GFCI trip point

[ptonsparky]

180324-0943 EDT

ptonsparky:

My post last night has an error in the way I worded it that I will correct later.

That you had a much lower source voltage in your experiment was greatly to your benefit. But it still is an incorrect way to run the test. You need to do the test approaching the trip point from a lower current. More on this later.

.

ok.

 

[kwired]

Biggest problem I see with your test method is the device may get triggered at one level, but response time until opening of contacts could still let more current through and your meter may catch that current before the switch is opened, leading you to believe the trip threshold is higher then it really is.

As has been mentioned there is a time factor in that trip curve. If you suddenly have a 100mA fault it probably trips pretty fast - still have 100 mA of current though. GFCI doesn't limit current it just responds to it.

 

[gar]

180324-1126 EDT

Correction to wording of my post #16.

The single number trip current rating of a GFCI is based on the lowest current applied over a long time, several seconds, that will not trip the GFCI. There is a lower and upper limit for this spec. The value 6 mA is a sort of nominal value.

This should have read more like:

The single number trip current rating of a GFCI is based on a value that is slightly above the highest current applied over a long time, several seconds, that will not trip the GFCI. There is a lower and upper limit for this spec. The value 6 mA is a sort of nominal value.

Also 6 mA may be 5 mA. I have not checked.

Below some current threshold the GFCI will never trip, but go above this threshold, and after some moderately short time, seconds, the GFCI will trip. If the duration of the unbalanced current is short, 8 mS for example, then a much higher current is required to trip the GFCI. The inverse time characteristic. Fundamentally this is for noise filtering.

.

 

[gar]

180324-1144 EDT

An experiment:

Leviton GFCI receptacle. AC power, nominal 120 V, continuously applied to Line terminals. No load on load terminals, or recreptacle sockets.

A series circuit of a 6 V transformer secondary (transformer provides isolation), a 5 k pot in series with a 220 ohms 1/2 W resistor, and a Fluke 27 in AC mA mode. This series circuit is connected between "line neutral" and "load neutral" of the GFCI.

Power to the 6 V transformer is applied and removed to perform each test. This is not a requirement. My procedure had the transformer power off when the GFCI was reset.

I started each test at abourt 3 mA, not precise, not important, just needs to be somewhere below the trigger point. Note: I had no current at time of reset, but the series circuit is connected to the two neutral terminals.

Test:
Test current zero, but the test current circuit was connected.
GFCI RESET.
Applied test current starting at about 3 mA.
Gradually increased test current to around 4.8 mA where GFCI starts to try to trip, but does not.
At about 4.9 mA the GFCI trips. There seems to be less than about 0.1 mA variation in the trip point on multiple trials.
Remove test current.
RESET GFCI.
Repeat.

.

 

[don_resqcapt19]

not sure where the 4-6 mA comes from
it's a curve
depending on ground path, etc it could be 100 mA
but as long as fast enough no injury

you shorted N-gnd thru a meter set to mA?
what was the voltage?

The 4-6mA comes from the listing standard. The GFCI must not trip below 4mA, is permitted to trip between 4 and 6mA, and must trip above 6mA. All of those are subject the maximum permitted time to trip. UL 943 defines the maximum permitted trip time as "The maximum permitted time to trip in seconds is equal to the quantity (20/fault current in milliamps) raised to the 1.43 power. " I assume that the curve you posted is based on that math.

 

[Ingenieur]

to predict where it will trip you need the fault curve
assume a z and an x/r
from this a tc can be calculated and the i rate of rise
where the fault curve intersects the trip curve is the trip point and likely greater than 5-6 mA

if I recall x/r for a person is capacitive and ~1 and use z = 1000
r = x = 707 Ohm
C = 3.75e-6
tc 2.7e-3 or 0.16 cycle
i = 120/1000 + 120/707 x e^-(tc x t) = 170 x e^-(2.7e-3 x t)
0.1 tc or 0.26 ms i = 270 mA
0.5 tc, 1.3 ms, 220 mA
1 tc, 2.65 ms, 180 mA
2 tc, 5.3 ms, 140 ma
4 tc, 10.6 ms, 123 mA, assume steady state 120/1000

a rough sketch relay vs fault, fault is off page left and returms
in this case should trip ~180 mA in 9 mSec
easily survivable

allowable for 180 mA ~ 0.95 sec (from i mA = 175/sqrt t)
very close to the threshhold curve on the UL graph

 

[al hildenbrand]

Why would I blow a fuse when I am causing an imbalance for the GFCI to detect?
From the grounded conductor to the eguipment ground.
A Basic check when you have GFCI trouble on new installation is to look for that inadvertent load side bond. A 5 ma fault on the grounded conductor should trip the same as the hot side. All the GFCI looks for is an imbalance.

Why do the math when I have meters that tell me the current? I could connect a pot in series with the meter and slowly turn it until the GFCI trips but I don’t think that’s how it happens in real life.

The part that makes me question my test now is the trip level of my basement GFCI.

You are not just creating a path for an imbalance, you are shorting the secondary of a transformer, i.e., the "Grounded Neutral Transformer" inside the GFCI. This is a source of current that is limited only by the impedance of the complete secondary circuit, so it can be a lot more than 10, or even 100, milliamps.

 

[Ingenieur]

made a small error
steady state is v/z = 120/1000 = 120 mA not v/r

trip at 120 mA in 15 mSec
calculated max time allowable for fibrillation threshhold for 120 mA is ~ 2.1 sec
from UL curve ~ 2 sec

trip well below threshhold curve (15 mS vs 2000 mS) and safe
well below the max UL curve for 120 mA, 15 vs 80 mS
but the victim will feel it, clench, muscles contract, but no burns or heart issues
stunned possibly but will recover quickly

 

[ELA]

Scope provides better resolution

Scope provides better resolution

Here are a couple of pictures from testing performed back in 2012 on Leviton 7599.


The top one tests the secondary transformer Neu-PE
The bottom is a line(load side) to neutral - current limited fault.

 

[ELA]

Here are a couple of pictures from testing performed back in 2012 on Leviton 7599.

The top one tests the secondary transformer Neu-PE
The bottom is a line(load side) to (PE) - current limited fault.

Correction only

 

[ptonsparky]

You are not just creating a path for an imbalance, you are shorting the secondary of a transformer, i.e., the "Grounded Neutral Transformer" inside the GFCI. This is a source of current that is limited only by the impedance of the complete secondary circuit, so it can be a lot more than 10, or even 100, milliamps.
View attachment 19987

I am making only the “Improper neutral-to-case connection”. There is no way that will be any where close to 1 amp let alone 15. The only load on the GFCI while testing at the shop was the meter.

The basement GFCI does have additional load and that would explain why my trip current is higher here because I am paralleling the neutral load for a brief time via the meter.

 

[kwired]

I am making only the “Improper neutral-to-case connection”. There is no way that will be any where close to 1 amp let alone 15. The only load on the GFCI while testing at the shop was the meter.

The basement GFCI does have additional load and that would explain why my trip current is higher here because I am paralleling the neutral load for a brief time via the meter.

What he was saying (and I never thought of it earlier) was that there is a coil inside the device that will induce current that will unbalance the current sensing coil and cause trip should the neutral fault to ground - even with no connected load. Without this coil in there the device only trips on a neutral to ground fault if there is current flowing through the fault path, with the coil it forces current to flow and make it trip if there is no load otherwise connected.

If the device didn't have that neutral to ground fault extras - your testing may not have even tripped the unit - depends if there was any load current involved with your fault. With those extras - he is saying it may actually put up to 100 mA of fault current through the device.

The other issue is how long must current flow before your meter will catch it, vs how fast will the GFCI respond to the same current level? If the device is capable of responding faster then your meter you may not have caught how high the true current was.

 

[ptonsparky]

What he was saying (and I never thought of it earlier) was that there is a coil inside the device that will induce current that will unbalance the current sensing coil and cause trip should the neutral fault to ground - even with no connected load. Without this coil in there the device only trips on a neutral to ground fault if there is current flowing through the fault path, with the coil it forces current to flow and make it trip if there is no load otherwise connected.

If the device didn't have that neutral to ground fault extras - your testing may not have even tripped the unit - depends if there was any load current involved with your fault. With those extras - he is saying it may actually put up to 100 mA of fault current through the device.
Ok, I see it now. Thank you.

The other issue is how long must current flow before your meter will catch it, vs how fast will the GFCI respond to the same current level? If the device is capable of responding faster then your meter you may not have caught how high the true current was.

Adding a hair dryer to the load side of the GFCI did not significantly increase the trip current that I was catching at 100msec.

 

[ptonsparky]

Thank you all for your responses.

 

[gar]

180324-2149 EDT

ptonsparky:

As I said before your test method and resulting meter readings are next to useless in terms of measuring the GFCI current detection threshold (the nominal 4 to 6 mA spec).

When you short your very low impedance milliamp meter between the GFCI output (load side) neutral and an EGC wire there is only a low source voltage to drive current thru the meter. This voltage can be highly variable from one test to another, and therefore you can expect different readings. Further the results are likely to be different for different GFCIs.

This test is not a test of the trip current threshold of the GFCI.

.

 

[ptonsparky]

180324-2149 EDT

ptonsparky:

As I said before your test method and resulting meter readings are next to useless in terms of measuring the GFCI current detection threshold (the nominal 4 to 6 mA spec).

When you short your very low impedance milliamp meter between the GFCI output (load side) neutral and an EGC wire there is only a low source voltage to drive current thru the meter. This voltage, can be highly variable from one test to another, and therefore you can expect different readings. Further the results are likely to be different for different GFCIs. This was/is on my list of things to look at today. Check the voltage of the shop GFCI. N-gnd. It is a different brand.

This test is not a test of the trip current threshold of the GFCI.

.

The two GFCI receptacles do consistently trip at the same levels compared to the splitter, be it luck or not.

Thank you all for attempting to educate me, in spit of my kicking and screaming the whole way.

 

[gar]

180325-2331 EDT

ptonsparky:

Consistency of readings in your experiment does not prove that the readings are particularly correlated with the GFCI specifications. Your readings are no where close to the GFCI specification.

You need to study and understand how recent GFCIs work under different fault conditions.

If you have a new GFCI with a current injection transformer, with 120 V line side input, no load, and shorting the GFCI load side neutral to the GFCI line side neutral with any resistance from 0 to 5 ohms, then the GFCI should trip. The duration of the injected current will increase as the resistance is increased, and the magnitude of the current will decrease. At somewhat near the 5 ohm value the current will drop to somewhat above 6 mA.

With about 7.9 ohms of resistance, 1 of which is used for current measurement, a sample Leviton GFCI trips at about 200 mS duration with a scope measured current of about 11 mA, and a Fluke maximum voltage about 7 mV, or 7 mA. There is a fair amount of variation in trip time, and at this duration produces a substantial variation in the Fluke maximum reading.

This is not a good way to measure the GFCI trip current.

.

.