In the computer biz they specify that the IG conductors will originate at the ground busbar where the Grounding Electrode Conductor lands in the service disconnect enclosure and that it radiates to the equipment in a tree fashion from there with no parallel paths. It will be isolated in any intervening distribution panels. Since we never brought the neutral to the computer panels they would use the insulated neutral busbar for the IG. Green reidentification of all of the conductors made it apparent this was a grounding bus. Sometimes they were also labelled that way. Usually we also had a bare wire ground bonding conductor to the enclosures too.
We used to install isolated grounds alot in commercial fit ups but for the past couple of years I haven't seen them much. Now everybody specs seperate neutrals to all receptacles, no multi wire circuits. We are currently doing a building in which 7 floors are for Adobe (pc company) and they want seperate neutrals for all circuits. We had to put in big panels and conduit to accomodate all the wires! I've heard of isolated grounds causing problems like getting a potential from regular ground to isolated ground. Anyone ever heard of anything like this?
Eddy, I have not heard of that one. Then again, that doesn't surprise me.
The whole "isolated ground" theme is filled with all sorts of mystical theories and majic potions. The 'geek' types seem particularily imaginative in their explanations.
Personally, what I think is going on is this: Advances in technology allow us to notice things we never did, and bad practices are catching up with us.
IBM seemed to lead the pack with the 'isolated ground' nonsense. Since then, they've backed off considerably. I think what is needed, mainly by electronic components, is a good, low impedance ground path. Properly made up pipe ought to be sufficient.
In the real world, though, fittings are loose, locknuts are loose, etc. So, in the end, running a green wire straight from a critical component straight to the panel can be cheap 'insurance.' If nothing else, you've eliminated one cause of problems.
The same thing goes for separate neutrals. In a flash, you completely eliminate one cause of power quality problems (loose neutrals), and counter one effect of harmonic problems (excess current on the neutral).
Likewise, the use of steel pipe has been shown to eliminate EMF issues.
I realise that a lot of these problems would also be eliminated if the related trades did their jobs correctly; data problems can be quite interesting when the cables are draped over the fluorescent lights!
Ironically, though, most of these "new fixes" are nothing more than a return to older, prooven methods... and avoidance of clever little "handyman short-cuts."
IBM originally spec'ed IG for the big computer room panels. In reality our machines were the cause of most of the noise but they were in "I'd pee on a sparkplug to fix this" mode. Once they really started doing the engineering, sometime in the late 70s - early 80s, they figured out IG really wasn't fixing anything and it was removed from the Physical Planning manuals. Unfortunately old habits die hard, as do old legends and there are still people spinning some of the yarns that started in the 60s-70s about ground loops and "noise". Everyone seems to know a guy who knew a guy who fixed some mystery problem by installing IG. I bet it was really just getting a decent ground in there in the first place, which could have been done using standard 250.12x methods. When I got involved with the lightning protection end of physical planning we found ourselves bonding things six ways from Sunday and it had absolutely no negative effect on signal reliability, much to the chagrin of those "ground loop" nazis.
Anecdotal, but I've never spoken to anyone who told me about a problem definitely being solved by an IG. What I have heard a lot of was that when equipment is plugged into seperate IG circuits and connected with non-isolated data-com lines, there can be circulating ground current. Or in the event of a minor fault, the parallel path presented by the data-com lines can actually cause fires.
I can see this potentially happening with reqular equipment grounds, too, but I don't wonder if the main problem is IGs are relatively higher impedance.
My opinion today is that they're smoke-'n'-mirrors.
We use isolated ground on some sensitive serial data systems in critical applications, where a small difference in ground potential between two pieces of equiment would lead to problems. (Such a small difference in potential could easily be caused by a fault an unrelated system) So, we had a completely isolated system ground from the standard chassis ground. This is rapidly becoming obsolete, though, as new equipment is far less sensitive.
All the equipment had two grounds- one for electrical reference, and a 2nd for safety. The reference ground was only actually grounded at one point.
[This message has been edited by SteveFehr (edited 10-16-2006).]