Successfully fooled a Leviton GFCI into letting 100mA RMS to ground indefinitely without tripping

[Flicker Index]

The UL standard that covers GFCI's (UL 943) requires that they trip off within 1 second on a 6 mA (milliamp) fault
Is that no longer applicable if the leakage current is non-sinusoidal?

Using a classic Apple 1A/5W charger, I've succeeded fooled a newer style (the kind that won't reset without power on) Leviton GFCI into admitting nearly 100mA rms leakage into ground indefinitely without tripping.
Measured value was 92mA RMS (332mA peak, CCF of 3.6)
This GFCI is great at not nuisance tripping, but it allows certain waveform to leak through nearing 100mA RMS.

To replicate, connect a 5v 1A Apple or similar phone charger like the one shown below across line and ground into a Leviton GFCI outlet. Do it with the adapter unloaded so that its internal capacitor charge up. Sometimes it will charge up and hold. Sometimes the GFCI will trip during the charge up. It may take 2-3 tries.

Once the GFCI holds, apply load to 5v side.

I can load the 5v side pretty close to 1A and Leviton brand GFCI holds indefinitely without tripping. Let's just say not tripping at 0.5A on 5v side.
5V 0.5A is 2.5W, so no matter how you skin it, the input current can not be less than 20.83mArms at 120.0vac at 1.000 PF. That's more than 3 times the UL allowed current.

The implication here is that Leviton GFCI may not respond to a leakage close to 100mA if through a power supply. For example, if you were to touch a lifted neutral of a switch mode power supply load that is connected to a Leviton GFCI.

 

[winnie]

Very interesting find.

I recall that one of the approaches to dealing with nuisance tripping due to high frequency leakage to ground (eg. caused by VFDs) is to make GFCIs less sensitive to higher frequency residual current. This is justified by the claim that high frequency currents are less dangerous.

I don't have a good reference for the above; I quick search pulls up this graph of 'let go current threshold' vs frequency, with percentile curves to give an idea of population variability. https://www.researchgate.net/figure...-variability-of-let-go-current_fig2_308409107

I wonder if someone took this idea too far, and the GFCI is simply not sensitive to high current at high frequency.

I also wonder if the impact on the human body of this particular waveform is actually reduced because of its frequency spectrum. You didn't give a waveform trace, but I am guessing that it is narrow spikes aligned with the peaks of the 60 Hz AC waveform, with lots of high frequency energy. This might be a waveform where the high frequency components make people think the current is safe, but where the necessary 60Hz component makes it actually dangerous.

 

[don_resqcapt19]

The UL standard that covers GFCI's (UL 943) requires that they trip off within 1 second on a 6 mA (milliamp) fault
Is that no longer applicable if the leakage current is non-sinusoidal?

...

That is not correct. The maximum permitted time to trip for a 6 mA ground fault is 5.59 seconds.
The maximum permitted trip time per UL 943 is the quantity 20 divided by the fault current in mA and raised to the 1.43 power.

All trip testing is done with a simple resistor circuit.

 

[Flicker Index]

That is not correct. The maximum permitted time to trip for a 6 mA ground fault is 5.59 seconds.
The maximum permitted trip time per UL 943 is the quantity 20 divided by the fault current in mA and raised to the 1.43 power.

All trip testing is done with a simple resistor circuit.

I can not edit the post, so I will post that I stand corrected. I got the result from Google AI, which based the incorrect information on https://nationshomeinspections.com/ground-fault-circuit-interrupter-gfci-explained-2

From a more reputable source, the trip time in second should be (20/x) ^ 1.43; where x is fault current in mA. Source: https://www.se.com/us/en/faqs/FA105578/

My test was done with hot directly connected to charger, and ground directly connected to charger and demonstrated that I can load the 5v side to nearly 1A before tripping the Leviton GFCI.

I've duplicated this with several other 5v 1A chargers. Some allowed for the same, some tripped it.

They're all rated 100-240v AC (and probably intended to operate from 90 to 264v) In the situation of where neutral side of cord is broken and fault current flowing through the power adapter and through damaged neutral through a person isn't likely to sustain 90v input on the charger for it to continue power conversion process however if the adapter was designed with an input voltage range of 50-120v AC instead of 100-240v, it's much more likely that it will just chugg along, but such a device would realistically not exist.

Much more likely with a 230v supply in EU since it's more plausible for 230v supply to be able to maintain 90v or more volts at the input through a resistive fault to ground on neutral side.

 

[tortuga]

Interesting can try the test again off a GFCI breaker?

 

[ELA]

Hello Flicker Index,
Your test findings were interesting so I attempted to duplicate your results but did not have the same experience.

Testing as per your setup with the neutral load current going direct to PE(GND) I tested 4 different chargers. All but the Apple(10W) charger would trip the Leviton 7599 immediately every time they were plugged in, with no load on the charger. As per your findings the Apple would trip once and then hold when plugged in the second time ( its cap now being charged). The load was then applied once powered up.

I did not have an Apple 5W charger but tested using the Apple 10W with a 5W load (1amp).
At 1 Amp the 7599 would trip everytime without fail.
For interest here is the current waveform:




I performed the same test using a Leviton GFNT2 and also a portable TRC -GFCI with very similar results. All three GFCI's would trip every time the 1A load was applied to the Charger.
I then reduced the load on the charger until the GFCI would hold and I had to take the load all the way down in the range of 40 -60 ohms before they would not trip. The GFNT2 would hold a little earlier at 40 ohms and the 7599 would not hold until load was increased to 60 ohms.


Those load values correlate with approx. 8 -10ma of line current so even though the current waveform contains higher freq. content the GFCI's still performed quite well. Not clear why your GFCI performed so poorly?

 

[Flicker Index]

Hello Flicker Index,
Your test findings were interesting so I attempted to duplicate your results but did not have the same experience.

Testing as per your setup with the neutral load current going direct to PE(GND) I tested 4 different chargers. All but the Apple(10W) charger would trip the Leviton 7599 immediately every time they were plugged in, with no load on the charger. As per your findings the Apple would trip once and then hold when plugged in the second time ( its cap now being charged). The load was then applied once powered up.

I did not have an Apple 5W charger but tested using the Apple 10W with a 5W load (1amp).
At 1 Amp the 7599 would trip everytime without fail.
For interest here is the current waveform:

View attachment 2576442


I performed the same test using a Leviton GFNT2 and also a portable TRC -GFCI with very similar results. All three GFCI's would trip every time the 1A load was applied to the Charger.
I then reduced the load on the charger until the GFCI would hold and I had to take the load all the way down in the range of 40 -60 ohms before they would not trip. The GFNT2 would hold a little earlier at 40 ohms and the 7599 would not hold until load was increased to 60 ohms.


Those load values correlate with approx. 8 -10ma of line current so even though the current waveform contains higher freq. content the GFCI's still performed quite well. Not clear why your GFCI performed so poorly?

1A is right at the threshold where tripping happens. With my experience, it happens around 0.9A.
Try with a load between 0.5-0.75A. You also said you used an Apple 10W charger. I can comment that there were some chargers that tripped with the same load and some that did not. Have you tried with other chargers?

You may need to pre-charge the charger's internal bulk storage first by powering with no load, then re-connecting to L-G before charge is lost.

Anker 18W QC charger was an example of a charger that would trip even with the load that did not trip on Apple 5W charger.

I will repeat the test with another Leviton GFCI when I get a chance.

 

[Flicker Index]

@ELA

So, I repeated the test with the same product line of Leviton GFCI and the result was similar to yours. I was able to get it to hold with the USB charger energized, but could not, on the 5v end, anything close to 1A of load before tripping it.

The production code paper label on the top end of the Leviton GFCI is rather cryptic and I couldn't figure out what it means, but the one I was talking about in the original post is non-tamper resistant SmartLockPro, and the one I have had similar result as your test is tamper resistant type.

Not that shutters in TR makes any difference, but since TR was mandated in code for residential, it implicitly suggests the TR is likely to be of newer Rev.

Hardware wise, GFCI has detection coils, circuitry, and a solenoid. The interpretation of the detection coil signal falls on the GFCI controller chip. So maybe Leviton/chip manufacturer was already aware of this and revised somewhere along the line.