NEVER trust the test/reset buttons on a GFCI

[mbrooke]

I said it a while back and I will confidently say it again: LED indicators, SmartLock technology, internal self testing and the test/reset buttons do not accurately verify that a GFCI is functioning correctly. They do not tell you the speed at which a GFCI will trip, nor do they demonstrate that power has successfully been interrupted once the receptacle has tripped. Plugging in a load does not guarantee it either.


Case in point (turn up audio for maximum experience ) :

https://www.youtube.com/watch?v=MGQXsCc11qc

Happily buzzing away:

https://www.youtube.com/watch?v=XACCFgc1q-A


The GFCI above trips and resets as it should, its indicator LED goes out when tripped and so do line to neutral connected loads. However, when a fault is placed line to ground it trips but continuous to pass lethal levels of current while buzzing away. I verified the wiring as being correct. The unit in question is only 5 years old.


I did the same test on another but identical GFCI and it does not do the same after tripping:


https://www.youtube.com/watch?v=M8phlIi42qY

 

[mbrooke]

And oh, yes I am aware that I am pushing the test button at 1:57 in the first vid but its it demonstrate the fact the unit is indeed tripped. It resets as soon as I hit reset.

 

[JFletcher]

Wow. Even if it werent wired correctly, I would think a unit just 5 years old would lock out power and not reset at all. I'd send your video to Leviton; perhaps a product recall is in order.

 

[cpinetree]

And oh, yes I am aware that I am pushing the test button at 1:57 in the first vid but its it demonstrate the fact the unit is indeed tripped. It resets as soon as I hit reset.

Wow.

That's all I can say

 

[mgookin]

Wow is right. That is incredible.

 

[mbrooke]

Wow. Even if it werent wired correctly, I would think a unit just 5 years old would lock out power and not reset at all. I'd send your video to Leviton; perhaps a product recall is in order.

Wow.

That's all I can say

Wow is right. That is incredible.

Wow is not to far off, though I am more POed. Has anyone seen this before? My guess is either the hot contact will not separate (break open) internally or a semiconductor is passing current through the electronics energizing the output. And yes, this will probably go to Leviton.

 

[mgookin]

How did you discover that? What led you to start that test?

 

[gar]

160818-0920 EDT

mbrooke:

I suggest that you get another identical GFCI, take it apart, and inspect its construction.

I took apart a Leviton 7899 back in 2008 and showed photos of it on this forum. I can not find my posts with either Google or Mike Holt's search.

The design of the 7899 is such that it has a mechanical latch for the contacts. When tripped it can not be electrically reset. The trip operation is a solenoid that is momentarily activated by an SCR. Once tripped the contacts have to be mechanically reset by the mechanical operation of the reset button.

There is not a third wire thru the current transformer for injection of the test current. The test current is 120/15,000 = 8 mA which is only a small amount above the minimum requirement. With this test current tripping is almost immediate and thus a good test of trip time. For testing the 15,000 ohm resistor is connected between one input wire before the CT and the other wire on the output side of the CT.

This device either trips or it doesn't, and it is not resetable except by the mechanical reset button.

.

 

[kwired]

160818-0920 EDT

mbrooke:

I suggest that you get another identical GFCI, take it apart, and inspect its construction.

I took apart a Leviton 7899 back in 2008 and showed photos of it on this forum. I can not find my posts with either Google or Mike Holt's search.

The design of the 7899 is such that it has a mechanical latch for the contacts. When tripped it can not be electrically reset. The trip operation is a solenoid that is momentarily activated by an SCR. Once tripped the contacts have to be mechanically reset by the mechanical operation of the reset button.

There is not a third wire thru the current transformer for injection of the test current. The test current is 120/15,000 = 8 mA which is only a small amount above the minimum requirement. With this test current tripping is almost immediate and thus a good test of trip time. For testing the 15,000 ohm resistor is connected between one input wire before the CT and the other wire on the output side of the CT.

This device either trips or it doesn't, and it is not resetable except by the mechanical reset button.

.

But why does he have voltage at the ungrounded terminal of the receptacle when if it is tripped that should be isolated from the supply?


mbrooke is there "load wires" attached to the receptacle? If so is there any "backfeed" of voltage coming from them?

I have to assume the "humming" is the trip coil of the device attempting to operate, but the device is already mechanically tripped, or at least partially tripped. Ungrounded side contacts "welded" together but grounded side still opens?

 

[steve66]

I have to assume the "humming" is the trip coil of the device attempting to operate, but the device is already mechanically tripped, or at least partially tripped. Ungrounded side contacts "welded" together but grounded side still opens?

I'm wondering the same thing.

With the receptacle tripped, do you still measure voltage from the receptacle line to ground?

 

[kwired]

I'm wondering the same thing.

With the receptacle tripped, do you still measure voltage from the receptacle line to ground?

Apparently his test lamp did.

 

[mbrooke]

How did you discover that? What led you to start that test?

A plug in GFCI tester would trip it but much to my surprise the neons would still glow unchanged.

 

[mbrooke]

Apparently his test lamp did.

Its a 100 watt rough service bulb, so that is some legit voltage.

 

[K8MHZ]

Thanks for the vids, great stuff.

If it were me, I would be dissecting the faulty GFCI and taking pictures. Leviton could have copies of the pictures and the receptacle after I dissected it.

I wonder if the bad GFCI is actually a counterfeit, not a real Leviton.

 

[mbrooke]

160818-0920 EDT

mbrooke:

I suggest that you get another identical GFCI, take it apart, and inspect its construction.

I took apart a Leviton 7899 back in 2008 and showed photos of it on this forum. I can not find my posts with either Google or Mike Holt's search.

The design of the 7899 is such that it has a mechanical latch for the contacts. When tripped it can not be electrically reset. The trip operation is a solenoid that is momentarily activated by an SCR. Once tripped the contacts have to be mechanically reset by the mechanical operation of the reset button.

There is not a third wire thru the current transformer for injection of the test current. The test current is 120/15,000 = 8 mA which is only a small amount above the minimum requirement. With this test current tripping is almost immediate and thus a good test of trip time. For testing the 15,000 ohm resistor is connected between one input wire before the CT and the other wire on the output side of the CT.

This device either trips or it doesn't, and it is not resetable except by the mechanical reset button.

.

I am planning on doing just that. I replaced the GFCI, but I am debating whether or not I should take it to bits myself or send it to Leviton. I made one mistake burning something out in the process. I kept the lamp in there a good 30 seconds playing around with the test/reset to see if I could get the lamp to go out and it started to smell like burning plastic. When I tried to re-set it afterward it would no longer do so, the receptacle become dead. My best guess is that the solenoid coil overheated but there is only one way to find out...

 

[kwired]

I'm still betting on welded contact on the ungrounded side. It is fairly obvious it is opening the grounded side or else the lamp would work when tripped when plugged into the normal current carrying slots of the receptacle.

Did you reset (close the circuit) into a short circuit or full volts ground fault at some time?

 

[mbrooke]

Thanks for the vids, great stuff.

Welcome


If it were me, I would be dissecting the faulty GFCI and taking pictures. Leviton could have copies of the pictures and the receptacle after I dissected it.

I wonder if the bad GFCI is actually a counterfeit, not a real Leviton.

Im guessing bad GFCI, it looks legit in all respects. However, I am wondering how often this failure mode takes place. I stumbled upon this and low and behold it also looks like a Leviton:


https://www.nachi.org/forum/f19/g-f-tripped-and-still-has-power-81497/


The above is exactly what happened to me btw.

 

[mbrooke]

I'm still betting on welded contact on the ungrounded side. It is fairly obvious it is opening the grounded side or else the lamp would work when tripped when plugged into the normal current carrying slots of the receptacle.

My guess as well. Either welded together or somehow not parting. Either that the electronics feed power from both line and load which I doubt.

Did you reset (close the circuit) into a short circuit or full volts ground fault at some time?

[/QUOTE]

Not that I know of.

 

[steve66]

Apparently his test lamp did.

Yes, that is apparently what is going on, but it would be nice to verify somehow.

Either verify the voltages are what we expect (L-N = 0, L-G = 120, N-G =0), or even better - take the receptacle out and with an ohmmeter verify that the Line slot terminal is still connected to the hot input terminal, but the neutral is open.

You could probably still verify that even thought the receptacle shot craps.

 

[FionaZuppa]

how many folks actually test ocpd buttons for gfi afi? i pressed them in my own panel, on the 3rd one it just went hummmmmmm with some smoke, i knocked the main to off. took out the ocpd that was nice and warm and replaced it, sent the ocpd back to eaton, never heard back from them.