, too, because SO MUCH of the stranded wire insulation is melted - it seems to have melted all the way along the length, including a long distance from the wire nut. The "wire nut failure mode" is that heat is generated at the connection point within the nut - the stranded wire being melted so far back seems to not support this theory.

Is it possible there was a short between the stranded wire and another wire in the box? If the stranded wire had a nick in the insulation and was touching one of those bare ground wires, that could cause high current just in that segment of stranded wire and explain all of this.

You mention loss of a strand or two being a concern, but I don't think that's the case. Loss of a few strands would cause a local area of higher-resistance, but once those strands join the larger bundle, current will flow out into all strands.

From these pics, I really think there was some alternate path for current to flow through that stranded wire...

[roger wiegand]

The spec is 15A; 1800 watts or a 3/4 hp motor according to the manufacturer. It's a UL listed device, and I hope a company like Honeywell doesn't fudge the numbers. I talked with a pretty knowledgeable tech support person at Honeywell this afternoon. They couldn't pinpoint a problem, suggesting several of the same options as came up here and also suggested the possibility of a voltage surge from lightning. They recommended confirming the load and either splitting it or upgrading the wiring and switch to something heavier duty if it's close. Rewiring with 12 ga would be a lot harder than making it work on two independent circuits. They confirmed that it should be fine if my initial wattage calculation were correct.

I've just ordered myself an ammeter after needing one multiple times over the years and can actually see what the current load is. I'm now guessing it's going to be high. I'm thinking I'll probably have to split the lights onto two different circuits with two switches. LED conversion is another alternative, but the replacement bulbs I've found so far are only 60% as bright and still use half the wattage so I might end up having to buy and install more fixtures, something I'm loathe to do.

There is no way that timer switch is rated for 15A continuous current or even 80%. I don't know if the specs are written on it or not, but even if it were 3-7 stranded cores you'd be looking at TOPS 10A and most likely 7 or 8 looking at the pictures and assuming > 10 cores. Maybe it would spec 10-12A, but given the lights you have and assuming +/- variability, current the timer draws itself, temperature, etc., no way it can handle the load safely.

[roger wiegand]

Interesting thought. I too was surprised at the difference between the line and load wires from the switch. The load wire (the blue one) was perfect, no sign of overheating at all. I would have thought they should carry the same current. I saw nothing in the box that suggested a short opportunity, but the black wire was pretty well destroyed. The box was not crowded, because the switch is physically large I used a double box for the single switch. The ground wires were folded all the way to the back of the box, not close to the line wire that melted.

It's hard to tell from the pictures, but it sure looks like the stranded wire insulation is melted all along the length (from the nut to the switch), but that is the "line" wire, and the output ("load") wire from the timer looks to be in fine condition. If it were an overcurrent condition, you'd expect both of those to heat equally. So it seems safe to assume it's not related to excessive load (or the load wire would be melted, too).

The wire nut may be suspect, but it sure looks like it's well-twisted. The risk, here, is that poor contact between the wires causes high resistance and heating. This would affect both the solid (input) wire and stranded (output) wire - are both exhibiting melting? (Understandably, the solid wire may be less heated due to larger gauge or higher thermal mass). But I'm suspicious of this, too, because SO MUCH of the stranded wire insulation is melted - it seems to have melted all the way along the length, including a long distance from the wire nut. The "wire nut failure mode" is that heat is generated at the connection point within the nut - the stranded wire being melted so far back seems to not support this theory.

Is it possible there was a short between the stranded wire and another wire in the box? If the stranded wire had a nick in the insulation and was touching one of those bare ground wires, that could cause high current just in that segment of stranded wire and explain all of this.

You mention loss of a strand or two being a concern, but I don't think that's the case. Loss of a few strands would cause a local area of higher-resistance, but once those strands join the larger bundle, current will flow out into all strands.

From these pics, I really think there was some alternate path for current to flow through that stranded wire...

[Dan Friedrichs]

That switch, itself, just contains a relay that opens and closes on command. Ideally, it generates no heat, itself, so 15A or 20A or even 25A should have a trivial impact on it. It's not like an old rheostat or something that is going to get hot. So: I really doubt this has anything at all to do with overloading the circuit (presuming the circuit is even overloaded, which it may not be).

Bizarre, scary situation, for sure.

[Mike Soaper]

I'm wondering about the corrosion , ie. was there an earlier event that cooked the insulation and exposed the wire so it corroded , or was the corrosion on the wire inside of the insulation. Either way maybe the corrosion caused enough of copper loss, or broken conductors reduce the wire's current capacity.

Maybe strip a section of the insulated part of the corroded wire near the device to see if it's corroded inside of the insulation

[Karl Loeblein]

I replace my fluorescent lights with lower wattage LEDs. Much brighter and no delays starting up in the cold winter. My only regrets is not doing it sooner.

[roger wiegand]

I replace my fluorescent lights with lower wattage LEDs. Much brighter and no delays starting up in the cold winter. My only regrets is not doing it sooner.

I'd like to find such a thing, The T5HO bulbs are 5000 lumens each, the brightest LED replacement I've found for them is 3500. Lumens/watt is quite close between the two-- the t5HO bulbs are quite efficient. The available replacement bulbs I've found also have an even worse CRI, not great for finishing.

[Karl Loeblein]

Roger, I found it cheaper to replace the entire fluorescent fixture and it looks better too. You also get a lot more lumen and CRI options with replacement fixtures.

This is much brighter in person, but my iPad auto darkened the pic because to light was too bright for the camera.

IMG_0428.jpg

[Alan Lightstone]

Pre-twisting provides a very positive connection, IMHO. I've also now moved to Wago connectors, but if I do a wire nut, I always pre-twist with lineman's pliers, clean/trim the end and then apply the wire nut followed by tape.

It will be interesting to see what you find when you peel that apart to see why it's arcing.

I use similar connections for all my low voltage cabling connections. Didn't know about the Wagos. Do they take up more room in an outlet box? How do you size them for two different size wires in same connection (like hooking up the timer)?

[Jim Becker]

I use similar connections for all my low voltage cabling connections. Didn't know about the Wagos. Do they take up more room in an outlet box? How do you size them for two different size wires in same connection (like hooking up the timer)?

While I have the 10 gage version which will not work on smaller gage wire, the smaller sized version works for both 14 and 12 gage, I believe. There are some similar products at the home center (not Wago brand) that are 14/12, I believe) S

So if you need to join smaller wire to #10 for some reason, I suspect you'll still need to use the wire nuts. But that should not be a common scenario I would think...

[Michael Burnside]

Looks like it's Honeywell rpls740b, and the manual for it does say 15A @ 120V...

I wouldn't trust it to handle 15A continuous with the number of strands. If the circuit breaker is 15A and isn't tripping, then it's clear the weak point is the timer. A current meter would be a good thing to have in this scenario. My recommendation would be to replace the lighting with LED anyway.