> On Aug 26, 2015, at 4:07 PM, Noel Chiappa <j...@mercury.lcs.mit.edu> wrote:
>
>> From: Paul Koning
>
>> What happened is that the "grounds" were offset enough, and with enough
>> of a current supply, that the ground strap that's supposed to connect
>> the row of RP06 drives melted.
>> This sort of thing is a major electric code violation: you can certainly have
>> multiple services, but all the grounds are required to be connected by
>> substantial wire; you're not allowed to stick ground rods in at
>> multiple places and leave it at that.
>
> I'm pretty blown away that the various grounds could be offset by that much,
> to produce that kind of current when they were tied together. Wow.
I can't vouch for the truth of the story; I heard it a long time ago from a
fairly reliable source.
But consider this theory. Suppose you have two service drops, fed from
transformers off the utility high voltage line. The neutral is simply defined
by the ground rod at the transformer and at the service entry to the building.
If the building is a steel frame and all service entries are bonded to the
steel, and the steel is generally conductive, you have a single neutral. But
if some aren't bonded, or the building isn't conductive, then you have two
separate ground references. Also, the green wire (protective ground) is
connected to the neutral at the service entry.
Now suppose that you have unbalanced phases, which will generally be the case.
That produces a neutral current, which dissipates through the ground rods. If
you have two services, the resulting neutral voltages will not be in phase.
If you now tie these two neutrals (grounds) through an unplanned wire, as in
this story, the phase unbalance voltage (the difference between the two
neutrals) will produce a current that's split between that wire, and the ground
around the building, in proportion to the impedance of the two paths. The wire
impedance is likely to be far lower than that of the ground (especially in New
England), and the phase unbalance current in a large building might well amount
to a lot of amps.
paul
