Testing ungrounded receptacles on GFCI circuit

[Steven B]

So you are using an external tester?

Yes, a GFCI plug tester which doesn't work in this situation

 

[iwire]

Sorry, I misunderstood, thought you meant the test button on the plug tester

None of the GFCI device directions will tell you to use an external tester of any sort. One of the reasons for this is that external testers will not work on a two wire circuit.

 

[MD84]

Good idea about using a different test method. The resistance of a light bulb would do the trick. I just was not expecting the GFCI to be slower than the breaker. I am going to try to find the trip times for each.

The GFCI should trip within 25ms on a 500ohm fault.

 

[JFletcher]

I am currently working in an older home with ungrounded 2 wire circuits. I have installed a GFCI receptacle to protect 3 wire ungrounded receptacles downstream. I know that a plug tester with a GFCI test button will not trip the GFCI receptacle but I assumed I could test the downstream receptacles by shorting the hot side of the receptacle to a nearby piece of grounded metal. I thought that it would simply trip the upstream GFCI well before the breaker would trip. To my suprise, I got a nice flash and the breaker tripped. Aside from any criticisms of my test methods, can anyone tell me why the breaker tripped instead of the GFCI? I am pretty sure that everything is connected properly. The test button on the GFCI works and kills power to the downstream receptacles.

Thanks for any help.

To my knowledge, what I bolded is the only proper/approved way to test downstream GFCI protection. As others wrote, shorting to ground probably caused the breaker to see a current that resulted in an insta-trip; it reacted faster than the GFCI.

I wont criticize the method other than to say that breakers dont always trip, and you might be surprised to see what kind of arc can come out of a 12 or 14ga wire.

 

[Steven B]

None of the GFCI device directions will tell you to use an external tester of any sort. One of the reasons for this is that external testers will not work on a two wire circuit.

Right. As discussed in my first post

 

[MD84]

There is no need of that.

The accepted method is to put a load in at the downstream receptacle and push the trip button at the GFCI device.

If this is some hobby type exercise at your house that is great but if you are doing this on the job the tests you are describing will not count for anything if comes down to that.

Nothing you cobble together will be looked at as a reliable test.

The manufacturers testing instructions can fall short IMO. The GFCI can be miswired and still pass the suggested test method. I think it is important to think about what we are doing and if necessary go beyond what is recommended.

I agree with the test method that you have suggested however that may not be the recommended test method per the manufacturers instructions.

Testing GFCI receptacles on the job does not require any special testing equipment. In the case of a protected downstream receptacle on a grounded receptacle I will use a GFCI tester. In the case of an ungrounded receptacle I will use the test button and verify loss of power at downstream receptacles.

I understand that the tester I described is not a professional tool. Please don't imply that I am a cobbler. There are cases in which I have fabricated lab quality test instruments where required by my customer.

 

[iwire]

The manufacturers testing instructions can fall short IMO. The GFCI can be miswired and still pass the suggested test method. I think it is important to think about what we are doing and if necessary go beyond what is recommended.

If I press the test button on the GFCI and the load shuts down what miswire are you thinking of?

I agree with the test method that you have suggested however that may not be the recommended test method per the manufacturers instructions.

The method I described is recommend by the manufacturers and the CSPC

Testing GFCI receptacles on the job does not require any special testing equipment.

No but if you are doing the testing and keeping a log as OSHA expects you had better be using a method instructed by the manufacturer.

I understand that the tester I described is not a professional tool. Please don't imply that I am a cobbler. There are cases in which I have fabricated lab quality test instruments where required by my customer.

You may be Tesla himself but I would still use the method that is found in the device instructions.

(It gets more interesting if you look at AFCI 'testers' UL only lists those as AFCI 'indicators' as they are not really capable of testing an AFCI yet they are sold as testers)

 

[don_resqcapt19]

The GFCI should be faster, yes. You never gave the GFCI a chance to see any imbalance between the hot and neutral going to ground. You left the neutral out of the circuit.

I don't think the GFCI is faster. The trip time for a GFCI is often stated at 1/4 second. That is 15 cycles and a breaker can clear a short much faster than that.

 

[don_resqcapt19]

The GFCI should trip within 25ms on a 500ohm fault.

The UL standard would permit the trip time to be 0.78 seconds for that load.

 

[Steven B]

Thanks all for your input. I will post back if I find any conclusive info on GFCI reaction time vs. breaker. For now I will have to be satisfied with the test button on the upstream GFCI, although I may do some experimenting with less "flashy" methods of testing the downstream recepts.

 

[jim dungar]

Thanks all for your input. I will post back if I find any conclusive info on GFCI reaction time vs. breaker.

The formula for a UL Listed Class A GFCI is:

T=seconds, I=mA of current.
I believe 300mA is the upper limit for the formula, larger currents will not produce faster times.

 

[ActionDave]

Many people are amazed at how fast circuit breakers actually are, when tripping in their instantaneous region, like with a direct short.
Likewise, many are surprised that GFCI's actually have a time delay in their operation.

I don't think the GFCI is faster. The trip time for a GFCI is often stated at 1/4 second. That is 15 cycles and a breaker can clear a short much faster than that.

Thank you to you both for the clarification. I think it is safe to say that if you want to test GFCI circuitry a dead short to ground is not the way to do it.

 

[iwire]

Thank you to you both for the clarification. I think it is safe to say that if you want to test GFCI circuitry a dead short to ground is not the way to do it.

Yeah, when I was getting started I figured it would not be a spark over 7ma if shorted. My screwdriver suffered the consequences and I learned things do not open before the sparks fly. :dunce:

 

[don_resqcapt19]

Yeah, when I was getting started I figured it would not be a spark over 7ma if shorted. My screwdriver suffered the consequences and I learned things do not open before the sparks fly. :dunce:

I did that in the past also....while the GCFI will limit the duration of the fault, it does not limit the current that flows into the fault.

 

[wwhitney]

The formula for a UL Listed Class A GFCI is:

T=seconds, I=mA of current.
I believe 300mA is the upper limit for the formula, larger currents will not produce faster times.

That formula implies the GFCI is supposed to open in under 19.7 ms when the imbalance is 300 mA or more (e.g. the OP's dead short). Apparently the instantaneous trip on a circuit breaker can be faster than that?

Cheers, Wayne

 

[don_resqcapt19]

That formula implies the GFCI is supposed to open in under 19.7 ms when the imbalance is 300 mA or more (e.g. the OP's dead short). Apparently the instantaneous trip on a circuit breaker can be faster than that?

Cheers, Wayne

One time trip curve that I looked at shows the instantaneous trip time to be between 1/2 and 1 cycle, so less than 16.6 ms.

 

[MD84]

If I press the test button on the GFCI and the load shuts down what miswire are you thinking of?

The method I described is recommend by the manufacturers and the CSPC

The method you are describing should catch a miswire. This is not however the test that is recommended by Leviton for example. See here. It is possible to miswire and the prescribed test would not catch it. I should not test any further however because of 110.3(B)

The UL standard would permit the trip time to be 0.78 seconds for that load.

That seems a lot longer then I would expect. How is that derived?

I did that in the past also....while the GCFI will limit the duration of the fault, it does not limit the current that flows into the fault.

I think it is a common misconception that a human will only be exposed to 6mA since that is the trip threshold of a GFCI. In fact it could be much more for a finite amount of time.

 

[iwire]

The method you are describing should catch a miswire. This is not however the test that is recommended by Leviton for example. See here. It is possible to miswire and the prescribed test would not catch it. I should not test any further however because of 110.3(B)

I am not seeing the diffreance in what Leviton recomends

(c) If you installed your GFCI using step 7B press the TEST button, then plug
a lamp or radio into surrounding receptacles to see which one(s), in
addition to the GFCI, lost power when you pressed the TEST button. DO
NOT plug life saving devices into any of the receptacles that lost power.
Place a "GFCI PROTECTED OUTLET" sticker on every receptacle that
lost power, then press the RESET button to reset the GFCI.

You can test 'more' but only the manufacturers recommend test will count for anything.


(Still want to know what miswire this will not catch that your method will)

 

[jim dungar]

One time trip curve that I looked at shows the instantaneous trip time to be between 1/2 and 1 cycle, so less than 16.6 ms.

Yes,
Trip curves only go to 10ms (.o1sec), many molded case breakers begin to operate faster than this (1/2 cycle is not uncommon for small breakers) which can cause the fault current to decrease. This is one reason that selective coordination and series combination rating are usually based on actual testing and series ratings.

 

[roger]

I thought it was >6mA? What you are saying is not incorrect since 7>6. I thought anything over 6mA and the GFCI should trip.

See this UL article

Excerpt

"It states that a Class A GFCI trips when the current to ground has a value in the range of
4 milliamps to 6 milliamps, and references UL
943, the Standard for Safety for Ground-Fault
Circuit-Interrupters."

Roger