NEVER trust the test/reset buttons on a GFCI

Status
Not open for further replies.
mbrooke said:
Dude! your being Un-American! :( Thats why self testing was mandated by UL:


http://www.leviton.com/OA_HTML/SectionDisplay.jsp?section=66915&minisite=10251



:p:p:)






Doesn't mean they can't learn.
Click to expand...
Does the "self test" check time/current curves? I highly doubt it, it likely just verifies that the electronics of the device are functional at any tolerance level. If the control contact(s) get stuck/welded like yours apparently has, you still still get similar results as you had - even if that problem were detected how would it "unstick" the contact?

I trust even the first generation of GFCI to lessen injuries/save lives over not using GFCI at all, over the years we just happened to make some improvements to them. They are man made and are subject to failure. Why do the receptacle types require the self test feature to be listed but the GFCI breakers do not and AFAIK are about the same thing as they were years ago?

Considering you don't usually see a lot of GFCI breakers compared to GFCI receptacles, I'd say their failure to pass their own push to test is pretty high on older units compared to receptacles from my experiences.

HI's and electrical - IMO they need to stick to pressing the test button and if it doesn't trip put that in their report and not get any deeper then that. If we ever decide we need to test time/current curves and that a HI is going to test it, they need a simple plug in device that does the testing and gives either a go or no go result. Sure there are a few with knowledge to do more then that, there are also many that may not even get that simple test right.
 
Last edited:
Wasnt there a mandate last year to have GFCIs self test and trip/lock out if they failed? Maybe the issue has already been addressed.

Not sure if it's possible to design a truly failsafe GFCI. If the solenoid hangs/freezes, and keeps the hot side hot like a McDLT, then what? Is it physically impossible due to the way it's constructed for this to happen?
 
kwired said:
Does the "self test" check time/current curves? I highly doubt it, it likely just verifies that the electronics of the device are functional at any tolerance level. If the control contact(s) get stuck/welded like yours apparently has, you still still get similar results as you had - even if that problem were detected how would it "unstick" the contact?
Click to expand...


My point exactly. Self testing does not and will not address mechanical failures. It simply adds another feel good price point without offering a guarantee fail safe.

I trust even the first generation of GFCI to lessen injuries/save lives over not using GFCI at all, over the years we just happened to make some improvements to them. They are man made and are subject to failure. Why do the receptacle types require the self test feature to be listed but the GFCI breakers do not and AFAIK are about the same thing as they were years ago?

Considering you don't usually see a lot of GFCI breakers compared to GFCI receptacles, I'd say their failure to pass their own push to test is pretty high on older units compared to receptacles from my experiences.
Click to expand...


Bear in mind it was the NEC's failure to mandate grounding receptacles immediately after WWII that made them so essential. Thats not to say they have no place in protecting lives such as with a frayed cord, but most electrocutions took place from energized appliance frames.
 
JFletcher said:
Wasnt there a mandate last year to have GFCIs self test and trip/lock out if they failed? Maybe the issue has already been addressed.
Click to expand...

There was:


http://www.leviton.com/OA_HTML/SectionDisplay.jsp?section=66915&minisite=10251

Not sure if it's possible to design a truly failsafe GFCI. If the solenoid hangs/freezes, and keeps the hot side hot like a McDLT, then what? Is it physically impossible due to the way it's constructed for this to happen?
Click to expand...


Three points I like to add:

1. I think better construction of the GFCI could have prevented this.

2. The mechanism should be deigned such the reset button only pops out if both contacts have fully departed.

3. Actual testing involving the EGC is the only way to tell power has been cut. I long believed the test/reset was enough, but apparently it can deceive as this is not the first time its been mentioned:

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


In this video, which is the manufacture's recommendation for testing, how would such demonstrate bonefied proof the GFCI will not pass lethal current when called upon?


https://www.youtube.com/watch?v=5mSroP-HWZ0


To be blunt every GFCI could fail in this manner and no one would know until an actual incident took place and a through investigation was completed. Till then a very serious problem may be going unnoticed.














 
Last edited:
mbrooke said:
1. I think better construction of the GFCI could have prevented this.

2. Actual testing is the only way to tell power has been cut. I long believed the test/reset was enough, but apparently it can deceive as this is not the first time its been mentioned:
Click to expand...

Better construction is always an option, where do you stop with "what if's"

Actual testing is the only way to verify power has been cut whether it is a GFCI, a switch, a contactor... things can and will fail.

I have played with GFCI breakers to see what simple failures can occur - I particularly remember a 2 pole Homeline will close contacts when you turn the switch on, even if you have lost voltage on one of the lines or even if you don't have an input neutral. I had also concluded from that experience finding out that if you lost the ungrounded conductor to the side that powers the GFCI sensoring device or lose input neutral, you have no GFCI protection on any remaining functioning input voltage. I see this as a potential design problem but UL seems more worried about what seems to be lesser issues with GFCI receptacles.
 
kwired said:
Better construction is always an option, where do you stop with "what if's"

Actual testing is the only way to verify power has been cut whether it is a GFCI, a switch, a contactor... things can and will fail.

I have played with GFCI breakers to see what simple failures can occur - I particularly remember a 2 pole Homeline will close contacts when you turn the switch on, even if you have lost voltage on one of the lines or even if you don't have an input neutral. I had also concluded from that experience finding out that if you lost the ungrounded conductor to the side that powers the GFCI sensoring device or lose input neutral, you have no GFCI protection on any remaining functioning input voltage. I see this as a potential design problem but UL seems more worried about what seems to be lesser issues with GFCI receptacles.
Click to expand...



Of course, anything can fail. Which raises a concern for me. One method this could be avoided (at least the chances reduced) is a requirement that the reset button will not pop out when either contact does not fully depart. IMO this would have added more to safety than the self test requirement.

I understand that nothing is in theory 100% fail safe, however various design considerations can significantly reduce the risk as with carbon monoxide and smoke detectors. How many smoke and carbon monoxide alarms today fail to operate when called upon? Very few with good batteries and not over 10 years old. In fact nearly all of them fail in false alarm mode which is what we prefer. Most carbon units will actually beep when a 10 year timer runs out.
 
mbrooke said:
In this video, which is the manufacture's recommendation for testing, how would such demonstrate bonefied proof the GFCI will not pass lethal current when called upon?
Click to expand...

you are right, its really no diff then any other probability of failure that applies to everything. heck, my 10yr old lexus has a really weird "maybe" issue just discovered, if you start it and put trans into gear within 6sec of starting, there is a possibility the airbag will deploy unexpectedly !! its a software issue, but i have not yet received the fix notice.

due to sampling procedures we get some confidence that the crud works as expected. a few bad ones will get by, or, a few of them may become bad after installation. does anyone fly in airplane? risk for everything, how much risk is tolerable to you?

can a gfi recept be made better, surely yes, it will just cost more, etc. same applies to std ocpd's, they can house really nifty electronics, etc.
 
FionaZuppa said:
you are right, its really no diff then any other probability of failure that applies to everything. heck, my 10yr old lexus has a really weird "maybe" issue just discovered, if you start it and put trans into gear within 6sec of starting, there is a possibility the airbag will deploy unexpectedly !! its a software issue, but i have not yet received the fix notice.

due to sampling procedures we get some confidence that the crud works as expected. a few bad ones will get by, or, a few of them may become bad after installation. does anyone fly in airplane? risk for everything, how much risk is tolerable to you?

can a gfi recept be made better, surely yes, it will just cost more, etc. same applies to std ocpd's, they can house really nifty electronics, etc.
Click to expand...

Which is why GFCIs should be thoroughly tested.
 
mbrooke said:
Which is why GFCIs should be thoroughly tested.
Click to expand...

tested how? regression testing? one-time test? production line sampling? every unit? full blown field testing?

what would you like to see as "thoroughly tested" from a testing perspective.
 
FionaZuppa said:
tested how? regression testing? one-time test? production line sampling? every unit? full blown field testing?

what would you like to see as "thoroughly tested" from a testing perspective.
Click to expand...

Testing that in the very least involves the EGC. I think the manufacture's recommendation and UL's recommendation on relying on the test/reset button is asking for failures to go unnoticed. I am starting to come across more and more threads/videos where the exact same thing is seen: the GFCI passes UL's and the manufacture's recommended end user testing but is still somehow able to pass lethal current.
 
Last edited:
Very interesting thread.

I would like to find out if the problem with the stuck contacts are limited to one manufacturer.

The next thing I would do would be to try to find out if the failed units were feeding other receptacles.

Of course, taking them apart would be in order.

I have now, officially, changed the way I test GFCI receptacles. After hitting the test button, I will check for power to ground. That won't be easy if there is no ground, but I will figure something out. I would like to get my hands on a failed unit that I could used on a two wire system to figure out the easiest way to test them.
 
mbrooke said:
Testing that in the very least involves the EGC. I think the manufacture's recommendation and UL's recommendation on relying on the test/reset button is asking for failures to go unnoticed. I am starting to come across more and more threads/videos where the exact same thing is seen: the GFCI passes UL's and the manufacture's recommended end user testing but is still somehow able to pass lethal current.
Click to expand...
can you expand some about what you mean "involves egc"?

9846d1240350660-why-do-gfci-outlets-have-4-lead-4-load-holes-gfci-schematic1.jpg



welded contacts issue already has a patent
see https://www.google.com/patents/US7336457
 
Last edited:
K8MHZ said:
Very interesting thread.

I would like to find out if the problem with the stuck contacts are limited to one manufacturer.

The next thing I would do would be to try to find out if the failed units were feeding other receptacles.
Click to expand...

This one was.


Of course, taking them apart would be in order.

I have now, officially, changed the way I test GFCI receptacles. After hitting the test button, I will check for power to ground. That won't be easy if there is no ground, but I will figure something out. I would like to get my hands on a failed unit that I could used on a two wire system to figure out the easiest way to test them.
Click to expand...



Good idea IMHO.
 
Status
Not open for further replies.
Top