I have a 48vdc to 120vac inverter with a 3 wire output (hot, neutral and ground). I installed a GFCI outlet using those output wires. I tried putting a resistor (I tested with 1k 100 and 22 ohm values) between the hot and ground but the outlet doesn't trip. Using the same resistor on another GCFI outlet connected to the grid, will trip the outlet. When I press the test button on the GFCI outlet connected to the inverter, the outlet will trip. The inverter ground is grounded. I get 120v from neutral to hot and 60v from neutral or hot to ground. It seems like the GCFI won't work if you were to make a circuit to ground. Any ideas of what is happening?

  • \$\begingroup\$ Can you test this particular GFCI on the grid? \$\endgroup\$ Commented May 20, 2020 at 15:33
  • 1
    \$\begingroup\$ @relayman357 Doesn't need to. The provided information adequately explains why the GFCI is not tripping. It's not an (isolated and bonded) power source of the type GFCI TEST circuits are designed to work in. \$\endgroup\$ Commented May 21, 2020 at 18:21
  • \$\begingroup\$ it was a brand new Eaton outlet \$\endgroup\$ Commented May 22, 2020 at 16:18

5 Answers 5


I get 120v from neutral to hot to neutral and 60v from neutral or hot to ground.

The important thing to note here is that you \$\color{red}{\text{get 60v from neutral or hot to ground}}\$

This tells me that the output is likely floating BUT loosely coupled to ground with EMI capacitors of a few 100 nF. Something like this: -

enter image description here

This means that a regular GFCI won't work because you are just not able to drag the current through the resistor. It's a bit like connecting a battery to ground via a 1 kohm resistor - no current will flow because you haven't made a DC circuit.

Your inverter's output is galvanically isolated from ground hence a single point connection from either hot or neutral to ground won't cause a significant current draw that would trip the GFCI.

  • \$\begingroup\$ Hacktastical said the neutral should be tied to ground before the gfci. Will that allow the gfci to function properly? \$\endgroup\$ Commented May 20, 2020 at 17:51
  • \$\begingroup\$ @DanielCaoili You have to do two things, one, obey what recommendations are given by the inverter manufacturer and two, obey what local regulations might apply to how inverter outputs can be configured. In principle yes, but, do some research. \$\endgroup\$
    – Andy aka
    Commented May 20, 2020 at 19:01
  • \$\begingroup\$ its a Chinese made "Reliable Inverter". I tried contacting them about this a month ago but got no response so far. The manual didn't really say anything about the ground situation. \$\endgroup\$ Commented May 20, 2020 at 19:48
  • 1
    \$\begingroup\$ Well, you get what you pay for, or a great deal less when dealing with China, now that they've learned they can jack prices a lot and we'll still pay em. I think the Honda issue is your likeliest bet. (it's reasonably safe; the UK does a similar thing on worksite power, requiring 110V center-earthed). If this is a worksite where safety matters, you really need to focus on grounding the ground, not the neutral. \$\endgroup\$ Commented May 20, 2020 at 20:48

Some (many?) small inverters (and Honda's "inverter" type generators) have a hot neutral.

During one half-cycle, the Hot wire will go from zero to +170 V and back to zero. On the other half-cycle, the Neutral wire will go from zero to +170 V and back to zero. This arrangement avoids the need to make a negative high voltage supply, and normal loads can't tell the difference between this and normal commercial power.

Grounding the Neutral in this case will cause Bad Things to happen!

I don't know how a GFCI would react when faced with this sort of supply.

  • \$\begingroup\$ bad things like a short circuit? \$\endgroup\$ Commented May 20, 2020 at 19:54
  • \$\begingroup\$ Yes - if you connect the "Neutral" to the inverter's Ground, you will have a short circuit. \$\endgroup\$ Commented May 20, 2020 at 20:27
  • \$\begingroup\$ I guess those manufactures don't feel the need to add gfci protection, huh? It seems like it should be required on those because people usually use generators outside. \$\endgroup\$ Commented May 22, 2020 at 16:25
  • \$\begingroup\$ "During one half-cycle, the Hot wire will go from zero to +170 V and back to zero. On the other half-cycle, the Neutral wire will go from zero to +170 V and back to zero" It seems that this is not the case though since i read 60vac from hot to ground one the inverter \$\endgroup\$ Commented May 22, 2020 at 21:49
  • \$\begingroup\$ @DanielCaoili: your meter will read the RMS voltage, not the peak voltage. Since the Hot and Neutral wires each carry alternate half-cycles of the AC waveform, I'd expect your meter to read about half of the full RMS voltage when measuring from either wire to inverter ground. \$\endgroup\$ Commented May 22, 2020 at 21:59

GFCI trips are only relevant when your supply has normal Neutral-Ground bonding back at the service.

So the normal way we do things, is, at precisely one point: the service point, there is an equipotential bond between neutral from supply/onward, and ground from the Grounding Electrode System/onward. The whole point of that exercise is to

  • a) assure that, say, a transformer leak doesn't cause hots+neutral to float at some wild voltage thousands of volts above ground (which would overwhelm insulation in most appliances). So it pegs neutral to near earth. And

  • b) to assure that fault current that winds up on ground efficiently finds its way back to neutral, with enough current flow to get an overcurrent breaker trip. (of course GFCI makes this second role irrelevant, since it'll trip at 5ma).

That bond is called the Neutral-Ground Equipotential Bond due to that first function.

Your thing is not doing it that way, so I don't know what to tell you. Normally in a service installation, only hots+neutral are sourced from the power company. You bring your own ground. And you bond to neutral yourself.

So by that definition, your inverter is the supply from utility. You take only hot and neutral from it. You then locally derive safety ground from your ground rods or whatever. Then you bond neutral and ground at your service point.

My concern is, this could result in the chassis of the inverter (and the battery system powering it) to now become a bouncing ball, wildly floating as high as 170V+48V above earth. That means all of it would become a shock hazard.

Harper's law is that non-UL-listed Chinese cr*p does not ever touch AC mains. If I had to make it safe, I would have the inverter output into an isolation transformer, and put both inside a grounded metal box. Then I'd wire the service as above on the secondary side. 240V/480V-120/240V supply transformers are commonplace, and you could run one at half voltage (and half power) easily enough. Also if you ever upgrade your system, you could go 240V with the bigger system, and re-jumper the secondary for split-phase.

At that point, you could feed into a normal service panel like an Eaton CH, do N-G bonding in a normal way, and have circuits come off in a normal way, and the GFCI should work ducky-doo.

  • \$\begingroup\$ How would the chassis of the inverter and everything else become a "bouncing ball" of high voltage. It seems like it'd be no different than a house where a refridgerator chassis is connected to neutral and ground \$\endgroup\$ Commented May 21, 2020 at 16:33
  • 1
    \$\begingroup\$ @DanielCaoili A refrigerator chassis is not connected to neutral. As for "how" see what Peter Bennett is saying. \$\endgroup\$ Commented May 21, 2020 at 18:19
  • \$\begingroup\$ if neutral is connected to ground at the breaker box and the chassis is connected to ground then it is also connected to neutral, no? \$\endgroup\$ Commented May 22, 2020 at 16:20
  • \$\begingroup\$ @DanielCaoili No, that's a false equivalency. But a very common belief, and one that is hard to overcome. The two being connected at the panel is not the same as the two being connected here. That only applies to a bona-fide electrical service... when we say "bouncing ball" etc. we are talking about your inverter, which is not a bona-fide service and would probably melt down if hooked up like one. \$\endgroup\$ Commented May 22, 2020 at 16:28
  • \$\begingroup\$ if the chassis is not connected to a gfci i fail to see how there's a difference between ground and neutral. \$\endgroup\$ Commented May 22, 2020 at 19:00

GFCIs work by sensing the imbalance between line and neutral. If the supply is floating (like it would be with an inverter), making a path to earth wont change the line/neutral currents with respect to each other, it will only shift the floating voltage around.

Tie inverter neutral to safety ground, before the GFCI, and test again. It should trip.

  • \$\begingroup\$ Is safety ground different from ground? \$\endgroup\$ Commented May 20, 2020 at 17:48
  • \$\begingroup\$ Same thing. Also called ‘earthing’ ground. \$\endgroup\$ Commented May 20, 2020 at 17:59
  • \$\begingroup\$ With a normal electrical system, neutral and earth/safety ground are tied at the panel. Your system isn’t like that: neutral is floating. That’s why the GFCI doesn’t trip. \$\endgroup\$ Commented May 20, 2020 at 18:01
  • 1
    \$\begingroup\$ @DanielCaoili Yes. Ac mains Equipment Safety Ground (aka EGC) is in contact with the earth and does not carry current by design; current ONLY flows during fault conditions. Electronics "ground" is the common that current returns to and in AC power that thing is called neutral. Completely different animals; n'ere the twain shall meet (except during fault conditions of course). \$\endgroup\$ Commented May 20, 2020 at 20:43

So I got a response from the manufacurer after sending this picture enter image description here

The response:


we do not recommend to connect the yellow wire with black one, it will make the inverter shell has the electricity.

best regards"


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.