Arc Fault Circuit Interrupter Triping
I just installed, or tried to install, an AFCI on an existing circuit that was added onto for a second floor bedroom remodel. The original circuit fed a downstairs bedroom and had a feed roughed in to the second floor for later expansion. The downstairs wiring has been in place for 7 years. I installed the AFCI and it keeps tripping. I disconnected all the new second floor wiring and the AFCI still keeps tripping. I even bought a second breaker to rule out a defective breaker. The panel in my house is a sub-panel fed underground from my garage where the main panel is.
Could this be caused by some ground/neutral interactions? I suppose it is possible that I have a problem with the original wiring, but I installed it myself and was very careful and there have been no issues for 7 years. Only outlets are fed from this circuit and every thing is unplugged.
A second AFCI installed on a second circuit functions without tripping. The test button works as intended.
Any hints as to what to look for? If there is some arcing even with nothing connected, how do you isolate the source?
Tim
Replies
Is it tripping immedately?
An AFCI has a ground fault detector, much like a GFCI, but it trips at 60 ma (IIRC) unbalanced current instead of 5 ma. The idea is to detect failing insulation.
So if the neutral is connected to the ground or the neutral of another circuit downstream then it will trip because of the unbalanced currents.
Likewise if the hot from antoher circuit returns through the neutral of this one it will trip.
Yes, the circuit is tripping immediately, even with nothing connected. The strange thing is I seem to read 0.7V AC with a meter even with nothing connected.
I guess I can go outlet by outlet and break the hot feeder from the rest of the circuit to see if I can isolate.
Not sure how to proceed.
TimYou buy a cheap tool twice and then you're still stuck with a cheap tool!
Peter,
You're barking up the right tree, in terms of troubleshooting. But instead of going outlet by outlet in sequence, pick an outlet in the middle of the circuit and break the circuit (all three conductors) there. This is called the "Rule of Halves"--disconnecting the circuit at the mid-point will tell you whether the problem is in the near half (nearer the panel) or in the far half.
This is a little more complicated than just identifying a short or open, because the AFCI has a GFPE (ground fault for protection of equipment) feature. It trips at about 30 mA of leakage out of the circuit. The trip could result from a hot-neutral, hot-ground, or neutral-ground connection or contact. The first two (short and ground fault) will trip a standard breaker if the fault is solid.
The classic approach is to cut the circuit in half and see if that localizes the problem to one half or the other. So, cut power, and at about the middle of the circuit, open all three conductors. Use an ohmmeter or DMM to check resistance of hot-neutral, hot-ground, and neutral-ground in the segment of the circuit away from the panel. You may find a neutral-ground connection, i.e., resistance will be something other than infinity. That's what I often find when AFCIs act up on power up.
Don't look in the direction of the panel if the AFCI breaker is in the circuit, because the AFCI breaker has circuitry in it to detect a ground neutral fault, basically an injector coil that imposes a little voltage on the neutral even when there's no load. GFCIs have this too. It means that if you read resistance neutral-ground of an AFCI or the load side of a GFCI, you'll see a couple thousand ohms. Which many continuity testers will read as continuity, and you'll then mistakenly interpret as a neutral-ground short.
If indeed you find the fault in the far segment, go to the approximate middle of that segment, break the circuit, and repeat. The point is to narrow the location of the fault until you locate the mistaken connection, or the ground wire touching the neutral or hot terminal screw. The thing about this last situation is that in pulling the receptacle out of the box, you clear the fault.
A completely different approach to this is to install a standard breaker in place of the AFCI and see if it holds. If not, you've got a ground fault or short, and you apply the rule of halves to find the contact 'tween the hot and the neitral or the ground. If the standard breeaker holds, it's a neutral-ground connection, or a neutral shared with another circuit. And then it's applying the rule of halves to find it.
Oh, and make sure ALL cord-and-plug connected loads are disconnected from the circuit. Unplug everything from the receptacles. Watch out for things like lighted light switches, dimmers, night lights, doorbell transformers...these will read as a high-resistance short.
I find shared neutrals a lot too when retrofitting AFCIs, and the GFPE feature of the AFCI will sure detect it.
Some guys evidently didn't learn that you bond all the grounds in a box (even from different circuits), but you have to keep the neutrals of different circuits separate.
The idea is to have the outgoing and the return current in the same cable or conduit so EMF fields cancel; if they don't, this can cause inductive heating of metal boxes etc.
Neutrals of separate circuits need to be kept separate for safety too; an inadvertantly shared neutral can cause a good shock or worse if you're working on a circuit where you think you've cut power, but the shared neutral (that 's dead when you test) then goes hot when you're working in it because some upstream load is turned on.
While troubleshooting, keep in mind that there can be more than one fault and more than one kind of fault in a circuit. Your DMM is your friend.
A word of advice--do as much troubleshooting as possible on dead circuits. Yea, sometimes it's necessary to check voltage or see if the breaker or AFCI breaker holds, BUT it's very easy to forget the status of the circuit when you're on the trail of a fault. It can be a shocking experience when you forget that the circuit is hot.
Lowly old 120V can kill. Do you feel lucky, punk?
Have fun, and good luck--
Cliff
Great advice.
Just to add that I've seen a new wire with a small nick under the insulation, from the factory, trip these arc faults.
You may be pulling, carefully, a new run.Troy Sprout
"D@mn... forgot the screws."
My existing breaker holds and has held the old part of the circuit for 7 years and the new part for at least 6 months. No hard short of hot to neutral or ground. No shared neutrals either shouldn't be any shared grounds except in the panel. I installed the wiring and only 1 circuit in any of the fed outlets (only outlets on this circuit).
Any issues with the panel being a sub with seperate ground and neutral busses? The 1st AFCI works, so I would guess this isn't an issue. I also switched postions for the AFCI to make sure it didn't have something to do with the leg of the 220V I was on.
Thanks for the tip on seperating the grounds and neutrals too! I only disconnected the hot in my new circuit, so I'll disconnect all and retry.
I suppose I could install a GFCI breaker to check for ground faults, this would be easier than pulling all the outlets (using the rule of halves as you call it -- or successive approximation!!).
Life is a journy isn't it.
TimYou buy a cheap tool twice and then you're still stuck with a cheap tool!
Tim,
No point in installing a GFI breaker. The AFCI trips at 30 mA leakage, and the GFI at 5 mA. Both do the same thing, it's just a matter of how sensitive they are to leakage.
Good point about the possibility that there may be an insulation flaw (either from the manufacturer or caused during installation) sufficient to trip the AFCI.
If that's the case, it'll become apparent once you've broken the circuit into segments and find a fault between boxes. Then it's time to pull new cable. I have a TDR and have used it to pinpoint such faults (both in power and communications cables), and the solution is the same--pull new cable.
Keep in mind that the fault may be in a device (receptacle or switch) or a fixture, and not in the wiring itself.
Once you isolate the fault to a segment of the circuit, then it's time to look at every aspect of the current pathway--devices, fixtures, damaged condoctor insulation (when stripping the sheath off of romex), an errant staple, a sheetrock screw, a cabinet mounting screw. I just finished a kitchen remodel where a cabinet mounting screw caused a ground fault in the romex feeding a section of plugmold. I was happy to have checked the circuit when I did--before the custom tile backsplash went up. And I was very happy that the fault was at the very bottom of the upper cabinet, where I could reach it through a hole without having the cabinets removed.
The AFCI breaker being in a subpanel (with a floating neutral bus) doesn't matter.
One other idea, take one of the AFCI breakers that trip on the circuit in question and install 'em in the subpanel, but do not land any branch circuit conductors on 'em. Just clip the breakers in and connect the neutral pigtails to the neutral bus. See if the breakers hold. Then try each on the circuit that's already successfully on the AFCI breaker.
There's always the possiblity that you've got two bad breakers--it happens. Rule out the "bad" AFCI breakers first. That's a lot easier than trying to find a problem in the branch circuit when there may not be one!
Troubleshooting is fun--once you've found the fault.
Cliff
No point in installing a GFI breaker. The AFCI trips at 30 mA leakage, and the GFI at 5 mA. Both do the same thing, it's just a matter of how sensitive they are to leakage.
They both detect ground faults the same way. But AFCI's also look at the transient behavior of the circuit, in an attempt to detect arcing. They have an embedded microprocessor in them, that makes a real-time analysis (BTW, there's a small power supply embedded in the breaker to power the microprocessor, which is why these things run a little warm, even with no load).
I think the suggestion to replace it with a GFCI is to ascertain whether the trip is due to a ground fault, or the arc fault detection feature.
Ths talk about frustrating troubleshooting reminds me of an experience i had with my car awhile back. I'd be driving along and the thing would just die. It was always a blown fuel pump fuse. The wierd thing was that, if I replaced the fuse, it would blow right away. However, if I waited "awhile", I could replace the fuse and it would be just fine. It might stay fine for days, weeks, or months at a time. When it blew, sometimes it would be fine again in an hour; sometimes I had to wait overnight. 'drove me nuts.
I finally found out that the problem was that, when the dealer had replaced the fuel pump when the orginal went bad, they pinched the wiring between the fuel tank and the chassis when they put the tank back in place (you need to drop the tank to replace the pump). Vibration had worn a hole in the insulation on the wire. Every once in a while, a short to the chassis would develop through the hole. If you left it long enough, corrosion would develop between the chassis and the conductor, which would clear the short. Until next time, that is.
CliffThere was a recent thread in the tavern about troubleshooting. It started about a story about a problem with a diesel engine on a ship.But I had this story, thought that you would appreciate it.....TW, I write software for embedded microprocessors. One of my clients makes the electronic control panels for newspaper "damping system" and sells it to the company that makes, installs, and services the damping system. (Part of the ofset printing process the rolls are dampend with a water mist.)They have a master control panel and each press tower has local panels with the drive electronics for the nozzles.These are all tied together with a serial communications line.At this one news paper they had problems. They system would work fine until the presses got up to speed and then they lost communications. Worked fine at the low speeds that they run when they first start, but then they start accelerating the presses.The service techs had worked and worked on it. They said that it was electrical noise. So they installed noise filters on the power wiring. Still had problems.Brought in the electricain and they ran new grounds and check all of the power connections. Still have problems.To this day I don't know why I was selected to trouble shoot this as I mostly work with the software and not the hardware.But I went.Now I could not control the presses as they eat up too much paper. I had to wait until they would starting a run.Put a scope on the com line and as expected everything was letter perfect.They the presses started and still good. Then the press speed was increased util they where really rumbling and the com lines went to sh*t. But it was not normal looking noise.So I started isolation the lines and the problem was coming from one side of the presses.Start isolating that side by opening up the control boxes and removing the com connections.Started removing the screws from one box and it started working. Tightened them back up and it stop.Found a couple of strands of wire from the com line that did not get in the connector. When the presses where runing the vibration caused it to short out the the case.
Bill,
That's a good one. The smallest things can foul you up.
A couple of strands of 30 gage wire in the wrong place, ouch!
Similar to a common problem in the telecom trade...at least in years of yore.
The standard wire gage is AWG 22 for a voice or data line (a "pair"). In the days when the leads at the station jack were connected via terminal screws, you had to strip the 22 gage and wrap it around the screw. Standard technique was to strip 2" or more of insulation off each lead to make it easier to wrap it around the terminal--on a good jack, there are brass washers under the screw head and one wire gets sandwiched 'tween two washers. Makes for a solid connection. Anyway, if you forgot to cut that long wire off after you'd tightened the terminal, sometimes the bare copper would touch another terminal, causing a problem.
Sometimes it'd take a while for the wire to move due to vibration or just gravity and make contact with another terminal. So, things would be OK at the time of the install, only to develop trouble later. These stray wires are called "shiners"--sometimes the best way to find 'em is to move the jack back and forth with a flashlight beam on it, and look for the shine from the bare copper.
All a moot point now, it's all insulation displacement (punch down) terminations on commercial jacks.
Cliff
AFCI [Arc Fault Circuit Interrupters] have a GFI [Ground Fault Interrupt] built into them. This is the main mechanism of both. You will probably have some connection between a ground wire and a neutral or white [or, to be technical, grounded conductor]. Replace the AFCI with a GFI and you will 99.999% of the time get the same results.
Somewhere there is a slight connection between your ground wire or even earth and your neutral. To trace it down would require the usual trouble shooting procedures. 4 Lorn should give you a pre-article essay on this, [if he please]. It could be a nail or screw hidden in your wall. puncturing the Romex.
AFCIs are a legally required rip-off. If you want the protectection, replace them with a GFI.
To ne continued...
~Peter
Either Seattle collapses tomorrow or remote viewing doesn't work.