How well will a GFCI circuit be able to detect current flows, if both the "hot" input and the (supposedly) "neutral" input to a GFCI outlet are energized?

The reason I am interested in this is that I am renting an old (1915) townhome and have discovered that an (ungrounded) wall outlet and a ceiling fixture are both controlled by a wall switch that opens the netural wire instead of the hot wire. The result is that when switch is off, the direct neutral line is open, but the netrual input of the GFCI is connected to hot. When I reported this to the landlord, their electrical contractor put in a GFCI, but they do not want to correct the open neutral. (I think it may be a knob and tube circuit with only the neutral line going through the wall switch, so fixing it may require pulling apart the wall and/or rerunning the entire circuit.)

For what it is worth, when the switch is off (and both inputs are energized), the GFCI test button will trigger a disconnect, but the reset button is not functional. But I'm not sure if the test button guarantees that the GFCI is providing effective protection in this scenario, hence my question: will a GFCI circuit work when both inputs are energized and it has no real neutral and no ground?

  • It isn't real clear from your question: Is the outlet (which is where the GFCI is presumably installed) switched or unswitched?
    – Dave Tweed
    Aug 16, 2013 at 20:42
  • Both the outlet and the light fixture are downstream from and get turned "off" by the switch opening neutral.
    – Jeff
    Aug 16, 2013 at 21:30
  • And yes, the GFCI is integrated with the outlet receptacle.
    – Jeff
    Aug 16, 2013 at 21:53
  • If the junction box housing the switch receives both the hot and neutral, from upstream, it should be easy to fix. It should; I think it is against code (USA, Canada) for the two conductors of the same circuit to go through different routes in the walls. E.g. a just a neutral going to a switch and then toward the appliances, with the hot taking some other path that is nowhere near the switch.
    – Kaz
    Aug 16, 2013 at 22:10
  • @Kaz: I'm pretty certain the US codes explicitly allow for a hot/neutral pair to go to a light fixture, and for unswitched and switched hot to proceed from there to a junction box without a neutral tagging along. It's interesting to note that in that case, even though both wires going to the switch are on the hot side, the current through them should be equal and opposite.
    – supercat
    Aug 16, 2013 at 22:17

2 Answers 2


A GFCI is only measuring currents and does not care about the voltages on either of the two wires going through it.

As long as the net current is always zero, it really doesn't matter how it is switched on and off.

The real problem with a "switched neutral" is that the contacts inside the fixture are live with respect to ground even when the switch is off. If you should accidentally touch the contacts while grounded, the GFCI will protect you by tripping. But if the fixture were wired correctly, that particular risk wouldn't exist at all.

  • Thanks for the warning. I actually don't think the GFCI was wired to put the light fixture downstream, so I think it is still unprotected.
    – Jeff
    Aug 16, 2013 at 21:51
  • For a GFCI to provide protection, the line neutral wire must be connected any time the line hot side is connected. Disconnecting just the neutral side is very dangerous, since even in the presence of lethal fault currents the GFCI would have insufficient power to operate the solenoid to disconnect the mains.
    – supercat
    Aug 16, 2013 at 22:55
  • 1
    I just tried the test button again with the switch off (i.e., neutral open) and the test button does indeed trip the GFCI.
    – Jeff
    Aug 17, 2013 at 2:15
  • 1
    I've left this answer as accepted for now, because I think it is the best answer to the narrow question as to whether the GFCI can detect faults when both the hot and neutral line terminals are energized. But I do still have concerns about whether it is able to reliably disconnect the circuit when a fault occurs. The fact that the test button works suggests to me that it probably does, but I'd feel better if I understood how it was working. For now I've been just been using the test functionality to deactivate the receptacle.
    – Jeff
    Aug 23, 2013 at 19:41
  • 1
    BTW, I contacted Leviton, and their engineer confirmed that their GFCI's will provide protection in the my situation.
    – Jeff
    Nov 19, 2013 at 15:33

The situation you describe is dangerous and should be totally unacceptable.

A GFCI whose hot can be connected to mains hot without the neutral connection being connected to mains neutral cannot be relied upon to provide any protection. Suppose that a switched-off light bulb represents a 10-ohm resistance between the GFCI's load hot and load neutral terminals. Suppose further that a 450-ohm path develops between the base of that bulb and ground while the GFCI neutral is switched off. What's going to happen?

The GFCI hot terminals are going to be at mains potential, and the neutral terminals are going to be at about 98% of mains potential. Even with 250mA flowing through the fault path, the potential difference between the GFCI hot and neutral terminals will only be about 2.5 volts. Consequently, it's entirely possible that the GFCI won't trip even with a fault current that's well above a level generally regarded as lethal.

  • In the scenario you are describing, you say that the GFCI would not detect the current flowing out, because the currents on the load hot and load neutral terminals are similar, but Dave Tweed suggests that a GFCI detects "net current" flows. In your scenario, both flows go out from the GFCI through the bulb to ground. Can a GFCI not detect this?
    – Jeff
    Aug 17, 2013 at 2:39
  • GFIs trigger on any current imbalance over 6 mA, often 4 to 5. (0.004 Amps) See 1 and 2
    – wallyk
    Aug 19, 2013 at 14:52
  • @jeff: A typical GFCI includes a relatively sensitive sensing circuit which, when a current imbalances is detected, feeds mains power into a solenoid which then opens the contact. In the scenario I described, the sensing circuit would have no trouble "detecting" 250mA, but would be unable to open the contact in response.
    – supercat
    Aug 19, 2013 at 15:11
  • @supercat I see what you are saying. I had thought you were saying that the 2.5V difference wouldn't be enough for the detector circuit, but you are saying the 2.5V difference wouldn't allow the GFCI to draw enough power from the mains to physically open the contact. I share that concern, but I'm having trouble imaging how the test button is working unless the GFCI device is designed to store sufficient potential energy operate the disconnect.
    – Jeff
    Aug 23, 2013 at 19:48

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.