refrigerator tripping GFI

[MarineTech]

Just for reference I have included a link to the LM1851.

http://www.ti.com/lit/ds/snis158/snis158.pdf

 

[ELA]

Very familiar with the LM 1851 data sheet.
Like to view a link to the LM 1951 data sheet that was mentioned. Not the 1 amp switch, but a device that replaces the 1851?

 

[hurk27]

We agree and yet we do not.

Unless you are talking about times on the order of one nanosecond, the time it takes light to travel about a foot, (corresponding to a frequency on the order of 1 GHz), there is no way that the currents can be unequal at two different points in a single closed loop circuit. The electrons have to be going somewhere.

What causes the effects that you observe with spikes, inductive kickbacks, and other transients is current through relatively small stray capacitances that make no noticeable difference at 60Hz but become important at higher frequencies. So when I am talking about reactive current paths, I am including ones in the range that you would not notice unless you were specifically looking at transients and high frequencies.

Let's now assume that the exact same current waveform was applied through both windings of the GFCI current sensor. If the coils are not necessarily identical in characteristics across that wide frequency range, you could then get a non-zero difference signal. One simple way for this to happen would be if one of the coils has a little more intra-winding capacitance than the other, so some of the current at high frequency is actually bypassing the coil instead of flowing through it.

That does not change the fact that the same current is flowing in both wires, it just means that the GFCI is not well enough engineered to behave well at that frequency. And, as you point out indirectly, it is easier to use a time delay to ignore those signals than to engineer them out, especially if part of the problem is capacitive couplings within the appliance. This delay takes care of both problems with one stone.

I have no idea why I went back to thinking that it was a sync of two different frequency sources causing this problem, I think it was you or ELA in another thread that got me turned in the right direction as to it being a more of an inductance problem to possibly the EGC since it is closest to the hot in NM type wiring or the conduit, I now think it maybe because of the high frequency nature of the inductive kick back that may cause a inductive coupling to the parallel EGC or conduit and that is causing the imbalance when this happens, he also pointed out that I had mistakenly said it was a 3 cycle delay (50ms) which I was wrong on, but more like a ? cycle to just over a ? cycle (8 to 10ms) because of the trip time of the mechanical trip mechanism that would keep the trip time within the UL required 25ms but it does very proportionally to the amount of current of the ground fault or current taking a different path, I never did calculate the approximate frequency of the inductive kick back current but on the scope it would look like it was somewhere above 500hz to maybe at least 1mhz unfortunately the quad trace I have is not a capture scope and doesn't have an output so I can post the wave forms.

I think what got me to thinking it was a sync problem is all newer (after early 90's) GFCI's do have two CT's but the second one is called a 120hz drive coil and is used to inject a 120hz current into the circuit ahead of the imbalance detection CT to detect a neutral to ground fault without any current flow which is why we see a GFCI trip as soon as a neutral to ground fault happens or as soon as it is energized even if there is no load on the circuit, AFCI's do not have this coil and is why a grounded neutral only happens after a load is applied, but for some reason I was thinking that the 120hz was how it worked but now I see that this would not be the case without any 60hz load applied as it still works.

I posted here that the GFCI can trip faster ( under normal conditions) than some were/are stating as supposed fact.

http://forums.mikeholt.com/showthread.php?t=149465&page=2&p=1445387#post1445387

I posted here that GFCI's can trip even more immediately due to EMI. In many of these cases the differential sensing core may be completely bypassed.

http://forums.mikeholt.com/showthread.php?t=152331&page=3&p=1473338#post1473338

In this thread a differential current meter should be used to confirm that excessive leakage is, or is not, present.

The problem is or was, I should have used CT's to measure the current in all conductors I had a quad trace scope but I only used a CT on the hot and assumed it was just a sync problem between the 60hz and the higher frequency of the inductive kick back, it I had a recording scope and CT's on the hot EGC and neutral I would have seen my thinking was wrong, I still think that inductive kick back is a problem as I saw this on the scope as I repeatedly tripped the GFCI using a solenoid I think I used one from a washer fill valve I had for parts would trip the GFCI about 60% of the time, also I tried it with my shop vac but it wouldn't trip as often, also it was always when I turned it off never when turning on.

Not sure when they started using the LM 1851 (LM1951 was a typo above) but it had to be about that same time they added the second CT to detect grounded neutrals, but I noticed that after this point most inductive kick back problems went away until I ran into a problem at a house that had the cheap Menard's ones made from China, had an apartment with them everywhere and several inductive loads not even on the same circuit would cause them to trip, I found putting a cap or an MOV across the switch (line to load) would lower the tripping to almost none at all, but putting them between the hot and neutral on the load side of the switch did nothing I wound up just changing out all the GFCI's to Leviton's and never had a problem after that, but I never checked out as to why putting the MOV across switch drastically lowered the tripping maybe I should do that someday.

Point taken. Certainly the small sensor coil in an outlet-mounted GFCI works exactly as you state.
Larger imbalance detection systems at a source or breaker panel have other options, but they can also be built in exactly the same way.

I added a PDF below that gives a good idea of how the GFCI's work and most are manufactured like this as well as the breaker type.

 

[hurk27]

Very familiar with the LM 1851 data sheet.
Like to view a link to the LM 1951 data sheet that was mentioned. Not the 1 amp switch, but a device that replaces the 1851?

Yep that was a typo:ashamed1:

 

[svemike]

Same thing

Same thing

I temporarily disconnected the frame ground wire from the cordset, and refer worked fine, no trip, reconnected, tripped instantly. Obvious current leak to ground. Worked good in non-GFCI protected circuit. Appliance repair company said not to put on GFCI, and after talking to 3 technicians, all with a different story on how GFCI's work, (None Correct), I finally was able to relocate the GFCI downstream from Refer, but still covering all countertop recepts. Tile Floor, now waiting for HO to get shocked. Leviton says they do not recommend them on refers, but did not say not to. How about the thousands that work OK, including mine.

 

[kwired]

I temporarily disconnected the frame ground wire from the cordset, and refer worked fine, no trip, reconnected, tripped instantly. Obvious current leak to ground. Worked good in non-GFCI protected circuit. Appliance repair company said not to put on GFCI, and after talking to 3 technicians, all with a different story on how GFCI's work, (None Correct), I finally was able to relocate the GFCI downstream from Refer, but still covering all countertop recepts. Tile Floor, now waiting for HO to get shocked. Leviton says they do not recommend them on refers, but did not say not to. How about the thousands that work OK, including mine.

You have something malfunctioning in this appliance. Do not listen to appliance repair guys. I think I said it earlier in this thread, I have on more than one occasion had to find the problem myself. Appliance repair guys in general have no idea what a GFCI is all about. I have found melted wiring from a failing connection end up faulting, I have found bad defrost heater elements. All tripping GFCI's and appliance repair man told HO that they needed to have an electrician replace the GFCI. Bill customer for what you do and let them deal with the repairman for his incompetence.

 

[Rewire]

Had a service call for fridge tripping GFI in garage. I unpluged it rest GFI tested it and it worked properly. I thn removed the backplate and disconnected the ground fron the frame. I plugged fridge in and it worked with th HO watching I placed my meter from the frame to the disconnectd cord ground and registered 120V I unplugged fridge reconnectd ground and explained to customer that if he had touchd the frame he would have became the ground and 120V would then pass through him . He replaced the fridge with no questions.

 

[hurk27]

I temporarily disconnected the frame ground wire from the cordset, and refer worked fine, no trip, reconnected, tripped instantly. Obvious current leak to ground. Worked good in non-GFCI protected circuit. Appliance repair company said not to put on GFCI, and after talking to 3 technicians, all with a different story on how GFCI's work, (None Correct), I finally was able to relocate the GFCI downstream from Refer, but still covering all countertop recepts. Tile Floor, now waiting for HO to get shocked. Leviton says they do not recommend them on refers, but did not say not to. How about the thousands that work OK, including mine.

I have just one question, in you original post you said you were installing 3-prong receptacles on a two wire system, and that there were no EGC's in the NM feeding these receptacles, if there is no EGC or anything connected to the grounding terminal then disconnecting the frame ground bond should not have done anything, the ground terminal on a GFCI is not connected to anything in the receptacle except the ground prong, so if they are fed by a two wire NM there should have been no difference, if they are fed by BX and the BX casing is grounded then yes this would have had an effect, are you sure there is no ground, some older NM had a small 16 awg EGC, some of these it was even aluminum and they didn't bring it into the box, they would double it back on top of the sheathing and would be connected when the NM box connector was tighten down, did you look in the panel to see if there were any EGC's if so then the above might be what you have?

 

[mdpolan]

GFCI & the Frig

GFCI & the Frig

OK we all know how a GFCI works. SO there are a limited numbers of possibilities as to why the GFCI trips.The current MUST be going some where. Yes we all have herd the story about the old refrigerators tripping GFCI's so what gives? I assume the plug on the frig is a 2 prong? I would check the frame of the compressor with a known good ground point. That might give you some clues. Hey if the frige is that old it's costing bucks to operate it. New frigs use like 450 watts compared to some older ones at 1200 watts.