39

I just had a near-disaster while trying to install new smoke detectors. My house has AC hardwire connections for the smoke detectors. I wasn't sure which circuit the smoke detector AC connection was on, but after going through the breakers one at a time I found that they were on the same breaker as the dryer for some reason.

I verified that the circuit was off using two methods:

  • While the circuit was on, the smoke detectors all showed a green LED indicating AC power. After shutting off the breaker, the green LED went off on all smoke detectors.
  • While the circuit was on, my non-contact voltage detector went off when I held it near the hot wire connected to one of the smoke detectors. After the circuit was off, the voltage detector gets no reading from the hot wire.

I started with the smoke detector I'd used the voltage detector on. I unplugged the old smoke detector and checked to make sure the new smoke detector fit. I discovered that the new detector wouldn't fit because it has a bulge that would need to stick up into the junction box, but there was no way to make it fit because the wires in the box were so long (roughly 8 inches from the ceiling + wire nuts and another 2 inches to the AC connector) that they filled up the whole box. I decided to trim the wires back a few inches.

I grabbed my wire tool and cut the red interconnect wire first. Then I did something which in retrospect was very stupid: I cut the hot and neutral wires at the same time, thinking I was safe because the circuit was definitely off.

As soon as the cutter made it through the rubber insulation, it bridged the connection between hot and neutral and EXPLODED. A huge shower of sparks burst from the wires and I fell off the ladder screaming. By some miracle, I escaped unharmed and the sparks didn't set anything on fire.

I'm guessing the only reason I'm not in the hospital or dead right now is because I was on a fiberglass ladder and the wire tool has an insulated rubber handle. You can see here where the arc burned through the metal and lightly singed my finger.

The arc burned a hole in the metal and singed my finger.

After I recovered, I checked the hot wire with the voltage detector (still no reading), checked the other smoke detectors to make sure they weren't getting power (still no green light), and finally went outside to check the breaker (still off).

What went wrong here?

Edit: I keep thinking I figured out what circuit these damn things are on and then finding that the wires are still hot. There have been moments where it seemed like the most plausible explanation was that gremlins were switching the smoke detectors to different circuits at random.

I discovered a compounding factor. In the bedrooms with recessed ceiling lights, if the lights are off, my voltage detector doesn't read anything from the smoke detector wiring. If the lights are on, the NCVD detects current all three wires (hot, neutral, and interconnect). That might be the explanation for some of my confusion. However, my multimeter doesn't show any current flowing through the wires even when the lights are on (I double-checked it against a regular outlet to verify it's working), so maybe the NCVD is wrong.

5
  • Comments are not for extended discussion; this conversation has been moved to chat.
    – Michael Karas
    Sep 9 at 5:22
  • @MichaelKaras My original answer, which I had crossed out, turned out to be substantially correct. I've edited it to un-cross it out, but because you deleted it, it does not show up. Could you un-delete it?
    – Kevin
    Sep 9 at 5:50
  • Is your house wired for 120 or 230 V? Dryers in 120 V countries are often multi-phase; if cross wired with 120 V circuits, strange things could happen.
    – Yakk
    Sep 9 at 17:53
  • 1
    @DaniëlvandenBerg It's funny for people that get the joke, but honestly I think it's better not to spread that around because someone who didn't get the joke might actually try something like that...
    – Kevin
    Sep 10 at 19:12
  • @kevin good call, deleted. Sep 11 at 20:16
54

After writing out the question, I realized I had forgotten to check something. I went back to the panel to see if any other breakers besides the dryer circuit had flipped off, and sure enough, the "garage light" breaker had been tripped. After testing more, I determined that the smoke detectors are actually on the "garage light" circuit breaker, not the "dryer" circuit breaker.

So apparently in all of my going back and forth to the panel and testing different breakers, I somehow confused myself on which circuit the smoke detectors were on. False positives from my NCVD may have initially prevented me from realizing when I did have the correct breaker shut off, and definitely caused me further confusion after the initial accident.

Lessons learned:

  • Triple-check that the wires are still not hot before cutting them
  • Don't rely entirely on a non-contact voltage detector to tell if the line is hot or not. False positives can be nearly as problematic as false negatives, because you might not realize when you've shut off the correct circuit.
  • If at all possible, have a partner help determine which circuit the wire is on. Not only could this avoid a mistake, it's a lot faster!
  • Don't do wiring late in the evening after a hard day of mentally demanding work.
15
  • 25
    And don't cut two wires at the same time. Especially neutral and earth with gfci - the neutral has a voltage due to oth4er devices running...
    – Solar Mike
    Sep 7 at 9:28
  • 4
    @SolarMike I think OP experienced the best possible outcome to be honest. Your suggestion would have resulted in OP using their body as an arcing mechanism when they inevitably use their hand as a bridge for the hot and neutral. I think your suggestion is better suited for those brazen enough to purposely work on live circuits. OP's tools saved them where their wits failed them.
    – MonkeyZeus
    Sep 7 at 15:24
  • 14
    Great follow-up answer, but I don't think it's complete yet. Simple confusion explains why the circuit was still hot when the breaker was off (looking at the wrong breaker), but as it stands now, it doesn't explain why the circuit still seemed to respond to that other breaker as described in the original question. Maybe some bad wiring in addition to the confusion?
    – AaronD
    Sep 7 at 15:38
  • 13
    This might warrant its own question, but I sometimes will take the supposedly off hot wire and ground it against neutral or ground "just to be sure." I have a screwdriver with a bit of copper permanently welded to it from once when I got it wrong.
    – rrauenza
    Sep 7 at 15:41
  • 5
    "Triple-check that the wires are still not hot before cutting them" Once the circuit doesn't run anything anymore, the cap is off the box, the wires are pulled out, then a wirenut is removed: test the bared wire with a known working voltmeter to a known working ground or neutral wire before touching it. Test every wirenutted connection after the nut is off, since wire colors aren't always a solid clue. Sep 7 at 17:20
35

Could be an unmarked Multi-Wire Branch Circuit

We often look at people's panels for other reasons, and we often say "hey, see the red+black wires from the same cable, going to 2 independent breakers? Those need a handle-tie. So when you shut one off for maintenance, they both shut off". And people say "thanks" but think "why bother?"

Well, this. This is why.

A Multi-wire branch circuit (MWBC) is where 2 hot wires share the same neutral. (or 3 in 3-phase wiring). This works because they are on different phases -- neutral carries only difference current - if one leg is drawing 16A and the other 15A, neutral carries 1A, not 31A. (use of a handle-tie also helps enforce phase placement).

These don't require common trip for safety, so they don't need to be on a 2-pole (3-pole) circuit breaker. But they should be handle-tied so maintainers are forced to shut off the entire circuit.

The procedure you did - turning off breakers until "hot" current goes away - is normal and common. However, if a MWBC's two hot breakers are not handle-tied, then the other half (2/3) of the circuit is still hot! And there is current flow on neutral! If you short another hot to neutral or ground, kaboom! Or if you interrupt neutral as part of maintenance, you have series arcing when it parts, and the "downline end" of the neutral now has 120V on it. (the downline loads on the other phase are trying to return current there).

So that certainly could have caused what you describe. The cure is to search the panel for "shared neutral" situations (if wired in cable, the sure sign is a red wire on a 120V breaker) and make sure to fit those handle ties!

Regardless, treat everything as live

I treat all wires of all circuits as "live", even neutral. (I don't handle ground much since my world is EMT metal conduit; I get ground the moment it's built). This saved my bacon once.

I had shut off a circuit and started to work on it. I know this circuit like the back of my hand. Also, it's in surface mount conduit, plainly visible, and fed a large fluorescent light that was on. (fluorescent lights are in the greater family of arc-discharge lighting, ranging from Batman's searchlight beacon to neon to those orange street lights). I knew I had the right breaker because the light went off.

I was sure enough that I actually used a screwdriver to flash hot to the metal box. (which I would never do if there was any chance of it being on). I braced for an arc flash, turning my eyes upward/away. I got an arc flash right in the face, but not the one I was expecting.

The fluorescent light turned on, striking its arc in the normal way.

Whaaaaaaa???

So I did it again. Tiny blue arc on the hot wire, and the fluorescent light turned on.

No. There was no hot-neutral reverse. The breaker was off anyway.

Turns out it was a double failure: The panel neutral-ground bond had failed, AND, something on the other phase had suffered a bolted fault. Effectively that fault became the neutral-ground bond, or rather, the Hot L1-ground bond. With neutral being 120V to ground and L2 being 240V to ground.

So it turns out, electrical is like a box of chocolates: you never know which one will kill you.

8
  • 2
    I'm not sure yet if there's a MWBC or just a bunch of crossed wires in the attic. Will go through the comments in more detail tomorrow.
    – Kevin
    Sep 8 at 7:24
  • 1
    For further context, the "garage light" and "dryer" breakers are not adjacent and could not be handle-tied if they were in a MWBC.
    – Kevin
    Sep 8 at 18:29
  • Would you explain the kaboom! part better? Perhaps I misunderstood the scenario. Supposing a MWBC with just one wire de-energized, if I go and short that de-energized hot wire to the circuit's neutral "nothing" would happen. That hot is already connected to the neutral via whatever loads are wired between them; there's no voltage potential between the de-energized hot and the neutral. If I short the neutral to ground about half the neutral's current will divert to ground, but such fault might reach 10A and won't make a shower of sparks and burn a notch in pliers.
    – Greg Hill
    Sep 8 at 19:47
  • On the other hand if the neutral had been opened at a junction box while one MWBC leg is energized then one neutral wire has pull-up to 120v while the other neutral wire is grounded/real neutral. There's certainly a shock hazard possibly at several places in the circuit. But shorting that pulled-up "kind of hot" neutral to the de-energized hot will make a spark no bigger than the one that would have happened when the neutral was opened in the first place. Still not a sufficiently large fault to melt pliers..?
    – Greg Hill
    Sep 8 at 19:51
  • 2
    After eliminating all of the false positives I was getting from my NCVD, I finally got this worked out. It's not a MWBC, I just screwed up and mixed up the circuits originally. They're definitely only on the "garage light" circuit and not the "dryer" circuit.
    – Kevin
    Sep 9 at 5:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.