ECN Forum Forums Code Related Discussion NEC & other Code issues IT room grounding

How in this context, would one interpet this code for size requirement, and point of connection? i.e. To nearest equipment ground, MPOE (TelCo), Main service, and what size?

BTW this is for Ladder rack and and Equipment rack grounding, served by 70A UPS

Any comments would be helpful

TANKS Mark

Mark:
IMHO, you cannot go wrong with installing a bond jumper (rated at 70 amp) from rack to ladder frame, and to bldg steel, or h2o piping.

Another 'gray' area???

John

I agree with John. The general rule on grounding and bonding is you size the grounding conductor based on the circuit that is likely to energize the protected equipment. In your case I would say it was the 70a UPS and 250.122 says that is a #8 copper or #6 al. This should be bonded to the EGC of the UPS supply conductors or back to the panel grounding bus, in my opinion.
I suppose you could go to building steel if you can assure that it is effectively bonded back to the grounding system but a water pipe is not usually an acceptable ground these days. Too much plastic in the plumber's trade.

[This message has been edited by gfretwell (edited 11-11-2004).]

OK the reason I asked....

LV sub doing the install of the ladder rack asked for 1/0 AWG???? To the main!!!!

An engineering note mentions 4/0! No location specified.

Now some other info that I left out, this is an it room covering an tenent buildout area fed by two 200A panels.

I can be one to follow spec's, but when I get conflicting ones I get itchy. Then again, I like to know why things are they way they ask. I have put these in several different ways now, as spec's dictate.

So what dictates the size of this ground, and where it comes from. Personally I think a #6 equipment ground would be fine, or for that matter to the MPOE ground.

There is lots of flexibility in the size (as long as you exceed the minimum) and exactly where it terminates as permitted by Article 250.
The Engineer may have a reason he/she wants 4/0AWG, don't reduce it without a nice credit

If this guy is going for reducing transients on the ladder rack I can see wanting to go back to the grounding electrode (star wired ground system) but I think the wire size may be excessive. These days noise rejection is a lot better in IT equipment than it was in the 70s but a lot of people have not adjusted their thinking.
There is more urban legend in IT than any other industry I can think of. A lot of these <overkill> grounding practices have simply taken on a life of their own and no amount of logic will get people to budge.

 
One origin of giant grounding conductors is the telecomm biz.

At least the blind-copycat boilerplate does not call for 4/0AWG seven-strand or does it? Another good one is exothermic welds to aluminum {or fiber-glass!} cable tray.

Two general references for power—signal ground ‘interfacing’ are IEEE emerald book described at http://shop.ieee.org/ieeestore/Product.aspx?product_no=SS94741 and the parchment scrolls of FIPS PUB 94 www.itl.nist.gov/fipspubs/fip94.pdf




[This message has been edited by Bjarney (edited 11-11-2004).]

In the "old days", there was a lot of RS-232 and similar stuff running around. RS-232 connections are direct electrical connections between the equipment, using the equipment ground as a voltage reference. A small difference in the ground voltage can kill equipment like this, and it doesn't take much noise in the grounding system to cause data errors.

More recent stuff, like Ethernet, uses differential, transformer-isolated connections. These transformers are rated to withstand a 2,000 volt potential difference between the communication line and the equipment. Because the connection is differential, it has something like a 60-90 dB rejection of common-mode voltage. It's really hard for a grounding problem to cause a problem in the equipment. And even if really severe noise on the grounds causes occasional data errors, most modern communication protocols have error detection and recovery mechanisms built in, so errors normally just result in a minor degradation in throughput.

Interesting historical note: When Xerox installed their first prototype Ethernet network, they found that there was something like a 90V difference in ground potential from one end of the building to the other! That affected our thinking about isolation, so when we designed the 10 megabit Ethernet standard, we made sure that everything was transformer isolated.


[This message has been edited by SolarPowered (edited 11-12-2004).]

I guess you never worked in Florida ;-)

I agree these LANs will do pretty well with huge DC and 60hz ground shifts but when you get short rise time transients like you see with nearby lightning strikes the only way to fix it is with copper. If we ran between different buildings or services with separate grounding systems we bonded them together. In some cases we even bonded individual machines together.
These are really those dreaded "ground loops" but as you say, these things reject that noise on the data side. This is actually a power hit that fries transistors. It may just be a regional thing tho. I don't remember this kind of thing happening much in Md

Hey, we have lightning storms here, too! I think I remember one sometime in the last three or four years.