GFCI tripping after a while

[vr608]

Hey folks,

I'm still working on setting up my BIAB kettle, and it appears I'm having some GFCI issues. I recently exchanged my 2-pole 30A breaker (used to test the new line I installed) with one with GFCI, and now I cannot get run the system for more than a few minutes before the GFCI trips.

When I originally installed the GFCI, I bonded the pigtail to the neutral bar as normal, but also connected the load neutral (I used 10/3 wire for the run since I wanted to future proof, but my heating element just uses the two hots plus ground). The neutral is unconnected at the receptacle end (NEMA L6-30R). Under this configuration, the kettle would start to heat but would trip just before it (presumably) started to boil.

So today, after reading some threads on the topic and some online articles over the past couple of days, I decided to disconnect the load neutral from the GFCI, but under this new configuration the kettle won't run for longer than a minute or so before tripping the GFCI.

No control panel being is being used, I'm just connecting the element straight into receptacle. So far I've confirmed there is a good ground connection between the ground lug and the kettle (~0.05 ohms; using an 2-gang electrical box ala Kal), and no continuity between any of the hots and the kettle. My element reads about 10.6ohms between both hots. In an earlier thread I was dealing with some leaking issues associated with bad sealing of my kettle but that no longer is an issue, I have found no evidence of water leaking after customizing my own gaskets out of silicone.

Any ideas? I'm tempted to think there may be some residual water somewhere in my heating element from when I was testing out the kettle sealing, but otherwise have no other leads. Any insight would be helpful; I know this issue has come up often, but I know there are some electrical gurus on this forum that may that should be able to solve this quickly. Thanks in advance!

 

[ajdelange]

Clearly you have some current going where it shouldn't. As the neutral is and was not connected in either configuration it should not be a player and the fact that things trip faster now that it is disconnected at both ends as opposed to at just the distal end is strange indeed.

I assume the kettle itself is well grounded? It should be connected, directly or indirectly, to the bare wire in your feeder cable.

Disconnect the feeder wires from the kettle element and measure the impedance between the heater terminals and the kettle. This should be very high (M &#937. If it is not of that order then that's the source of your tripping. A voltage of 120V (even though the feed is 240V the maximum voltage reference to ground is 120V) will drive hundreds of uA through an impedance of 1 M Ω. You might also want to check the impedance of your kettle to ground. To do this get a remote ground from an extension cord plugged into another outlet. Measure between the kettle and the round hole in the extension cord's outlet. It should be at most an Ω or 2.

If the terminal to kettle impedance is less than M Ω then you have found your ground fault. The fix is probably a new element. The fact that you get a trip after a time delay suggests that the heating of the element is causing something to move thus allowing water to seep in and create the leakage path. That is, at this point, of course pure hypothesis.

Try those measurements and come back. We will go from there depending on what you find.

 

[vr608]

Clearly you have some current going where it shouldn't. As the neutral is and was not connected in either configuration it should not be a player and the fact that things trip faster now that it is disconnected at both ends as opposed to at just the distal end is strange indeed.

I assume the kettle itself is well grounded? It should be connected, directly or indirectly, to the bare wire in your feeder cable.

Disconnect the feeder wires from the kettle element and measure the impedance between the heater terminals and the kettle. This should be very high (M &#937. If it is not of that order then that's the source of your tripping. A voltage of 120V (even though the feed is 240V the maximum voltage reference to ground is 120V) will drive hundreds of uA through an impedance of 1 M Ω. You might also want to check the impedance of your kettle to ground. To do this get a remote ground from an extension cord plugged into another outlet. Measure between the kettle and the round hole in the extension cord's outlet. It should be at most an Ω or 2.

If the terminal to kettle impedance is less than M Ω then you have found your ground fault. The fix is probably a new element. The fact that you get a trip after a time delay suggests that the heating of the element is causing something to move thus allowing water to seep in and create the leakage path. That is, at this point, of course pure hypothesis.

Try those measurements and come back. We will go from there depending on what you find.

A quick update on this: I was able to get to a full boil using a 120V connection to a receptacle-based GFCI that is next to my 240 line. This is actually the 2nd time I've been able to get to a boil on 120V, in addition to the test boil I did after laying my 240 line (without the GFCI, dangerous, I know).

As for the tests above, I'm measuring infinity between both the element leads and the kettle, and my ground is still reading betweein 0.03 and 0.05Ω.

The bad news is that I now have a leak between the kettle and the box , so I have to go back to the drawing board trying to find a fool-proof way of sealing the kettle smh.

 

[benrenden]

Could be a faulty GFCI. Or Maybe check to see if your cord end connectors aren't too tight. I've once found someone had over-tightened a male cord end cap and it caused a gfci breaker to trip... If it always seems to happen after a certain amount of time then my bet is a faulty breaker. The way a gfci device works, load has no effect. It will only trip if there is a imbalance between current in each line (if you have any sort of leak to ground then you will have an imbalance).

In my setup i use a 40a gfci spa pack and i have the neutral tail connected even though my element is straight 240V. Its always worked well for me.

Getting it to not leak was also an issue for me, i realized my hole was not as clean as it should be and the slightest little "burr" of metal against your silicone gasket can cause a leak. Also a hot water element socket cams in real handy (less than 10$ at the home depot i think).
Hope any of this helped!

 

[ajdelange]

.

As for the tests above, I'm measuring infinity between both the element leads and the kettle, and my ground is still reading betweein 0.03 and 0.05Ω.

The bad news ....

Yes but the measurements are good news. Now we know that the bad current isn't sneaking through the element.

Next thing to try is, with the element still disconnected, open the breaker, check the voltage at the output terminals with a non contact checker or voltmeter and, if verified 0, then measure the impedance from each lug to ground (not neutral). Again the impedance should be very high. We are looking for leakage in the cabling and connectors at this point. If there isn't any we have no path in the wiring/connectors and no path in the element/kettle and that leaves no path and a great mystery.

Just to review - a GFCI contains a little doughnut through which three wires, the two hots and the neutral run. If the sum of the currents through those wires is other than 0 (or, in reality, close enough to zero) the breaker trips. What this means is that any current flowing out of the breaker (through a hot) must return through the breaker on either the other hot or the neutral. If any current going out on a hot makes its way back to the service entrance through any other path (the protective ground, i.e. bare wire, earth itself, a water pipe...) the breaker trips. Also to review: protective ground and neutral are separate wires joined only at the service entrance and nowhere else. I mention this because there is often misunderstanding of what a GFCI is (and isn't) and confusion as to the roles of the 'grounded conductor' (neutral) and 'grounding conductor' (bare wire) and, with names like that can you blame folks? This sometimes leads to wiring errors so check that.

 

[vr608]

Yes but the measurements are good news. Now we know that the bad current isn't sneaking through the element.

Next thing to try is, with the element still disconnected, open the breaker, check the voltage at the output terminals with a non contact checker or voltmeter and, if verified 0, then measure the impedance from each lug to ground (not neutral). Again the impedance should be very high. We are looking for leakage in the cabling and connectors at this point. If there isn't any we have no path in the wiring/connectors and no path in the element/kettle and that leaves no path and a great mystery.

Good thinking. It would appear that both lugs measure infinity to ground, so looks like there should be no current leakage.

Could be a faulty GFCI. Or Maybe check to see if your cord end connectors aren't too tight. I've once found someone had over-tightened a male cord end cap and it caused a gfci breaker to trip... If it always seems to happen after a certain amount of time then my bet is a faulty breaker. The way a gfci device works, load has no effect. It will only trip if there is a imbalance between current in each line (if you have any sort of leak to ground then you will have an imbalance).

I'm starting to think the GFCI may be the one at fault, although I'm open to more suggestions. I even tried plugging in the kettle into another 120V GFCI outlet, and I managed to get similar results to the other one, ie no trips for the duration of the tests (to compare apples-apples, I would need to let it run until boiling though).

Also, I think you're onto something with very tight cord caps; I opened mine up to double-check the connections and re-tightened it, and the GFCI immediately tripped on the next test. Once thinks were loosened it went back to its normal operation of tripping after about 30 seconds.

Getting it to not leak was also an issue for me, i realized my hole was not as clean as it should be and the slightest little "burr" of metal against your silicone gasket can cause a leak. Also a hot water element socket cams in real handy (less than 10$ at the home depot i think).
Hope any of this helped!

The leaking issues still continue, although I've confirmed it is definitely between the kettle and the box, ie no water is present inside the box or around the heating element connections. I already have an element socket wrench and it has been a very useful tool thus far.

Any other things to try? Definitely suspicious of the GFCI breaker at this point.

 

[ajdelange]

Also, I think you're onto something with very tight cord caps; I opened mine up to double-check the connections and re-tightened it, and the GFCI immediately tripped on the next test. Once thinks were loosened it went back to its normal operation of tripping after about 30 seconds.

That's telling. Disconnect the ground wire at the breaker and see what happens (be careful). If the breaker does not trip then you have found your fault. It's at the plug where the cord cap is squeezing one of the hots into sufficient contact with the earth wire to to permit trip level fault current to flow. I think the fix might be then to cut off a couple of inches of cord and rewire the plug being very careful when stripping. I assume you are using something suitable like SJOOW cable?

 

[vr608]

That's telling. Disconnect the ground wire at the breaker and see what happens (be careful). If the breaker does not trip then you have found your fault. It's at the plug where the cord cap is squeezing one of the hots into sufficient contact with the earth wire to to permit trip level fault current to flow. I think the fix might be then to cut off a couple of inches of cord and rewire the plug being very careful when stripping. I assume you are using something suitable like SJOOW cable?

Does it matter if I disconnect the ground wire at the breaker or at the element box, as long as its disconnected? I'd rather not touch the panel unless I have to (overly paranoid and will need to cut the main breaker in order to feel comfortable moving wires), and just unscrewing the lug on the box is easier.

I'm using a 6ft 10/3 dryer cord (neutral is taped off), its pretty thick.

 

[vr608]

So a few updates:

1. I tried the suggestion above of running the system ungrounded but the GFCI still tripped after about 30 seconds. I reattached the ground.
2. Next, I tried swapping the lugs on the element and retrying; the GFCI immediately tripped. I then removed the cord cover at the receptacle end, re-tightened the wire connections and retried. GFCI tripped just before 30 seconds.
3. Next, I stripped off about an inch of wire, re-did the receptacle connection and retried. Tripped after 30 seconds.

I then decided to test voltages without the element; HOT A showed about 121.6V max, HOT B peaked at 122.0. Interestingly enough, while waiting for HOT B to stabilize, the GFCI tripped!

My next test is to just try leaving the GFCI on without doing anything to see if it trips on its own. If it does, I'll reconnect the neutral cable and retry. If it still trips, I'll concede that its faulty and just send it back for replacement.

 

[ajdelange]

Does it matter if I disconnect the ground wire at the breaker or at the element box, as long as its disconnected?

To check out this thesis (and please remember that we are grasping at straws here) it would have to be disconnected at the breaker. I am theorizing that current is getting into the ground wire because the wires are being squeezed together tightly enough by the plug cap that an otherwise non problematic insulation flaw becomes problematic. This current is presumed to be finding it's way back to the source through the ground wire to the service entrance. We need to break that path on the way back to the panel to see if that's the case. The evidence is that the fault only develops after some period of time and an explanation of that consistent with this theory is that as current flows the conductors warm and expand increasing the pressure even more.

It would be a good plan to disconnect the ground wire at the element too as leakage current flowing into the ground wire could return through the kettle and stand through the actual ground if that path is of low enough impedance (which it could well be in a garage or basement but isn't so likely to be in a kitchen).

I'd rather not touch the panel unless I have to (overly paranoid and will need to cut the main breaker in order to feel comfortable moving wires), and just unscrewing the lug on the box is easier.

I don't want to encourage you to work in the panel if you are not comfortable doing so in fact I don't encourage anyone without professional training or a thorough understanding of what's going on inside one acquired through other means to do so.

If you don't want to work in the panel then you could try a new dryer cord or cut a few inches off the one you have and rewire the connector.

I'm using a 6ft 10/3 dryer cord (neutral is taped off), its pretty thick.

Just so we are clear: this is an assembly bought at Despot or some similar outlet that include 6 ft of wire with black, red, white and green (perhaps with yellow stripe) conductors and a plug which has a cap to retain the wires held on with screws that can be tightened and loosened. I thought these came with molded plug and am thus confused. Where exactly are these 'cap' screws?

 

[dyqik]

Left-field suggestion: Is there any way to tell the difference between the GFCI breaker tripping due to ground fault or it tripping due to overcurrent (like a regular breaker)?

The 30 second delay sounds a bit like a weak overcurrent breaker operated just above its overcurrent limit, which can take time to trip. Obviously this would require that the overcurrent portion of the breaker be bad, and actually tripping around 25A.

The fix would still be to replace the breaker, as it would if the GFCI circuit was faulty.

 

[vr608]

Left-field suggestion: Is there any way to tell the difference between the GFCI breaker tripping due to ground fault or it tripping due to overcurrent (like a regular breaker)?

The 30 second delay sounds a bit like a weak overcurrent breaker operated just above its overcurrent limit, which can take time to trip. Obviously this would require that the overcurrent portion of the breaker be bad, and actually tripping around 25A.

The fix would still be to replace the breaker, as it would if the GFCI circuit was faulty.

I was definitely thinking of the same when the element was connected, ie over current protection and not knowing whether that was the root cause or not.

Just so we are clear: this is an assembly bought at Despot or some similar outlet that include 6 ft of wire with black, red, white and green (perhaps with yellow stripe) conductors and a plug which has a cap to retain the wires held on with screws that can be tightened and loosened. I thought these came with molded plug and am thus confused. Where exactly are these 'cap' screws?

I actually got cheap and just ordered a 4-wire dryer cord and cut off the molded plug, replacing it with a NEMA L6-30P plug. So when I meant redoing the connections at the receptacle, it meant removing my new cap, inspecting/adjusting the wires from the dryer cord and replacing the cap.

I have some news; I think the GFCI is faulty. I disconnected both hot loads on the breaker (left pigtail connected, obviously) and just turned it on, and it still is tripping between 20-30 seconds. I even tried moving it to a different position in the panel, same result.

Its possible it is a refurbished unit and marketed as new (got off Amazon), or it could just be a faulty unit. Its a GE THQL2130GFTP in case that helps, not available at Home Depot so had to be ordered.

Of course, I'm open to more suggestions, but I think we have a smoking gun. I may consider a compatible Siemens breaker (since a good portion of the other breakers are Siemens) but would rather get a replacement GE breaker, since that is what my panel sheet recommends.

 

[dyqik]

Yeah, that sounds pretty conclusively like a bad breaker.

For absolute pedantic completeness, I'd check again with the neutral on the output side of the breaker disconnected if it wasn't before - a current from somewhere else returning down the neutral leg would be expected to trip the GFCI as it's an unbalanced current. Leave the pigtail connected. Quite where that could come from, I don't know, but while you're disconnecting the breaker, you might as well take a minute to check it.

 

[vr608]

Yeah, that sounds pretty conclusively like a bad breaker.

For absolute pedantic completeness, I'd check again with the neutral on the output side of the breaker disconnected if it wasn't before - a current from somewhere else returning down the neutral leg would be expected to trip the GFCI as it's an unbalanced current. Leave the pigtail connected. Quite where that could come from, I don't know, but while you're disconnecting the breaker, you might as well take a minute to check it.

Understood. Yes, all my testing since yesterday was with the load neutral capped, so there is no current coming from anywhere else. The last test seems pretty convincing, ie no load leads connected and still a trip after 30 seconds.

I wonder though whether a larger issue is at play, ie different voltages between the bus bars? How much of a deviation (if any) would cause a situation like this to occur? I guess I'm worried about replacing the breaker with another one and having the same situation occur.

 

[dyqik]

Different voltages on the bus bars shouldn't cause any issue with the GFCI - the current flow in and out of each leg will be proportional to the voltage differences between the various legs, but no current should be lost.

 

[ajdelange]

Yes, that sounds like a smoking gun to me (which is better than a smoking panel or smoking anything in your system).

You can get SJOOW by the foot at the Home Despot so if you want to do it right that is always a possibility.

 

[vr608]

Different voltages on the bus bars shouldn't cause any issue with the GFCI - the current flow in and out of each leg will be proportional to the voltage differences between the various legs, but no current should be lost.

That is good to know, I figured that should be the case but didn't know for sure.

Yes, that sounds like a smoking gun to me (which is better than a smoking panel or smoking anything in your system).

You can get SJOOW by the foot at the Home Despot so if you want to do it right that is always a possibility.

Thanks, I think I'll buy a few feet, since the 6ft cable I have is a wee bit short for where my kettle will be located.

I'll report back when I have a new breaker installed. Thanks guys!

 

[ajdelange]

Left-field suggestion: Is there any way to tell the difference between the GFCI breaker tripping due to ground fault or it tripping due to overcurrent (like a regular breaker)?

Ground fault trips are from mA of imnbalance. Thermal/magnetic trips are from large current magnitude in one or the other or both of the phases (there are two breakers in there toggled together so that both flip if either is tripped). GFCI's have a test button. Pressing it will trip it in ground fault mode. Since such tiny currents are involved you usually only hear a click. OTOH thermal/magnetic trips are usually accompanied by loud buzzing just before the flip. If you have an I/R camera the breaker and its wires will appear hot just after a thermal/magnetic trip. Of course it will too if there has been a large, but below trip level on it for a while before the ground fault occurs.

Just as a point of interest the last thermal/magnetic 'trip' I observed was when I miswired a 240 heater thermostat incorrectly putting the two hots across the thermostat contacts. When I went back to the panel and flipped the breaker it made lots of noise but did not trip choosing instead to fail open. It did interrupt the fault current, which turned out to be 500 amperes on each phase, in making this supreme sacrifice so I guess we have to say it did what it was supposed to do. 20 Amp breaker on a 200 Amp panel with 400 Amps service. Neither the panel nor service disconnect breakers (200A) tripped.

 

[vr608]

Update on this:
Definitely was the breaker, my new one has withstood multiple boil attempts as I've worked on getting my kettle leak sealed.

Again, thanks guys for the help in narrowing this down!