GFI Caused a Fire!

[gfretwell]

The science of AFCIs is proprietary and they are not all the same but they look at current, detecting short duration spikes

[Ashton Crusher]

I may never sleep again.

[Robert Green]

I'm not sure I understand the last sentence. Do you mean that the AFCI's don't use the powerline to transmit the "noise" created by an arc to the detector/signal processor in the AFCI? From what I've been reading at least some part of the detection process is based on detecting the RF noise from an arc. I also read that detecting an arc is difficult because it may not be drawing detectibly larger amounts of current than a normal device might.

One of my CCTV cameras is badly shielded an acts as an arc detector. Whenever I run a motor with brushes near it, the image starts to show random blips across the screen. It's really noticeable when an old power drill is operated anywhere near the camera. I've been assuming that noise is one part of what an AFCI "looks" for in trying to determine if arcing is present.

I'm asking because I've got three different items that appear to use the home powerline as "antennas" although perhaps that's not the right term. Each devices uses the powerlines in a slightly different way. X10 home automation controllers inject a 120kHz RF signal onto the powerlines to communicate commands. My Netgear Ethernet adapter sends networking signals over the home powerlines with plug-in adapters (sometimes - it's not very reliable):

I've seen numerous discussion refer to the HomePlug technology as "transmitting" ethernet signals over the powerline. Is the powerline acting as an antenna in those cases or is their a better terminology for what's happening?

I would also guess that the AFCI's are looking for several parameters to determine if there's an arc fault. I would suspect that sort of detection has improved greatly since the first units hit the market.

This site (from 1999!) claims:

Haven't been able to find information about how AFCIs and arc welders interact but it seems most run off 240VAC and would have their own circuit, probably without an AFCI.

This site talks about the original bedroom requirements and says it's because that's where many home electrical fires start:

The same site also implies that arc welders do not generate nuisance trips with the newer AFCIs.

The same site has a good roundup of arc fault types and possible causes:

What is an arcing fault?

According to UL 1699, Standard for Arc Fault Circuit Interrupters, an arcing fault is an unintentional arcing condition in a circuit. Arcing is defined as a luminous discharge of electricity across an insulating medium, usually accompanied by the partial volatilization of electrodes.

There are 3 basic types of arcing faults: line-to-neutral, line-to-ground, and series arcing.

What causes an arc fault?

Arc faults may occur anywhere in the electrical system and may be a result of the following:

a.. worn electrical insulation or damaged wire b.. misapplied or damaged plug in appliance cords and equipment c.. loose electrical connections d.. drill bits, nails, or screws driven into the wire e.. wire staples driven too deep f.. furniture pressing against electrical cords g.. broken wires h.. frayed wires More reading is in order before I bite the bullet and begin replacing the rather new breakers I installed in the panel just recently. Does anyone reading this know if AFCI's come in "dual skinny" formats? One of the site above talks about how AFCI's run warmer than normal breakers because of the built-in power supply for the electronics. I wonder if they can squeeze all the required electronics into the dual space-saving breakers?

Thanks for your input, Bud!

[Robert Green]

I'm confused now. Do the AFCI's use RF at all to detect arcing conditions? I've been trying to find some descriptions or patents that detail how they work without much success. One interesting site I found:

said that one user, who thought his AFCI was "nuisance tripping" discovered that it was actually tripping because of a dangerous condition in his bathroom.

Maybe it IS time to switch to AFCIs!

[Nate Nagel]

In my limited experience with them, I haven't had any issues. Every time it's tripped it's been because of a legitimate problem.

Also, the amount of improper wiring that I found in my last house, combined with simple age and the tendency for old 60C wiring to become brittle in ceiling light boxes, makes me think that in general they're a good idea. I lived in my last place for almost a year before going up in the attic and doing a serious investigation of the wiring, what if something had happened before I started fixing everything? And since everything was working, a homeowner less inclined to do it themselves could have lived there a decade or more unaware of any potential issues (as I believe to be the case of the previous owners.)

No, they're not a replacement for proper wiring products and methods, but in a structure where the wiring is apparently in good shape but known to be old and not up to today's code, it makes sense. For newer houses where everything's right, it might be overkill, but won't hurt anything either.

nate

[bud--]

RF from the arc is not picked up by circuit conductors acting as antennas.

AFCIs look at the current on the conductor. The arc "signal" is created by current variations through the arc.

That is the current variations created by the arc.

A "series" arc (loose connection) is limited by the normal load that is on the circuit. The AFCIs that are used now may detect a 5A or larger arc.

Parallel arcs (H-N and H-G) can be a lot larger. The current can be up to the available short circuit current at the point of the arc. If I remember right, an investigation found that is very likely over 60A out to 6 ft of line cord in a house.

The signal from the camera isn't very large so a small noise can be quite visible.

AFCIs need a 5A or larger "noise" on power wiring that runs at much higher voltage and current than the camera.

[bud--]

Looking at the current in the wire, a "spike" is made up of frequencies that are much higher than 60 Hz. My guess is AFCIs look at the low end of RF and what is in the audio frequencies. I would also guess AFCIs pick up the current signal with a current transformer on the hot wire. (GFCIs use a current transformer around both the hot and neutral.)

[gfretwell]

Most nuisance trips can be traced to a ground fault, not an arc, usually in the neutral.

[Robert Green]

I wonder if it's really fair to call them nuisance trips if there's a real, underlying electrical fault causing the breaker to open? (-: The more case studies I read, the more I am convinced that installing AFCI's is a good thing if only because there almost always *is* a problem of some sort that they've revealed if they do seem to start tripping for no reason. While that may not have always been the case - I believe that the first AFCI's did trip too easily from what I am reading - it seems that now when people install them and they trip with regularity there's a good reason.

Now the question is - who makes the best AFCIs and do they make them in a format I can use (dual skinnies)? I can replace the bedroom breakers, which are mostly single, with full-sized AFCI breakers, but I'd like to protect other circuits that can carry northward of 10A regularly to be protected as well.

More Googling required. (It's hard to remember a time when if you wanted to look something up, you had to go to the library!)

[Robert Green]

It seems that's the case for many people who have installed them.

You wouldn't believe the things I discovered when I ripped out the poorly finished basement's walls and ceiling. The additional wiring was done with equal disrespect. Neutrals pulled from separate circuits, wires nicked by bad stripping, grounded outlets installed without any grounds, wire nuts improperly installed, speaker wire instead of Romex, etc.

I have an old house, too, built when WWII was just starting and all sorts of compromises had to be made because of war shortages (no building paper between the floors, for example). I can see great benefit in installing AFCI's to help detect any concealed faults in the wiring system.

I've added a number of new circuits for high-current devices over the years because I didn't feel comfortable taxing the old, cloth covered wiring with loads it wasn't designed for, but those circuits are still active. If I switched them to AFCI's and they started tripping, I would probably just disconnect them and put in a new circuit with Romex because tracing a fault through the old wiring would be very hard. The old wiring runs up to the attic and then down again and everything's behind plaster and lathe walls and ceilings.

In any event, I agree with you that with old wiring, installing AFCI's could be a lifesaver.

How much work was involved when you checked/rewired the attic, Nate? I decided that is was far easier, although slightly more expensive, to just put in new, grounded 20A circuits from the basement up wherever I could to replace the 15A cloth-covered wires going up in the attic. I found that the cloth-covered wires were still mostly OK, except around ceiling junction boxes and switches that had been replaced. Considerably fraying occurred wherever the wire had been disturbed for any reason.

Agreed.

[gfretwell]

A ground fault on the neutral will not open the overcurrent device but it will trip anything with ground fault protection. A common suspect is that big kludge of white wires under a wirenut in a ceiling box. That is a common cause of the "fan"problem. The fan vibrates the box and a bare ground bumps into an exposed part of a white wire. This is a case where taping up a wirenut may be a good idea.

The size of the AFCI precludes them being in a skinny format ... so far.

[gregz]

I installed arc fault breakers on two legs, which included both bedrooms. This was old wiring. One leg would trip on occasion. I called that leg the big mess, because if you saw it, that's what yo would call it. Most if that has been replaced, so far so good.

An arc fault breaker is not also a ground fault is it. Are there some ?

Greg

[Robert Green]

Yes, many sites I've scanned talk about vibration loosening connections in junction boxes. When a wire pops out of a wire nut it can really create havoc.

It looks like my options are not good for replacing all the current breakers with AFCI's. They're expensive, too, at least the list prices I've seen are. In most of the new lines I've run, I know where the first outlet in the circuit is located and I can replace it with a AFCI outlet instead of a breaker except that most of those outlet boxes are occupied by GFCI's. )-: I am not fond of using outlet protection devices because it means you have to go all over the house when a GFCI/AFCI outlet "pops" - been there, done that. That reminds me to annotate the breaker box legend with the location of all the GFCI outlets . . .

Besides, it's not the new circuits I am worried about. It's the old, cloth-covered wire circuits that I am most concerned about. I'll have to review the circuit panel and decide which ones regularly carry more than 5A and will consider replacing those regular breakers with AFCI units. From what Bud has said, an AFCI requires at least that much current to trip.

[Robert Green]

60A through a 14GA cord is going to do some damage. I remember when I was a kid an electrician working on the circuit panel with a screw driver shorted it out inside the box with a huge spark that spattered molten metal inside the circuit panel. I still have that screw driver around somewhere. He just tossed it in the trash because the tip was cratered but I retrieved it because I thought it made a great souvenir and reminder to be careful.

It's clear now from what you've said and what I've read that it's not a "radio" process but a monitoring of the instantaneous current draw that detects the arc. I wonder if the radio telegraphs used in the Marconi era would work today with all the background RFI that exists today?

The camera is only of use in detecting things like motors with brushes that need replacing. It even detects when a gasoline-powered leaf blower is nearby.

[Robert Green]

Yeah, I had a similar circuit in the basement that had been hacked into by hack amateurs so many times that yanking it was the only thing to do. It was the circuit that had speaker wire coming out of one junction box and going into a flourescent shoplite without any strain relief. The wires went right into a sharp-edged sheet metal hole that had already begun to saw through the insulation. Yikes!

Good question. I seem to recall reading something about one, but I can't recall whether it was an actual product or a prediction. I also came across someone who's promoting a scheme whereby all outlets and all plugs have RFID chips in them so they can communicate with each other and establish what the normal current load should be. That would make it easier to detect way out-of-bounds over-current draws, but the scheme sounds too complicated to ever be adapted universally. If I find a combo AFCI/GFCI unit, I will let you know. I suspect if it's available, it would only be in an outlet form-factor.

[Robert Green]

It's interesting that in all my research I've yet to come across a circuit diagram or anything technically detailed about how AFCIs actually work. I'm going to bite the bullet and order an AFCI outlet to run some tests with. They ain't cheap!

runs $33 from Amazon. If anyone knows of a cheaper vendor, please advise. Also, how would one go about creating a deliberate arc for testing purposes? I've got some carbon rods lying around somewhere that I used to create a carbon-arc light with. I suppose that should work . . .

[TimR]

I found this quote on a Inspectapedia, home inspection site:

"As of September 2008 we have found no test tool that reliably and complete ly tests the function of an AFCI. Only the integral test button tests the c ircuitry of the device as well as the trip mechanism. UL classes these "tes t" devices not as "testers", but as "indicators".

A problem is that some devices used to "inspect" an AFCI, in trying to prod uce a simulated arc fault condition, may fail to cause the AFCI device to t rip even though it is perfectly fine.

Literature from the manufacturer of a popular "test tool" tells the user of the tool to go to the electric panel and use the test button on the AFCI d evice to make sure it trips. In other words the inspector cannot rely on th e separate test tool. For this reason you will see such tools referred to a s "indicators" rather than "testers": they are not a complete and reliable test instrument for AFCIs"

So, trust the test button, because there is NO external way to verify the d evice works? Huh? Does that make sense?

[Robert Green]

I've noticed that almost all the AFCI breakers I've been looking at have pigtails attached. Those are going to make for a very messy looking installation in my cramped box, IMHO. I see that one of the newer circuit panels has accommodations for the AFCI pigtails but now we're talking serious expense and time to replace my old panel with an AFCI compatible one.

One of the sites I browsed made an interesting comment:

The breakers, to my surprise, all seem to be more expensive on the whole than a OBC (Outlet Branch Circuit) outlet-format protector. So $50 x 20 breakers is $1,000. The real problem is that now I know that the AFCI's require pigtails and full breaker slots, it means I would feel compelled to pull the old box, replace any cloth-wired circuits coming into back to their source (that's the long pole in the tent) and put in a new box designed for AFCI's and then probably have to heavy-up the service to the pole. So the math isn't quite as clean as $50 by 20. I'd think that's a multi $K job for a licensed electrician and at least $2K for a homeowner DIY'er if everything listed is done.

What do electricians charge to run a new branch circuit in old plaster/lathe construction? I've always done it myself so I confess, I have no idea.

The question now is how much is piece of mind worth? Is the money better spent on a panel full of GFCIs or on smoke, heat, CO detectors or other safety technology? AFCI's, once perfected, could just as easily ride the main breaker, couldn't they? The arcing "information/signature" should be detectable at the service entrance, shouldn't it? Admittedly nuisance trips that bring the whole home "grid" down aren't going to be pleasant, but I know that the new home industry would rather not have to add over $1000 in AFCI costs to new homes if there's a cheaper way to do it.

One site said that there are at least 111 arc-fault based fires in the US each year. Statistically, that's really not a light. Can't vouch for the accuracy of the number. Seems like it should be higher. Could be CRS or the webmaster took a WAG. (-:

Is it time to switch out the old breakers or will there be combo AFCI/GFCI breakers sometime down the line? I see that some of the AFCI's incorporate a form of GFCI, but the trigger appears to be 30mA and GFCI's appear to operate at a more life-saving 5mA level.

I did notice that more than one site (Leviton and Eaton/CH, I think) took pains to point out that they are continuously improving their arc detection capabilities. Cynical people might take that as an acknowledgment that they feel they are "not quite there yet" in terms of 100% reliability.

[Robert Green]

"As of September 2008 we have found no test tool that reliably and completely tests the function of an AFCI. Only the integral test button tests the circuitry of the device as well as the trip mechanism. UL classes these "test" devices not as "testers", but as "indicators".

Funny you should bring that up because the Eaton/CH site I was reading mentioned using a circuit analyzer *before* installing an AFCI to help scan for potential problems that would cause a nuisance trip. So I started looking at circuit analyzers that could tell me if there was a problem with the in-wall wiring. The cheapest I found was $90 for this:

NEEWER TRMS Voltage GFCI RCD Tester Circuit Analyzer MS5908A but it doesn't appear to test AFCIs although it does test for Residual Current Devices which turn out to be like a GFCI, but speak with a British accent . From Mike Holt's forums:

< RCD is the term normally used in the UK, and the most common trip rating is 30ma though many other ratings exist. A UK type RCD normaly only protects against leakage to earth/ground, and does not protect against overload or short circuit, a fuse or MCB must also be employed to protect against short circuit or overload. A combined earth leakage and overload protective device is available, these are known as RCBOs. A UK RCD built into an outlet does NOT protect downstream outlets. The common 30ma trip current does in practice give good (but not total) protection against shock, especialy as they normally trip at about 20/25ma, and are faster acting than USA GFCIs.>

The other tester they just got pricier from there: Amprobe INSP-3 Wiring Inspector Circuit Tester Was almost $300 and folks on Amazon had some complaints about it.

The Ideal Industries 61-165 SureTest Circuit Analyzer weighed in at over $330 but deals with both AF and GF CI's but users complained that it was "short" some necessary adapters.

So it looks, at first glance, like a circuit analyzer that can "test" AFCI's will set me back the cost of at least 6 AFCI breakers. The question now is will testing all the circuits reveal the same sorts of problems that nuisance tripping of AFCIs would also "detect?" Is money better spent on the AFCI's themselves or on a tool that can reveal potential hazards?

even though it is

Gives you that "warm, fuzzy" feeling about their overall effectiveness, doesn't it? That's precisely why I've been thinking of buying a single OBC outlet AFCI unit and testing it under real world arcing conditions that I create - the kind of testing that drives SWMBO mad. (-:

of the tool to go

sure it trips.

this reason you will

not a complete and

Great. This AFCI investigation is rapidly spreading out into a murky wamp. )-;

device works?

Only in Bizzaro land. (-: What bothers me most about the lack of testers that can actually TEST and not just INDICATE is that I would always be suspicious that the AFCI's might not react to a real arc. There's something just not scientific about not being able to create a reliable, repeatable tester for a device intended to save your life. Sometimes, when manufacturers have problems like nuisance trips, their first corrective attempts overshoot the mark.

I guess I had just better order a single AFCI and start testing while I try to decide whether a $300 circuit tester would be a good investment.

[Robert Green]

Are you sure that's still true?