Beware of your differential circuit breakers (GFCI)!

[Calambres]

This is a topic to warn you about failing differential circuit breakers in your homely electric installation.

We all (must) know that the test button in those switches is there to perform a monthly test to see if the circuit breaker is working. My word of advice is to distrust this procedure and go a bit further based upon my recent experience. I, like most of us, refrain to perform the monthly test and only do it once in a blue moon or simply when it comes to mind. The other day I was lurking youtube and found one talking about a simple differential switch tester (a Triac, a Diac, 4 resistors, 3 capacitors, a potentiometer, a SDPT switch and a 50mA AC analog panel)  with which you can test not only if it works but "when" it works, I mean, how many mA are needed to trip the typical 30mA differential switch we all have at home for our security.

The fact was that I had a couple of "Very Expensive Super Immunized 30mA Differential Switches With Automatic Reconnection" at home from a "Very Reputable Brand", both installed on independent branches of the electrical grid at home and with no more than 4 years of use. Both of them pass the test button test OK but one didn't work at all as a differential switch    The one in my house tripped the switch at 25mA (very common for a 30mA breaker) and auto reconnected like a charm but the other in my "den"  didn't trip at all, not even at 50mA... although, as I said before, it passed the button test perfectly . This is the one responsible of protecting my work bench where I do all things electric and electronic, my garage and my mechanical shop 

My word of warning is to be aware that the "monthly test button test" may not be definitive and your safety may be in danger: Place a load of between 15mA to 30mA between the live and earth and see if the differential switch go down.

Sorry if this is not the correct place to put this thread. Feel free to move it to a better place if necessary.

[EDIT] I'm talking about a GFCI Breaker or Ground Fault Circuit Interrupter.

[TK]

I guess you mean A and not mA

[Dubbie]

No, this is not a over current circuit breaker, it is a ground leakage detector.

Basically it checks to see that the same current is flowing “out” through the phase as is coming “back” through the neutral. If there is a discrepancy, it means there is some fault that is letting current leak to ground. Maybe a finger in a light socket somewhere, or a drywall screw through some tps cable? There is a threshold of a few tens of mA so that ordinary inductive and insulative leakage doesn’t trip it constantly for no real fault.

Of course if your finger forms part of a circuit between phase and neutral, the power will still return through neutral and the thing won’t trip.

I suspect TK knows this already, so is more for benefit of others who may not.

[Monkeh]

I guess you mean A and not mA

No. No, he really doesn't.

[CatalinaWOW]

I guess you mean A and not mA

Very possible that you needed some coffee before making this comment.  If something like that was not the problem I would suggest staying far from hardware, and be cautious what you suggest others do in your systems analysis role.

I will also give you an out on language problems.  Here in the US these breakers are called GFI or GFCI breakers.  This forum is often widely separated by a common language.

[tautech]

If your den is on a submain distribution board your mains ground return path could be suspect.
Try checking den mains ground to a local earth and see if there's a voltage differential. If any it should be very low.

[TK]

(very common for a 30mA breaker)

What about here?  Is it 30A breaker (GFI) or 30mA?

I must admit it was not lack of coffee.  I know my limitations and I would never give any recommendation regarding anything related to mains or electric wiring or safety.  I still consider to be a total ignorant on this field... I was one of the ignorants that connected the ground clip of a Rigol scope to the 110V section of a power supply (Live / Neutral) and experienced a huge spark, the scope survived but took years to understand why it went wrong.

I know what a GFI is, but did not know how it operates... I learned something new today

[TK]

I suspect TK knows this already, so is more for benefit of others who may not.

I didn't know

[woody]

I always wonder if there are many people that test these things once a month.
Nowadays that means first shutting down pc's, a couple of NASes, the heating, the alarm system, a mail server, the router, and then run a real risk that one of the many, many connected SMTPs does not come back up after the test.
Who makes time for that 12 times a year?

[james_s]

I probably test my GFCI breakers once a year if that. Never in my life have I encountered one that failed to trip. I did have to replace one at my mom's place recently that would trip immediately with nothing plugged in. Replacing the GFCI cured the problem.

[Monkeh]

I probably test my GFCI breakers once a year if that. Never in my life have I encountered one that failed to trip. I did have to replace one at my mom's place recently that would trip immediately with nothing plugged in. Replacing the GFCI cured the problem.

You've never encountered one which fails to trip at the massively excessive current applied by the test button. How many times (if ever, since leaving the factory..) have any of those breakers actually had their trip characteristics verified?

[Gyro]

I probably test my GFCI breakers once a year if that. Never in my life have I encountered one that failed to trip. I did have to replace one at my mom's place recently that would trip immediately with nothing plugged in. Replacing the GFCI cured the problem.

I have. Admittedly it was a fairly old one but having experienced it, I now have that slightly uncomfortable feeling having just tested it, will it fail to trip next time?  It can lead to compulsive button pressing syndrome.

[Dwaine]

Instead of going with GFCI breakers in my home.  I just went to better plugs.  Some nice Hospital Grade plugs with filtering and Audio Alarm.

[Monkeh]

Instead of going with GFCI breakers in my home.  I just went to better plugs.  Some nice Hospital Grade plugs with filtering and Audio Alarm.

And how is a 'plug' (you'll find that's a socket) a replacement for a protective circuit?

[steverino]

I think the newer GFCI devices (outlets and breakers) are required to self test, at least in North America (UL943 2015).  Here's a common chip used in newer leviton gfci outlets (at least i think this is the chip used, the markings on the chip were removed).

http://www.onsemi.com/PowerSolutions/product.do?id=FAN4149

[Calambres]

If your den is on a submain distribution board your mains ground return path could be suspect.

That's the first thing I thought but it is not so. I replaced the failing switch with a new one and this works perfectly and trips at 22mA. Besides, I checked the ground path as well as the isolation from live to earth with a megger. All was OK.

Regarding the discussion about A vs. mA I think the problem is in the nomenclature. I'm not used to your namings because english is not my mother language. Of course I meant CFGI breakers. It was a 40A and 30mA sensitivity CFGI.

And yes, I'm afraid that the test button is not enough indication that the thing is working properly and hence my warning.

[tautech]

If your den is on a submain distribution board your mains ground return path could be suspect.

That's the first thing I thought but it is not so. I replaced the failing switch with a new one and this works perfectly and trips at 22mA. Besides, I checked the ground path as well as the isolation from live to earth with a megger. All was OK.

Great, then it was a dud breaker.
Was it checked as OK at initial installation ?

[Calambres]

Of course if your finger forms part of a circuit between phase and neutral, the power will still return through neutral and the thing won’t trip.

AFAIK, in Europe the neutral is ultimately connected to ground so in the case you refer, the CFGI breaker should trip. In Spain where I live it is imperative by law to have at least one of these with a sensitivity of at least 30mA in homes and small business and 300mA in industries.

[Calambres]

Was it checked as OK at initial installation ?

Yes, it was, and worked OK. Somehow it broke.

[tautech]

Was it checked as OK at initial installation ?

Yes, it was, and worked OK. Somehow it broke.

Hmmm.
Then we best take your advice and check them.
Thanks for bringing this to our attention.

[james_s]

I probably test my GFCI breakers once a year if that. Never in my life have I encountered one that failed to trip. I did have to replace one at my mom's place recently that would trip immediately with nothing plugged in. Replacing the GFCI cured the problem.

You've never encountered one which fails to trip at the massively excessive current applied by the test button. How many times (if ever, since leaving the factory..) have any of those breakers actually had their trip characteristics verified?

I have to assume somebody has verified that. How much current does the test button use? It would not be particularly difficult to apply a known live to ground leak to test it, IIRC the spec is something like 5mA.

Either way I'm not terribly worried about it. My house is old enough that only the bathroom and outside receptacles are GFCI protected, I haven't bothered to update the kitchen. I treat all circuits as if they are unprotected even when I know they are. The GFCI is there as a last resort to hopefully prevent electrocution but I make every effort to never end up in a situation where I need it. Sure it could happen, but so can lots of things.

[helius]

Of course if your finger forms part of a circuit between phase and neutral, the power will still return through neutral and the thing won’t trip.

AFAIK, in Europe the neutral is ultimately connected to ground so in the case you refer, the CFGI breaker should trip. In Spain where I live it is imperative by law to have at least one of these with a sensitivity of at least 30mA in homes and small business and 300mA in industries.

No, the ultimate connection doesn't matter here. What is relevant is the branch currents in the phase (live) and neutral as measured at the RCD/GFCI. Only when they are unequal in magnitude by more than 30mA is the breaker specified to trip. The only way for the branch currents to be unequal is if some current returns to the source by another path: for instance, by taking the protective earth wire if there is an insulation fault, or by taking a path through metal plumbing fixtures if a live wire falls into water. It will also trip if the current returns on the neutral wire of a different branch circuit.

A person putting their finger in a lamp socket, for instance, will bridge between live and neutral, but without a connection to protective earth, there is no way for the measured currents to be unequal in magnitude.

Do you have a pointer to the tester circuit you mentioned?

[Calambres]

I don't know much about the differences between the power grid in USA and Europe but there are no "branches" here. In Europe, if you measure continuity between neutral and earth you'll find a value near zero so if you put a finger in a lamp socket the GFCI should trip, but don't ask me to try   

Regarding the tester circuit I use, first of all I must say that it is not my design. It is a very simple "AC variable dummy load" using a triac:

I took it from a youtube tutorial:



the description is here:

https://100ciaencasa.blogspot.com.es/2018/02/

There you'll find the schematic and the BOM except for a load resistor: RL. If you want to know its value you'll have to get into patreon or... use the Ohm's law    It is there to limit the maximum load the circuit can deliver. I used 4K7 for 50mA with a nominal input tension of 240V (51mA to be precise) and used a 50mA AC analog milliammeter instead of the proposed 30mA panel to prevent it from "pegging" if the failing GFCI goes up to the nominal full scale. All capacitors are for 400V for security. The resistors are 2W except the load resistor (RL) which is 10W also for security reasons. Of course, both links are in spanish.

See my own rendition of that schematic, with english captions, below.

Bear in mind that it is designed for european standard electrical home installations, i.e.: 240V and Schuko base plugs which are not "polarized", that is, you can insert the plug in the base in one way or the other so you're not sure where goes the phase or the neutral and hence the need for the switch. In USA you may skip that section as plugs are "polarized". Here you insert the plug and toggle the switch until you see a nominal reading of about 7 or 8 mA, with the potentiometer fully counterclockwise, of course: then you've selected the phase as the input. If you select the neutral the panel will show no reading at all. Then you must rotate the pot clockwise very slowly reading the panel until the GFCI trips. One better option is to substitute the analog panel for a decent DMM with MAX/MIN. The DMM will then tell you the exact maximum mA just in the moment the GFCI tripped. Don't forget to turn the pot CCW again or disconnect the circuit or you won't be able to reconnect the GFCI back again.

I decided to keep it simple and made it all in a plastic box with the analog panel, which I bought very cheaply in ebay. I'll make a photo of it the next time I visit my bench.

[ElektroQuark]

Of course if your finger forms part of a circuit between phase and neutral, the power will still return through neutral and the thing won’t trip.

AFAIK, in Europe the neutral is ultimately connected to ground so in the case you refer, the CFGI breaker should trip. In Spain where I live it is imperative by law to have at least one of these with a sensitivity of at least 30mA in homes and small business and 300mA in industries.

I'm in Spain too.
I have no neutral but two lives.


BTW, here we call those "differential switches".

[Calambres]

I have to assume somebody has verified that. How much current does the test button use? It would not be particularly difficult to apply a known live to ground leak to test it, IIRC the spec is something like 5mA.

I read somewhere that the test circuit in these buggers are made by an internal 5K 2W resistor. In Europe with a typical 240V line it means that the current is 48mA which is above the nominal maximum current for them. Let's see if I can open the failing unit and look what's inside...