Stange Problem with GFCI

[Terabyte2007]

I am not an expert on GFCI, but I do know how they work, however I have had a strange problem in my lab with one in particular that keeps nuisance tripping. Here is the scenario, I have one part of the lab which uses a computer and about 8 to 12 low powered devices which not all on at the same time. They are fed from a 20A circuit to the GFCI and (2) standard outlets are supplied by the "LOAD" side of the GFCI for a total of (3) outlets. Equipment has been running for a few years on this setup with no problems until recently I started to get tripping of the GFCI several times a day. Now, here is where it gets weird, I have monitored the power coming into the GFCI and it appears to be good. I have checked for leakage current which is in the range of less than a few mA. The GFCI should not trip until at least 5mA> from what I know. This problem has never occurred at night, only in the middle of the day and not instantly when equipment is used. I can do all the same activity on the equipment in the evening with no issues at all. I check the power feed coming into the lab with an RFI line meter and get around 150mV or less of HF noise which is not bad considering I have seen upwards of 400-500 mV in new building installations. I have unplugged equipment, tried various combinations and cannot seem to isolate this problem. The current draw is maybe 2 amps at best and averages around 600mA to 1.2A. Since GFCI is not effected by current draw I don't think this is the problem. I guess it could be a bad GFCI, it's not to terribly old though. Their is an APC UPS on one of the "LOAD" outlets but it has been running for over 6 months now without issue.

Any recommendations to try before I pull this GFCI and try a new one, or even replace it with a standard outlet?

[ajb]

Are there significant temperature/humidity fluctuations between the daytimes and evenings?

Probably the most informative troubleshooting step would be to replace the GFCI, and if the problem persists it's a good indication that the problem is a result of the setup or environment rather than just a faulty/worn out GFCI.

[Terabyte2007]

Are there significant temperature/humidity fluctuations between the daytimes and evenings?

Probably the most informative troubleshooting step would be to replace the GFCI, and if the problem persists it's a good indication that the problem is a result of the setup or environment rather than just a faulty/worn out GFCI.

No, the temp/humidity is regulated and it stays pretty close to the same day and night.

I was trying to avoid swapping it if I could, but that may be my only option soon.

[SeanB]

Probably leakage current has increased in something, along with a failing GFCI which is getting more sensitive to noise. Easiest is to replace it, they are cheap enough. Better is to give each outlet it's own GFCI controlled circuit, that way the ground currents do not add up to a trip level.

I have had that issue as well with 20 SMPS units on a circuit that tripped at random times. Solved by adding a second GFCI and splitting the circuit into 2 parts each of which did half.

[Richard Crowley]

GFCI devices are relatively precision, but MASS produced commodity items.  And they are likely designed to "fail" safely. Which means that they may interrupt where they don't need to vs. the other option (to NOT interrupt where they should).

So just replace it. They aren't that expensive, and they weren't made to last forever.  When you get the old one out, do a tear-down video for us. I didn't find anything about "GFCI" when I searched the eevblog website.

[SeanB]

I think I should take an old 1960's technology one apart then, they actually have electronics inside like a BRY39 switch.

[128er]

First, a question ...

Do we talk about, what I know as a "RCD", residual current device?  All three, or only 1 phase and the neutral pass through a "summation transformer", the magnetic fields cancel each other out and there is no induction to trip the mechanism.

The differential current of your device is 5mA? fairly low, I think
How do you measured the leakage and where? If you have a current clamp, put the neutral and the phase in the clamp. There should be no current.

[Monkeh]

Do we talk about, what I know as a "RCD", residual current device?  All three, or only 1 phase and the neutral pass through a "summation transformer", the magnetic fields cancel each other out and there is no induction to trip the mechanism.

Yes.

The differential current of your device is 5mA? fairly low, I think

Normal in Yankistan.

[madires]

I have had that issue as well with 20 SMPS units on a circuit that tripped at random times. Solved by adding a second GFCI and splitting the circuit into 2 parts each of which did half.

The current through the Y filter caps adds up with each SMPS. The 5mA of the RCD could be easily exceeded. Therefore we got local restrictions for the max value of the Y caps to limit the current per device (30mA RCDs are standard).

[128er]

The differential current of your device is 5mA? fairly low, I think

Normal in Yankistan.

Ok, here in germany it is 30mA for circuits with sockets up to 32 Amps, that are accessible for electrical laymans (almost everyone). Lesser in medical enviorments for example. But the RCD's are allowed to trip at about 50% of the rated differential current. Most of them trip between 20-30 mA.

Edit:
madires is quite right about that

[Monkeh]

The differential current of your device is 5mA? fairly low, I think

Normal in Yankistan.

Ok, here in germany it is 30mA for circuits with sockets up to 32 Amps, that are accessible for electrical laymans (almost everyone). Lesser in medical enviorments for example. But the RCD's are allowed to trip at about 50% of the rated differential current. Most of them trip between 20-30 mA.

They don't protect circuits in the US, they protect outlets.

[128er]

They don't protect circuits in the US, they protect outlets.

With circuits i meant outlets. The RCD sits in the distribution board and protect the outlet. (Find the right term in english is not easy )

[Monkeh]

They don't protect circuits in the US, they protect outlets.

With circuits i meant outlets. The RCD sits in the distribution board and protect the outlet. (Find the right term in english is not easy )

Yes, I know what you meant. They don't do that in the US, they build the RCCB into the outlets.

[Richard Crowley]

They don't protect circuits in the US, they protect outlets.

Actually, there are GFCI breakers available to protect the entire circuit. They install just like regular breakers, but they implement the GFCI (or RCD) function by separate connection to the neutral bus for reference.

[Monkeh]

They don't protect circuits in the US, they protect outlets.

Actually, there are GFCI breakers available to protect the entire circuit.

And how often do you see those used? It is, from everything I've seen, not standard practice (like a great many other things..).

Not to mention protecting an entire circuit with a 5mA device would be lunacy.

Proper IEC 60898-1 and 60947-2 breakers are available in the US, too, but you don't use them in distribution.

[Richard Crowley]

Well, perhaps you don't see many there in the UK.  But here in the US, I have seen many of them. And I believe they are now legally required for kitchen, bath, laundry, and exterior circuits.

[Monkeh]

Well, perhaps you don't see many there in the UK.  But here in the US, I have seen many of them.

What we have here is a failure to communicate.

[madires]

Well, perhaps you don't see many there in the UK.  But here in the US, I have seen many of them. And I believe they are now legally required for kitchen, bath, laundry, and exterior circuits.

We've had that requirement for over 20 years. Nowadays the whole apartment needs to be protected by RCDs in the distribution panel. And you should install several ones to prevent a complete lights-off in the night and to deal with high load areas like a kitchen. Typical RCDs used are 3-phase 25, 40 and 63A (trip at 30mA max).

[Terabyte2007]

Probably leakage current has increased in something, along with a failing GFCI which is getting more sensitive to noise. Easiest is to replace it, they are cheap enough. Better is to give each outlet it's own GFCI controlled circuit, that way the ground currents do not add up to a trip level.

I have had that issue as well with 20 SMPS units on a circuit that tripped at random times. Solved by adding a second GFCI and splitting the circuit into 2 parts each of which did half.

I suspected something along these lines so I did a bit more troubleshooting today and found a Cheap Chinese Switch-Mode power wall wart to be likely putting the edge on the leakage current. It was a 2A/12V for a 20 dollar SATA drive caddy. When I plugged this PS in within 1-15 minutes it would trip. I don't think it was causing it by itself, but an additive effect with all the devices connected to the circuit. This just kinda put it over the edge.

[Monkeh]

Probably leakage current has increased in something, along with a failing GFCI which is getting more sensitive to noise. Easiest is to replace it, they are cheap enough. Better is to give each outlet it's own GFCI controlled circuit, that way the ground currents do not add up to a trip level.

I have had that issue as well with 20 SMPS units on a circuit that tripped at random times. Solved by adding a second GFCI and splitting the circuit into 2 parts each of which did half.

I suspected something along these lines so I did a bit more troubleshooting today and found a Cheap Chinese Switch-Mode power wall wart to be likely putting the edge on the leakage current. It was a 2A/12V for a 20 dollar SATA drive caddy. When I plugged this PS in within 1-15 minutes it would trip. I don't think it was causing it by itself, but an additive effect with all the devices connected to the circuit. This just kinda put it over the edge.

And this, folks, is why 5mA RCCBs are stupid.

[Terabyte2007]

GFCI devices are relatively precision, but MASS produced commodity items.  And they are likely designed to "fail" safely. Which means that they may interrupt where they don't need to vs. the other option (to NOT interrupt where they should).

So just replace it. They aren't that expensive, and they weren't made to last forever.  When you get the old one out, do a tear-down video for us. I didn't find anything about "GFCI" when I searched the eevblog website.

That sounds like a good idea, I will do so.

[denelec]

I've read that RCD were initially used in Europe to prevent electricity theft.
Because a load wired phase to ground instead of phase to neutral wouldn't be seen by the meter.
Can anybody confirm this?
In US and Canada, the meter is not referenced to ground.  It only sees the 240V (line-line) and the two line currents.
A phase to ground load will be measured.
GFCI are used only to prevent electrical shock.

[Monkeh]

I've read that RCD were initially used in Europe to prevent electricity theft.
Because a load wired phase to ground instead of phase to neutral wouldn't be seen by the meter.
Can anybody confirm this?

Load of crap. That's not how meters work or where RCDs are fitted.

[denelec]

I've read that RCD were initially used in Europe to prevent electricity theft.
Because a load wired phase to ground instead of phase to neutral wouldn't be seen by the meter.
Can anybody confirm this?

Load of crap. That's not how meters work or where RCDs are fitted.

This document from ABB seems to confirm the electricity theft theory:
http://www05.abb.com/global/scot/scot209.nsf/veritydisplay/c4e584f06cc6c4fbc1257ad800496193/$file/2CSC420004B0201_RCDs%20EN.pdf
See page 7 of the document: "It is not clear when and by whom the first RCD was developed, but it certainly appeared on the
market in the 1950s and was initially used by some utility companies to fight “energy theft” due
to the use of currents from phase to earth instead of phase to neutral."

So maybe I'm not so "full of it" somehow...

[Monkeh]

I've read that RCD were initially used in Europe to prevent electricity theft.
Because a load wired phase to ground instead of phase to neutral wouldn't be seen by the meter.
Can anybody confirm this?

Load of crap. That's not how meters work or where RCDs are fitted.

This document from ABB seems to confirm the electricity theft theory:
http://www05.abb.com/global/scot/scot209.nsf/veritydisplay/c4e584f06cc6c4fbc1257ad800496193/$file/2CSC420004B0201_RCDs%20EN.pdf
See page 7 of the document: "It is not clear when and by whom the first RCD was developed, but it certainly appeared on the
market in the 1950s and was initially used by some utility companies to fight “energy theft” due
to the use of currents from phase to earth instead of phase to neutral."

So maybe I'm not so "full of it" somehow...

I've never seen a meter with anything but a solid link for a neutral, so I'd love to know how that's meant to work..

[denelec]

I've never seen a meter with anything but a solid link for a neutral, so I'd love to know how that's meant to work..

Don't know.  Maybe old meters measured both the line and return current in the neutral?
Our meters don't measure the neutral current or line to neutral voltage.  Only the 2 line currents and the line to line voltage.
A line to earth load would appear the same as a line to neutral to our meters.
But I doubt that a modern meter would be blind to a line to earth load.

[Monkeh]

But I doubt that a modern meter would be blind to a line to earth load.

Neutral and earth in my house only diverge a few inches ahead of the meter. It would neither know nor care if I returned all load current via 'earth'.

[denelec]

But I doubt that a modern meter would be blind to a line to earth load.

Neutral and earth in my house only diverge a few inches ahead of the meter. It would neither know nor care if I returned all load current via 'earth'.

Yes. But what if you used another earth reference not physically connected to your electric panel such as a ground rod?

[Monkeh]

But I doubt that a modern meter would be blind to a line to earth load.

Neutral and earth in my house only diverge a few inches ahead of the meter. It would neither know nor care if I returned all load current via 'earth'.

Yes. But what if you used another earth reference not physically connected to your electric panel such as a ground rod?

Then the current would find its way through the earth to another rod connected to the PEN..

[denelec]

Then the current would find its way through the earth to another rod connected to the PEN..

You meter shouldn't see any difference unless it somehow monitored both the line and neutral currents.
We need an expert on old European meters...

[Monkeh]

Then the current would find its way through the earth to another rod connected to the PEN..

You meter shouldn't see any difference unless it somehow monitored both the line and neutral currents.

That's what  I said several posts ago... It, like every other single-phase meter I've looked at, has a solid link for neutral. It does not care whether load current passes through it or not.

[denelec]

That's what  I said several posts ago... It, like every other single-phase meter I've looked at, has a solid link for neutral. It does not care whether load current passes through it or not.

If you used one of our meter in your house, a 100W line to earth load would appear to it as 50W.
That's because a 120V load produce a current in only one line and a 240V load produce a current in both lines in our system.
Since our meters have no ground or neutral connection, L1 would be 240V and L2 would be neutral in your setup.
So both line and neutral current would be measured.
Maybe your old electromechanical meters were made like ours.

[Monkeh]

That's what  I said several posts ago... It, like every other single-phase meter I've looked at, has a solid link for neutral. It does not care whether load current passes through it or not.

If you used one of our meter in your house, a 100W line to earth load would appear to it as 50W.
That's because a 120V load produce a current in only one line and a 240V load produce a current in both lines in our system.
Since our meters have no ground or neutral connection, L1 would be 240V and L2 would be neutral in your setup.
So both line and neutral current would be measured.
Maybe your old electromechanical meters were made like ours.

We're not talking about your split-phase system. It has never been used here, we have never used a meter like that.

[SeanB]

Electricity thieves use a SWER system, as a single cable is easy to install and hide. Thus a ground return will show up as a fault.

[madires]

The common meter in Germany is a Ferraris type for 3 phases. It's connected to neutral, which is connected to the building's grounding system. Therefore the meter would also count any "stolen" power. RCDs were introduced to prevent electric shocks and not for detecting power theft.

[Terabyte2007]

Swapped the little bugger out last night. Here is the culprit. I will do a short teardown on this guy over the weekend. It's a Leviton with a 92 date code.

Also, this is a 20 Amp GFCI, can anyone spot the major difference to modern GFCI's??? Or even outlets?

[NiHaoMike]

A proper 20A outlet has that extra sideways slot. However, for some reason, it only seems to appear in commercial buildings.

[Richard Crowley]

The "sideways" slot implies that the full 20A is available from that outlet. It is very rare in a domestic situation to provide a branch circuit with a single outlet, typically there are  many outlets on a circuit. So even though the circuit may be 20A, you can't be assured that you are the only load, so those 20A outlets are inappropriate.

At least that is how mains wiring is done in North America. In UK, they have "ring" circuits with individually-fused outlets plugs. Dunno what they do down there in OZ?

Updated per correction from Monkeh.  Thanks!

[Monkeh]

At least that is how mains wiring is done in North America. In UK, they have "ring" circuits with individually-fused outlets.

Individually fused plugs.

Dunno what they do down there in OZ?

Normal single-phase radials.

[Terabyte2007]

Ok, folks. Here is the teardown pictures. I am sorry I did not have time to do a video teardown of this GFCI, but projects at work were keeping me busy.

Interesting find, the plastic outside plug case says that this GFCI is a 20 Amp but the stamp on the internal framework states 15 Amp, What's up Leviton?

BTW, the small bar with a resistor tacked to the end was the test button. This would actuate the protection circuit. Also, can anyone tell me what the RA9031A 8 pin IC is? I could not find the data sheet on this.

[SeanB]

Interesting that yours is mains powered. The ones used here are all powered by the imbalance current.Older ones ( late 1960's to early 1970's) used a BRY39 GCS thyristor to drive the trip relay, while the later ones up to the current generation use a simple biased coi to operate the trip mechanism, the trip drive energy being stored in a spring. They will operate with overvoltage, undervoltage and even with a broken connection to neutral on the supply side. There were a small range that had overvoltage and phase reversal protection, which used a 275V varistor on the line side and a 100v varistor on the neutral side, with the common along with the test resistor being connected to an earthing wire. There was also a range of prepayment meters which used the basic trip mechanism as a disconnector for the power, and they were quite common though they had a flaw in that you dould disable the trip by putting a needle into the unit through the plastic label they placed over the hole where the test switch was. Later on they put a plate there to block this hack. Newer meters also have back end checks as the payment is linked to the meter number, the old ones were a prepaid number that worked on any meter.