Superficial GFCI autopsy

In the summers I run something I call a "lightning detector", which is simply an AM radio tuned way down to around 550 where there's no hearable station. Any electrical disturbance tends to make it crackle, and I can hear thunderstorms building up fifty miles out or more so it serves as a nice early warning. It also gives off an occasional click when some electrical connection in my or the neighbor's house is made or broken, but that's relatively rare as most switching is reasonably clean.

So there I am, dead asleep at like 4 in the morning, and I come half awake hearing this loud "...fffzzzzzZZZZZZZT! <click>" and then silence. The timbre of the noise immediately let me know I'd heard it out of the radio speaker and not some actual device in the house, so I lay there for a little while thinking "wtf??" and not hearing [or smelling] anything further and eventually drifted back off to sleep, and thought nothing of it on waking up later that morning.

Until, upon going to look something up, I realized that my network connection was dead. "Oh noez, Google isn't working, teh intertubez must be down!" Now, my FiOS link has been pretty rock-solid reliable for several years so it was somewhat surprising to wake up one morning and discover a connectivity outage longer than about five seconds. While the optical termination box looked like it was still alive and had its network uplink, a little debugging finally showed that it was running on its battery backup and for some reason that causes it to drop the link on the customer-side Ethernet port after a few minutes. But the root problem that finally sank into my foggy pre-coffee brain was that it had no AC power for some reason.

My electrical panel was recently upgraded as part of an extensive house renovation project, and the fiber box power supply was simply running from a new convenience outlet mounted on the panel backboard. This outlet was a ground-fault interrupter type, as it's in the basement and code says it has have GFCI protection either at the breaker or the receptacle itself. The electrician chose to install a simple GFI outlet. Except that now it was dead, and wouldn't reset to provide power. It began to dawn on me that the death throes of this receptacle might have been what I'd heard in the middle of the night.

I ran some temporary power over to the fiber setup, and pulled the GFI receptacle out of its box. Network all back and happy again, no problem. I then sent email to the electrician who had done this job less than a month before to tell him about it; he assured me that it could be replaced under warranty that very afternoon because his guys were already in the area, so I didn't have to worry about digging into my own stock of [non-GFCI!] outlets.


GFCI receptacle that went bad As this had all become a little more interesting, the item in question landed on the bench to undergo an initial autopsy.

[Click any image for a larger copy]


GFI faceplate removed There isn't much under the outer faceplate; just the output contacts. The reset button has a strong spring under it and is very likely to pop out of whatever's anchoring it in there at this point.

Getting down to the innards The next layer reveals more of the innards, including the current transformers, output contacts, some of the latch mechanism that holds the device in the on state, and the trigger solenoid. We can already see that something's not right with the solenoid winding...

The child-resistant insert protection mechanism is rather clever: if an object pushes into *one* hole it tilts a little plate one way which then locks against sliding sideways with little barbs; if both sides are pushed evenly then it drops back and slides sideways against a little spring and allows the holes to open for the plug prongs. Code refers to this as a "tamper resistant" receptacle, designated by the "TR" logo between the slots. Obviously, a kid armed with TWO paperclips can still zap himself, and even a working GFI won't offer much protection against that. But if Junior is that determined and clever and manages to survive the resultant Voltage 101 lesson, then let's just start training him as an apprentice electrician.

Ground-fault sense board Everything else comes out as a module, and flipping it over reveals a surprisingly complex board full of electronics.

This is very different from the old GFI I took apart a few years back, which was basically a passive mechanism using a purely mechanical latch held in a delicate balance of three or four different springs and a high-current dual-winding coil with hot and neutral sent through it, to directly sense load imbalance and magnetically trigger the release like a little mousetrap. It's clear that there's a lot more going on with this thing.


A toasted solenoid However, it's obvious that all that newfangled whizbang stuff didn't protect the solenoid, which appears to have become very warm and eventually behaved like a fuse. Not good, folks!

Evidence of high component heat Visible on the back cover is evidence that some components were running really warm for a while -- that characteristic dulling of the plastic, possibly from a little bit of the magic smoke leaking out too. But sometimes this matte appearance can result just from long-term heat. These receptacles do run a little warm, which I originally attributed to a simple current-limiting means for the LED but I now see that there's plenty more opportunity to dissipate a couple of stray watts back here.

I couldn't really go much farther into it, because around this point I called Leviton's customer-service / engineering people and they wanted me to send the whole thing in to them for analysis. Gladly, I said, and simply put it back together [and I will say that getting the reset button back into the right place was a *bitch*], boxed it up and shipped it to the address of the QA department they sent me via email, with a short note explaining what happened and expressing strong curiosity as to what they'd find. So I never even quite figured out how the solenoid and release mechanism is actually supposed to work. The only other one of these that's installed is powering my ventilation system, so I don't want to rip it apart ... it's probably not important enough to go out and buy another new outlet of the same type, either. If Leviton actually gets back to me on what went wrong, I'll update this page.

I learned during my call that Leviton's spec-grade GFCI receptacles are manufactured *both* in China and the US, and there's an option to order stock from a particular source. Guess where this one was from. That, like so many other situations I keep running into these days, says volumes about what outsourcing is doing to us.


_H*   120718