So I noticed with my infrared thermometer that a two year old GFCI was eight degrees higher than ambient with nothing on it. Having seen this with standard outlets before that nearly caught fire, I wanted to take a look inside to make sure there was no loose connection. Flipped the breaker off, took a look, tightened the terminals, all was well, noticed the loop on the hot lead was kind of mangled and thought I should fix it in case there was a hairline crack causing resistance, far fetched but since I'm in there I made a pigtail, good tight pre-twist with a tight appropriately sized wire nut, and put it back on the hot terminal. Flipped the breaker back on, and when I hit the reset button it started buzzing LOUD. There was a 20 to 30 second delay for someone to run and flip the breaker off (I just told them to hit the main since it would be quicker), in those last moments the buzzing stopped and smoke started coming out. With the power off I quickly pulled it back out curious if the wires would be warm/hot. They seemed fine, no discoloration, no charring, no melting. The GFCI felt was warm though, mostly centered near the back and right side (hot side). Put a standard outlet in overnight and fixed it the next day with a new GFCI, new one functions fine.
I took the bad one apart (still smells burnt even now) and found some damage inside. Take a look. The varnish on the relay got melted and crispy. And there is one other little lightly charred area close to the hot terminal which should explain the warmth on the right side. Both problem spots are circled in red.

So what could have went wrong? Because everything was fine. No loose connection, no insulation under the terminal screws, hot was on hot, neutral on neutral, ground on ground. Is it coincidence it picked a time to fail after messing with it? Was it the testing mechanism stuck shut? Could that heat the solenoid? But I see no charring or fusing on the contact points, they are still clean. Could over tightening the terminal screw cause do something to damage or push on something inside? This was a Leviton GFCI, bought at Home Depot so it shouldn't be a knockoff or anything. Amp meter seems to indicate nothing else is on that outlet except that one 14 gauge wire. Here's my worry, theoretically if a breaker was unable to pop on an overload due to a short (like the tester getting stuck shut) would the warmest part of the wire be near the short or do I need to worry about the possibility of damaged wire insulation inside the wall? The breakers are pretty old. No visible damage on the wires, and they did not feel warm. So I THINK it was all contained in the GFCI, something faulty in the electronics. Lastly, the new one is still the same temperature so I suppose a GFCI a fair bit warmer than ambient, eight degrees isn't nothing.

 

I bent those top contacts out of the way for a better view. It appears there was arcing between these two pieces of metal, there is a small gap between them. Not sure if they ever touch, maybe this is how the tester works? Not sure.

 

More can be seen with the solenoid bent up out of the way here. You can also see there was some burning on the neutral side as well, but to a lesser degree.

 

There was likely a poor connection where the discoloration is located which causes arcing and heat. GFCI's do not trip on overload, they trip on the imbalance of current between two conductors.

I do not think 8 degrees is anything to worry about.

 

PCB creepage specs are designed to accommodate typical overvoltage events, such as lightning. Also, to accommodate high humidity conditions. The raw PCB itself has specs, resulting in passing tests called CTI, if you want to research that one. All that said, it appears this arc could have been initiated by an overvoltage even that exceeded the standard it was designed to. Was there any history of a close lightning hit? or a flood?
In the end; time to replace the unit.

PS; note the "FR4" that is silkscreened on the PCB. That indicates a fire resistant blend of materials was used in the fabrication of the raw pcb. If it didn't have that, your wall might have been involved too.

 

It doesn't really explain your situation, but I had a GFI that exhibited a similar death spiral following a nearby lightning strike (down a tree about 20' from the house). I didn't end up taking it apart, but it was charred, smoked a bit, and buzzed until it tripped the breaker.

The same strike took out a baseball-sized chunk of foundation and fried most of the electrical system of the wife's car too.

 

No history of over voltage or lightning strikes I know of that effected anything in the house in the last two years since this GFCI was installed. Several power outages a year though, live in a rural area, so who knows. Usually lightning strikes do damage to at least multiple appliances in the house.
After working on it, with the cover still off before the reset button was pushed I used my multi meter and tested the voltages from hot-neutral and hot-ground, it was 116 volts (very typical for this house). Before I worked on it the test and reset button worked too. When it was done being worked on when I went to hit the reset button I got the buttons mixed up and pressed test first with it off, realized my mistake and hit reset, after reset was pushed is when it went insane.
There is another wire just running to the box, but the hot and neutral and just capped with wirenuts. The hot is not live. I am almost positive this is where an old bathroom fan switch was once connected since it's old switch has no power going to it, my tester does not beep at either the switch or this old capped wire. The old bathroom fan just has cupboards going over it now, and it was functional before being covered, but I do know it was disconnected, this outlet is likely where it was disconnected, this was 15 years ago. The old and disconnected bathroom fan line and supple line do share grounds however, and the way it was done, it wasn't done with a pigtail and wirenut, but just a simple twist and then one wire (coming from the dead side) was the one looped around the ground. I wasn't too thrilled by how this twist was done, it didn't look too tight, so when I installed the new GFCI (the other day) I fixed it and pig tailed the grounds with a wire nut for a better connection, figured a shared ground shouldn't hurt and since it's been like this for 15 years (and over 30 years prior when the fan was connected) with no problems I'll leave it just in case because who knows, might as well leave the grounds connected like they've always been. But the way it was done might have given a loose ground connection, possibly maybe. So could a bad ground connection in a GFCI cause this failure to happen? I'm stumped, if it wasn't that it must have been something bad in the GFCI. Maybe tightening the terminal screw caused a solder joint to get a hairline crack that is invisible to see? I was working over a sink, I was trying to be careful not to get anything wet, but maybe a tiny drop of water managed to get on the terminal screw? I did have it all the way out at one point. Could that be it? Just simply faulty and time to go out with funny timing?
Again back to my first post, I like to think of circuit breakers as a backup device, I don't like to rely on them. So let's just assume the breaker supplying power to this outlet has something wrong with it and could never pop no matter what, pretend it's hardwired (it's not but let's say it is). In an over voltage or short circuit event that could cause the GFCI to fail like we see in the pictures, would the wire start to melt behind the outlet first or do I need to worry about damage to the insulation in the wall? There was 20 to 30 seconds before the breaker was thrown off. Wire does not appear to have gotten hot at the outlet or the breaker box (took a look with the panel pulled off).
Thanks for the responses and reading my insanely long posts by the way.

 

Do I just chalk it up to a malfunction that caused a loose connection then? The new one is thus far functioning normally, a bit nervous though since it too is a Leviton and I just took a look at the box and it says "Product of China" on it. Not amused. The reviews are mostly good, and it is the brand Home Depot is in contract with. But still... Had the power not been cut off what would have happened? Makes me hate GFCIs, I have been trying to read up on them and similar events have happened to others. They are too complex inside.

 

I have at least 8 of that model in my home. No issues to date. There are probably millions out there.

 

Big box stores will carry a particular name brand which also happens with supply houses. Certain manufactures will work with certain distributors. Leviton is a good brand just like Pass & Seymour/Legrand or Hubble. The ones that I would stay away from are odd brands like "Smart Electrician".