I wired a bath renovation with all power routed through a GFCI outlet. With power on, and all switches off, when I touch the white wire and the bare copper wire in a fixture box, the GFCI trips (did it accidentally at first).

Here is what I have:

  1. Dedicated circuit from the breaker panel to the bath GFCI
  2. Load out of GCFI to 3 gang switch box
  3. In switch box, COPPER WIRE from the LINE in from GFCI and COPPER WIRES from 3 fixtures connected together with a pigtail out to ground the box
  4. In the switch box, WHITE WIRE from the LINE in from GFCI and WHITE WIRES from 3 fixtures connected together
  5. In switch box, BLACK WIRE from Line in from GFCI connected to 3 BLACK WIRES pigtails, one of each then connected to each switch
  6. In switch box, BLACK WIRE from each fixture connected to one pole of each of the switches

All fixtures work fine without tripping the GFCI. If I hook up a non-GFCI outlet to a fixture box, and plug in a 3-prong appliance, it works fine. I checked to ensure the Line/Load and polarity on the GFCI is correct, no visible bare wires anywhere, all connections are tight. Tried a new GFCI and the same occurs.

So, with all that, with all switches off, why would the GFCI trip if I touch white to copper in one of the fixture boxes? Seems to me there shouldn’t be ANY current in the white wire, let alone bleeding in the ground. Why would the GFCI trip if I touch the wire ends, but not trip when I hook up an outlet and appliance?

Maybe it's a predictable anomaly. But I thought GFCI's worked on change in resistance in the ground, and would think a direct connection of the white and copper to have low resistance versus an appliance. Ideas or explanations?

  • Hi Paul! It looks like you’ve accidentally created a couple of accounts. If you’d like to merge them (there are some advantages to doing so), the instructions are here. Welcome to the site!
    – Niall C.
    Dec 3, 2014 at 18:13

4 Answers 4


GFCI devices work by comparing the current incoming on the hot wire to the current leaving on the neutral wire. If there's a difference, it means electricity is leaving the circuit somehow (a "ground fault") and the GFCI cuts power*. The GFCI doesn't do anything with the ground conductor—in fact, you don't even need to have one at all. One way of upgrading a 2-prong outlet (without ground) to a 3-prong outlet is by inserting a GFCI device upstream or using a GFCI outlet.

So I'm guessing there is a small amount of current in your scenario, and when you touch the white and ground together, some of that current goes into the ground and doesn't return through the GFCI neutral. Do any of you lights have timers, nightlight, etc.?

(BTW I'm not sure it matters for this question but your description of how the wires up is a little confusing to me. LINE and LOAD are opposite sides of the GFCI device, but you use them interchangeably. If you want a downstream device to be protected by the GFCI it must be connected to the LOAD side. Anything on the LINE side is considered upstream and is unaffected by the GFCI.)

  • THanks for the notes. I thikn I have it answered. You're right, I used line/load interchangably...the load out of the GFCI became the "line in" to the switch box
    – Paul
    Dec 3, 2014 at 20:11

The GFCI is tripping on its own power draw.

GFCI works by measuring the current on the hot and neutral wires entering the outlet. If there's more than a 5mA difference between the two, it trips. Doing that requires electronic circuitry that itself consumes power; coincidentally, this is typically in the range of 5mA-10mA. When you connect the neutral and ground upstream, some of that current returns down the ground wire instead of the neutral wire, and that can easily be enough to trip the GFCI.

  • Thanks for the response. This is exactly what I was hypothesizing...adding the copper ground to the equation essentially changes the grounding path and the the conductivity/resistance. Thanks.
    – Paul
    Dec 3, 2014 at 20:10
  • So out of curiosity, and only hypotehtically, if I tripped the GFCI intentionally, connectedc the white and copper, and then tried to reset the gfci, do you think it would still read it as a change or variance, or might the GFCI read it as a static starting point and not trip. I'm assumning it would still read it as variance between the hot and the ground
    – Paul
    Dec 3, 2014 at 20:13
  • It would still trip. It's just measuring current through the hot and neutral wires; it doesn't require any baseline to compare against.
    – Zhentar
    Dec 4, 2014 at 5:30

Because touching the neutral (white) to the ground (copper) is a ground fault. And that is what the GFCI (Ground Fault circuit interrupt) is designed to catch.

People often notice neutrals and grounds are connected to each other in the service panel (breaker box), and conclude "neutral equals ground". Nope. Stick a hair dryer on the farthest outlet in your house and check voltages with a DVM. (I'm assuming your house is wired correctly.)

I know it's a weird system and non-intuitive to anyone who's worked in other kinds of circuits, but it evolved over 100 years of picking through ashes figuring out why houses burned down, compromised with not using excessive copper or labor.

You are in error about how GFCIs work. Actually, GFCIs have absolutely nothing to do with the ground wire. They do not evaluate ground, GFCIs don't even need ground connected to them. The only reason to bother wiring ground to GFCIs is - in the case of a GFCI+receptacle combo device, to provide equipment safety ground to the receptacles. A GFCI deadfront will work fine without ground, and a GFCI+breaker combo device doesn't even have access to ground. Seriously.

GFCIs work by comparing current flow in hot and neutral, which should be equal. When you short neutral and ground, you are creating two return paths, and the returning current is being split between neutral and ground. So the GFCI sees hot and neutral are not equal, and SNAP. I wouldn't expect it to trip on no load either, but it sounds to me like you put all the lighting/fan/heat loads on the LOAD side of the GFCI without realizing it. There are your surprise loads.

Do not bootleg LOAD-side grounds off LOAD-side neutral: Doing the latter would "fix" your symptom, but would be catastrophic, creating a shock hazard.


I wired a storage barn with all power routed through a GFCI outlet which is the first device after the breaker and outside by my A/C unit for servicemen. With power on, and the UF cable black wire in the barn switched off, when I touch the white wire and the bare copper wire in a fixture box, the GFCI trips.

Go to ROOT CAUSE below to skip the detailed analysis and to find out how this actually happens.

Here is what I have:

Dedicated circuit from the breaker panel to the normally unused A/C GFCI receptacle and then out the load side to the barn.
Load out of GCFI goes to a barn mounted kill switch box as the first device in the barn.
In switch box, UF black feeder wire from the LINE terminal is from the GFCI.
In the switch box, WHITE WIRE from the silver white line from the GFCI and WHITE WIRES in the 2-gang switch box with a second device being a receptacle ares connected together.
In switch kill all power to the barn safety box, BLACK WIRE from Line after the GFCI is connected to 1 Brass terminal on the kill switch. The outer black wire goes to the first receptacle.
In the 2-gang switch box, BLACK WIRE output from the receptacle is connected to the rest of the barn receptacles, switches, and lights.  The kill switch cuts all power in the barn after that switch, ALL.

All fixtures work fine without tripping the GFCI and tested fine with a circuit connection tester. If I hook up a non-GFCI outlet receptacle to a fixture box, and plug in a 3-prong appliance, it works fine. So, I checked to ensure the Line in and the Load out has the polarity on the GFCI correct. All connections are tight.

So, with all that said, with the kill switch off (open black wire), why would the GFCI trip if I touch white to bare copper in any of the fixture boxes?

It seems to me there shouldn’t be ANY current flowing in the white wire, let alone bleeding into a second grounding point in the barn.

So, why is no power being used in the barn then tripping the GFCI to Off? The reason can be seen in this youtube.com video below at 4:00. The coils in the GFCI internally are circulating currents in the white and the bare or third wire green grounding wires in the GFCI. Touching white and bare in the barn causes an imbalance in the GFCI detector circuit. BTW, the voltage drop in the UF cable to the barn from the GFCI is less than 0.5 volts AC under a resistance light bulb load. No CFLs or LEDs were working in this load testing.

The key current path is shown at 4:00 in this video of a GFCI tear down. https://www.youtube.com/watch?v=DyeijLvh-yI

My kill switch is in the black line wire by the word load. There is no load current flowing in any load in my barn.

With no load connected and the kill switch open, no current is flowing from a load back to the GFCI on the neutral or the ground green at the common paths shown in turquoise by the wire resistances, Rn and Rg. Rg would be open until I touch the white and ground wires. Bang. GFCI trips.

There should be no current flowing through either Rn or Rg back to the service panel grounding strip shown in turquoise on the left. The only current flowing would seem to be in the coils in the GFCI that might induce a current in the returning white wire inside the GFCI itself that cause the chip to detect a fault. This is a false ground fault. This is not how the GFCI is tripping, IMO.

Handling any of the wires in a three plug inserted into the turned off receptacle in the kill switch box does not trip the GFCI. Shorting any combination does not trip the GFCI until I short the un-powered white neutral line and the bare (green third wire equivalent) wire together. POW. The GFCI trips with no current flowing in the barn. The voltage drop on the UF feeder wires is way below 2%. Leakage current through the UF wires to some theoretical earth grounding point is not enough to trip the GFCI.

In short, I believe that shorting the load side neutral and the bare ground wire terminals on the GFCI itself with nothing connected to the GFCI receptacle, would trip the GFCI internally. How does that happen? The GFCI is tripping on a false positive. The black load terminal on the GFCI is never touched in this thought problem. None of that stuff in the dotted line box in the video at 4:00 is even hooked up except for the turquoise loop showing the top turquoise white wire path connected to the bottom turquoise wire or ground wire.

Where is the hidden black wire current coming from and going to? It first flows through the solenoid coil winding. Then through the SCR device current, but is unlikely enough current to cause a trip. The other path after the solenoid coil is through the Rline 24K resistor to the chip pin 5 and always shows some current flowing with no load side GFCI devices hooked up. Where is the black wire voltage powering this chip? The transformer coils, I guess. That is another black wire current flowing all the time and not noticed.


The key current path is shown at 4:00 in this video GFCI tear down schematic.


Causing a short like in the turquoise path without a load, grounds out IC chip pin 4. The short directly jumpers the upper white wire connected to the bare ground wire and then to to the bottom chip pin 4. It also shorts anything flowing through the solenoid coil and components going to pins 5, 6, and 7 but with only 2 ohms of resistance. These are the possible root causes of the false triggering on shorting the white and bare wires in my barn.

This is a pretty complicated wiring scheme inside the GFCI that is giving the false tripping. Shorting the neutral and the ground wires causes the chip to default, no pun, to being tripping by the unseen circuitry inside the chip itself. Shorting pin 4 to the connected white and bare grounded wire will cause a trip. What is the resistance of a pot setup as a variable resister between white and bare? Anything below about 25,000 ohms trips my GFCI. You will never find this analysis of the root cause on the internet anywhere. Nothing after or outside the GFCI itself is causing the ghost ground loop fault. It is a simple shorting of pin 4 to ground anywhere that causes the tripping.

The only way to stop my GFCI from tripping is to install a DPDT 220 V AC switch that switches both the black wire on one pole and the white wire on the other pole. You never switch the white or bare ground wires, see the NEC. It is illegal. Don't do it. Trip the GFCI at the GFCI to rewire anything on the circuit down steam on the load side.

So, the root cause of this false tripping is the GFCI chip pin 4 being shorted to ground. All the quotes of erroneous magical, hidden, and mysterious ground loops in the earth, dirt, bare ground wires, or leakage currents are very misleading to me.

The problem is inside the GFCI device PCB itself. This partial manufacturer diagram was good enough to figure out my idea and confirmed the testing of the root cause of fake tripping at pin 4.



University 'electronic technology' degree, summa cum laude.

  • Hello, and welcome to Home Improvement. This is a whole lot of information for something that's already been pretty well-answered. And, you should probably take our tour so you'll know how best to contribute here. Dec 20, 2019 at 21:01
  • A couple of points: 1) switching the neutral in addition to all hot conductors is permitted by the Exception to NEC 404.2(B) (and is required in a few niche applications, even, such as fuel dispenser disconnects, see 514.11(A) for details). 2) more importantly a GFCI tripping on a N-G fault is intended behavior (and even required by the UL943 standard, AIUI), NOT a "false trip". 3) Pin 4 is simply the internal reference for the GFCI IC and thus is directly wired to the neutral as you see in your diagram. The actual mechanism involves the grounded neutral coil seen in that diagram. Dec 20, 2019 at 23:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.