GFCI vs. bonded neutral in subpanel

[CodeQuandary]

Here's a funny one.
Existing subpanel, single family residence. Run at 240V 30A with older (but apparently fine) 3-wire no ground MC cable. No additional ground lead in the cable.
The ground and neutral are tied together in the panel.
This is a no-no.
In the absence of being able to run a new ground (long difficult run: the client probably won't do it), it would be possible to put the entire subpanel on a GFCI back at the Main.
a. This is actually arguably safer than running a ground and not adding a GFCI to the circuit, I think... (change my mind!)
b. The only way this works is to disconnect the neutral from the box & all the grounds, I believe. Given old house wiring: then (I think) connect all grounds to the box, in case any of them (or the cable) get back to a real ground.
c. An unpredictable downside is nuisance trips, but the house got a whole bunch of GFCI's and AFCIs during some recent work, and doesn't seem to be having any problem with nuisance trips.

Not ideal, but (as inferred above) probably safer than a proper ground but no GFCI. Definitely better than as-is.
Another idea would be to put the (6) circuits each on their own GFCI, but I think that might be less safe than more (the box and cable are no longer protected) and it's definitely more expensive.
Other thoughts?

 

[Dennis Alwon]

I would leave it as it is. Many people don't think it is all that dangerous

 

[hillbilly1]

Where it becomes dangerous, it is another one of the "what if" scenarios. If for some reason the neutral is broken, everything bonded to it, becomes "hot" to ground, posing a shock or electrocution hazard. Likely hood of the neutral being broken is low, but still a possibility.

 

[jaggedben]

I think Dennis has a point. The neutral may be ensuring an effective ground fault path, which is a code requirement. If forced to choose between violating that code requirement, vs. others, I'd probably choose others.

 

[CodeQuandary]

Where it becomes dangerous, it is another one of the "what if" scenarios. If for some reason the neutral is broken, everything bonded to it, becomes "hot" to ground, posing a shock or electrocution hazard. Likely hood of the neutral being broken is low, but still a possibility.

Just to be clear (if obvious): if the neutral is broken, the GFCI trips immediately and demands repair. That's almost the perfect case for when this is helpful. That may have been your point.
Also, to Jaggedben:
Technically, when there is a load on a circuit, neutral is a current-carrying conductor. I don't think it counts as a ground fault path: am I wrong on this point?

 

[drcampbell]

First question: What does this subpanel serve? We need to know that before we can make intelligent, applicable suggestions.

If it's a garage, barn, et c. ... Putting the entire panel, including the lights, on a remote GFCI reduces the risk of a shock but increases the risk of tripping over something in the dark and breaking your nose. Plus the added nuisance of schlepping back to the house to reset it.
Maybe the answer is to put the convenience receptacles on individual GFCIs, but not the lights.

Which is not to suggest that the ehtire subpanel can't also be protected with a GFCI back at the main panel.

If it's a pool, boathouse, wharf ...

 

[hillbilly1]

Just to be clear (if obvious): if the neutral is broken, the GFCI trips immediately and demands repair. That's almost the perfect case for when this is helpful. That may have been your point.
Also, to Jaggedben:
Technically, when there is a load on a circuit, neutral is a current-carrying conductor. I don't think it counts as a ground fault path: am I wrong on this point?

If left connected as is, it will be the ground fault path, as long as it is still complete back to the source. If for some reason the neutral is broken, then everything bonded to it becomes hot.

 

[LarryFine]

Technically, when there is a load on a circuit, neutral is a current-carrying conductor. I don't think it counts as a ground fault path: am I wrong on this point?

You are, in a way. The service neutral performs that very function, as do neutrals during L-N faults.

 

[Greentagger]

If it’s three wire then case should be bonded.???To ensure panel is grounded.

 

[winnie]

Keep in mind that the spiral metal sheath of the old MC is not considered an effective ground fault current path, but _is_ a metallic connection from one end to the other, and I believe at one time was considered a proper EGC.

Perhaps when this was installed the sheath was considered the EGC, and then ground and neutral connected at a later date.

-Jon

 

[jaggedben]

...
Also, to Jaggedben:
Technically, when there is a load on a circuit, neutral is a current-carrying conductor. I don't think it counts as a ground fault path: am I wrong on this point?

The neutral is most definitely the ultimate path back to the source, and being the grounded conductor it is the ground-fault path from wherever ground-fault current joins it. That point is supposed to be a main bonding jumper or system bonding jumper. But in your case it would be the groundbar at the subpanel. Unbonding the metal parts from the neutral at that point leaves you without an effective groundfault path, unless there is another one existing. The MC cable sheath may or may not do the job, see 250.118(10) and/or do some testing on the site.

If you don't have an effective ground-fault path then a hot-to-ground fault (or rather, it will be a hot-to-ungrounded-metal parts fault) will energize the panel and other parts at line-voltage-to-ground. Without GFCI this is most certainly a much more dangerous situation than having objectionable current on the MC cable sheath or just violating the letter of the code on bonding locations. With GFCI I think it is arguable what is worse. But regardless, your GFCI stuff won't detect the fault unless some current actually finds another path back to the source neutral, and that path could end up being a human.

 

[CodeQuandary]

Hi Jaggedben,
I agree completely that in the absence of a GFCI, it's best to leave well enough alone.
And yes, GFCI only trips with stray current. But of course, that's exactly what it's there for: in the same situation, with just a ground, that current just keeps on cranking through the hypothetical human, unless it manages to trip the OCPD (that doesn't bode well).
I think I've come to the conclusion that a GFCI is the safer route. But only if you disconnect grounds from neutrals at the subpanel, whether or not there's a good ground-fault path. If the MC cable provides a decent ground, then that's a very significant bonus of course: it almost approaches code compliant!
Of course, the scenario is illustrating the imperfect situations that older existing installations often present.
I hope people don't feel like I'm ignoring your input, this was helpful. I just don't think I've heard a clear argument that persuades me away from the GFCI.

 

[curt swartz]

Are you sure this is MC cable and not AC cable? I have never seen MC that didn't have either a insulated ground or the new option that has a 10 bare AL in contact with the sheath. If its AC with the bond strip integrated in the sheathing that is a good EGC.

Either way you have a metal cable. Bonding the neutral to ground at the load end of the circuit put the sheath in parallel with the neutral which in not good.

 

[winnie]

Keep in mind that the spiral metal sheath of the old MC is not considered an effective ground fault current path, but _is_ a metallic connection from one end to the other, and I believe at one time was considered a proper EGC.

Whoops, Curt is quite correct. It is the old _AC_ cable where the sheath without the bonding strip is no longer considered an acceptable EGC.

-Jon

 

[CodeQuandary]

Ah, thanks. I didn't remember the details of cable nomenclature offhand: I think it is probably old AC then. There is no additional grounding element. I was there today, and separated the grounds/added a bonded ground bar, and confirmed that the box is at least nominally grounded: If I check between the (new) ground bar and the (mains) neutral, there is continuity between the neutral. The neutral bar turns out to not have a bonding jumper (it is isolated from the box). Equally, there is no continuity between any of the grounds and any of the branch circuit neutrals. I realized there are only 5 circuits (one unused breaker), and 2 of them are also in old AC without any sign of ground wires/tape/anything.
So yes, I think this is old armored cable without any additional ground provision, no longer a legal grounding system. However, the grounded state of the box indicates a nominal ground fault path, at least. I'm going to pick up a GFCI for it in the next day or two, and try to think what I can plug in that will best test if I should expect nuisance trips... I suppose I've never thought carefully about why nuisance trips happen: it can't be an all-out cross-connection between neutral & ground anywhere (or it would constantly trip), so... huh. Of course, it seems to me there's a lot less nuisance tripping of GFCIs vs. AFCIs, which are doing a vastly more complicated thing.
Thanks as always for helping me think & learn!