In the years working with design engineers for PQ instrumentation, occasionally all of the engineers would gather around a project that had a significant failure and resulted in a smoke cloud wafting up to the lab ceiling. We joked that all electronic components are made with smoke inside them. Sometimes, under the right circumstances, the smoke would be let out, and sometimes it did so with the bang of exploding components.

After a call today concerning an unusual electrical problem at a residence, I am beginning to think that even GFCIs have smoke built into them. The resident was attempting to replace a three-way switch with a three-way timer. With the circuit that fed the switches off, he used a portable LED light plugged into an extension cord, which was plugged into a GFCI. Shortly after starting, the light went out. A trip to the basement revealed that a 15A breaker had tripped. Resetting it, he went back to the task at hand, only to find the breaker tripped again. After several attempts to reset it, the resident noticed the GFCI had burned, melted and otherwise become inoperable.

The image to the right shows the Eagle 15A GFCI with its distinct sideways receptacles after being removed from the box, which was blackened along with parts of the wall. The cover’s backside was coated in melted plastic. Strangely, another breaker situated three positions above that breaker had also tripped and would not reset. It controlled the two-car garage circuit.

The investigation was similar to one for PQ. I started at the point of failure to see what could be learned from the damage about its cause, and then went to the source. The wires were removed from the two breakers to see if they had any internal failures, which they didn’t. The loose wire from the GFCI circuit had millivolts to ground. The wire from the other circuit was reading 0.00V on the digital voltmeter, a bit unusual as stray radiated voltage on open circuits usually shows a very low, but nonzero, number. The impedance of both circuits to ground was around 5–8 ohm—not usual. Reconnecting the wire and resetting the breaker caused it to trip immediately. There were two mysteries and an unknown relationship now.

Starting with the GFCI circuit and tracing all of the receptacles and lights that were out found them all in the room where the breaker panel and laundry are located as well as the GFCI used. It had wires off the load side of the GFCI to points unknown. Disconnecting the wires from the GFCI allowed the basement lights and receptacles to function again. No damage was observed on the line-side wires.

The neutral conductor on the load side showed 2 inches of melting insulation. There was evidence of arc damage between the neutral and ground contacts on the receptacle’s face. Opening up the GFCI found the neutral contact melted into several pieces and two balls of melted metal. One of them had lodged between the hot contact and the grounded tab of the GFCI. This is likely what caused the breaker to finally trip. Most of the inside of the GFCI was black ash, along with one of the pieces of the neutral contact.

The company that makes Eagle GFCIs is no longer in business. It’s likely that this GFCI was the original installed in 1997 when the house was built. The current homeowner could not recall having used this GFCI in the past 10 years.

We moved on to the second mystery circuit. Unplugging all equipment from the receptacles on the circuit and turning all light switches off allowed the breaker to stay on when re-energized. One at a time, each piece of equipment was plugged in after switching the breaker off. When re-energized, it stayed on. Fortunately, everything worked. Unfortunately, nothing was found to explain why the breaker tripped or wouldn’t stay on later.

The other mystery is where the load side of the GFCI wires go. Nothing else in the residence has been found to be unenergized with both breakers on. Before a new GFCI can be installed, this mystery needs to be solved. Both breakers are being replaced in case they were damaged or not operating properly. While it seems that the GFCI failed internally with the neutral load side being faulted to ground and eventually the hot side, the damage was so extensive that not enough was left intact to prove conclusively. Given the information (or misinformation) found on the web, other similar GFCIs in the residence are also being replaced.