Why is the neutral isolated in a sub box.

It keeps current from flowing in the ground (thus making it /ground). Note that you don't ground a sub-panel to earth directly either (it's grounded at the entrance panel).

If this is a detatched garage and there are no other metalic paths between them you can reground the neutral. Otherwise you would pull a 4 wire feeder and separate the neutral in the garage. In either case you will need another grounding electrode in the detached building, typically a driven rod, that connects to the grounding bus.

It is still code compliant to run a three wire feeder to a detached garage that has no metallic pathways to other buildings in areas governed by the US NEC but it is not good practice. If you use a three wire feeder to the panel in your garage and a high resistance or open condition develops on the neutral of that feeder than all of the exposed metallic surfaces of the garage wiring system will be at line voltage potential.

Regardless of which feeder is run the US NEC requires that a grounding electrode system consisting of at least one driven rod be built at the detached structure. To avoid the cost of testing a single rod most electricians will drive two rods. The grounding electrode system is for protection against lightning and accidental contact between the service wires and high voltage distribution wires it does not play any major role in clearing normal operating faults.

-- Tom

at least 6 feet apart.

NEC VIOLATION, you have just created a ground loop,

Otherwise you would pull a 4 wire feeder

this is the right way and the only correct method of doing a subpanel.

The NEC allows for a supplemental ground, your ground rod, as long as the grounding conductor is sized correctly for the service, not the feeders for the subpanel.

Any connection to another ground source without the proper conductors between the grounds is a ground loop,

Strangely enough, all the FAA TOWER and TRACON installations that I am familiar with isolate the GROUND, not the NEUTRAL in a sub-box. (Manchester, NH, Burlington, VT, and Boston, MA as examples.)

I suspect that it serves the same purpose that Keith explained, only executed in a different way.

Harry C.

Since 2 of us quoted virtually the identical language from NFPA 70 2002 article

250.32 (B) (2) I think we are right. How did you read that section? If you want to go back to 1996 where it was in 250-24 you will not even see the "no continuous metalic paths" language. In either case you still need a grounding electrode at the remote building.

Isolated grounds are very common in the communication industry. It is a further attempt to make sure that there is a low ohmic path to ground. I have seen many applications wanting that hated word a "clean" ground. If a ground system is properly installed "clean" is not an issue. The biggest thing I have seen recently is the push towards zero ohm resistance in the ground systems. I have designed and installed systems that maintain ~ 5 ohms. Less is really difficult here in the deserts of the SW.

Recently I saw a new house that they are using the rebar as their grounding electrode. Allowed, by the NEC but a ground rod clamp on the rebar and a piece of copper to the panel. UG not for me. I called a supervisor in the electrical department for the city in question and we had a discussion. Boiled down to not his department and nothing he could do about it. Cheap cheap cheap

I recommend SOARS book on grounding to everyone. I just got my new one for $35. Makes a complex subject clear. A new version is out for the new code which some municipalities are adopting

The Ufer ground is one of the best grounding electrodes you can get. Concrete is usually a lot more conductive than the dirt you are grounding to and in the footer of a house, slab and garage floor, you end up with plenty of ground contact. If nothing else you have created a ground plane that covers your whole house footprint. It might not be the same as your neighbor's but so what?

In the 96 Code, the N-G bond was optional at the remote building if there was an equipment grounding conductor run with the feeder. The handbook shows it dashed, and the language is "shall not be required", implying that there was nothing wrong with doing it if you wanted to. In 2002, it is clearly not allowed if there is any conductive connection between the buildings' grounding systems.

Ben Miller

Not really.

That is exactly the situation when the neutral from the power company opens. It is a PITA (because some of your house gets over voltage and some get lower voltage) but it doesn't present a safety hazard since your end of the neutral is still bonded to your local grounding rod.

Even if one HOT and the NEUTRAL break AND the remaining HOT touches the metal equipment in the garage you will have a situation where in the distance between garage and house (or, in the utility case, between your house and your neighbors house) there is a total of 120 volt drop.

BFD! At most a few worms might make it to the surface.

But what about you when you touch the box? If your standing on the concrete floor, and the box is 120V, don't you get hurt? With no grounding rod in the box, nor a fourth conductor back to the main service, seems like you could get a shock by touching the box?

Or is the box still require a grounding rod even when you only run three-wires from the main service? Maybe I missed that.

daestrom

A ground is required if the Garage is more than 25' from the house. However, that ground rod won't help you in you senario. It won't pull that

120v fault down to a safe level; the impedance is too high at that voltage to pass enough current to trip the OCD suppling the panel. The ground is rod really there to provide a ground path for high voltage such as from a lighting strike or transformer primary to secondary fault.

"Nukie Poo @verizon.net>"

--------- If the ground rod is insufficient under the currents produced by a 120V fault, then how will it provide safe touch and step potentials in the case of lightning which will inject much higher currents (and current -not voltage is the operative term)? I recognise the non-linearity of the ground-rod connection but even if the resistance is an order of magnitude lower for lightning, the current is likely to be more than an order of magnitude larger. Without a good handle on the nonlinearity in a particular case, we cannot make any definitive statements.

-- Don Kelly snipped-for-privacy@peeshaw.ca remove the urine to answer

I never said it would "...provide safe touch and step potentials in the case of lightning which will inject much higher currents..."I was simply stating that the rod would not pull down a 120v line to ground fault to a safe level and its purpose is provide a ground path for high voltage such as from a lighting strike or transformer primary to secondary fault - not pull that down to a safe level. Read the post!

It can't! That would require micro-Ohm grounding electrode resistance with several hundred thousand amps flowing. That is not the purpose of the grounding electrode. It will help to reduce the voltage magnitude in terms of insulator breakdown, etc.

One factor is that the paths are different. For the 120 volt fault, current is returning to the grounded side of the transformer, which could be several hundred feet away. The resistance of this path could be a few hundred ohms through the earth. This is why a metallic path is required, and most fault current will flow through the grounded conductor, not the earth.

For lightning current trying to reach the earth, the theory is that it happens fairly close to the electrode. According to IEEE std 142, half of the rod to earth resistance occurs within approx 0.5 feet of the rod. A

1-foot radius contains 68% of the total resistance, and a 5-foot radius contains 86%. The 94% point is 10 feet away, and the 99% point is 20 feet away (all are approximate). Step potentials will be severe within a few feet of the electrode. Touch potential from earth to objects bonded to the electrode will be worse the further away from the electrode you go.

The hope is that someone inside a properly bonded building only sees grounding electrode potentials, and therefore no significant touch or step potential, even though the entire system has risen in potential with respect to the "earth". In improperly bonded systems, or where illegal multiple grounding electrodes without a common bond are used, horizontal arcs of 5 to

6 feet have been reported between the two "ground" systems!

Ben Miller

So let me get this all straight....

1) An outlying building such as a garage, no other metal connection to main building.

2) three-wire (two 'hot' and neutral) circuit to sub-panel in garage.

3) More than 25', so by Nukie-Poo, I would need a separate ground rod(s) for the sub-panel, and the neutral can *not* be bonded to the grounding rod/conductor in the sub-panel.

4) So an 'open-neutral, hot-shorted to casing' scenario'.... a) would not necessarily trip breaker since the ground to earth may not conduct enough back to main service ground b) But may not be that dangerous since my feet (on garage floor) would be at nearly the same potential as the 'casing'???

5) Think I'll wire the outlying building with GFCI's anyway. (They are required in 'garage', but not so sure about other outlying buildings (such as a workshop). Or maybe a fourth conductor for the ground rods to bond back to main service panel? Or perhaps both.

daestrom

wrong, if you do go the 3 wire route, no metalic paths etc you DO reground the neutral and you ALWAYS put a ground electrode in a separate building. (regrounded 3 wire or separate 4 wire) I don''t have a clue where Nukie got the 25'. Certainly not in the NEC. The only exception to all of this is a single branch circuit feeding that second building. It is what I think of as the "extension cord" exception where the installer replaces the orange cord with some more substantial wiring method. If it gets any more involved than that he must treat it like another sub panel.

Ah.... THIS is the part I missed somewhere along the thread. So a four-wire, they (neutral & ground) are kept separated in the subpanel, but a three-wire (and no metalic pathways), you ground the neutral *again* in the sub panel with suitable grounding rod(s).

Thanks for clearing that up.

daestrom