Originally posted by Kevin-MI:
We have a very large customer that is a financial institution. They have set standards for the data racks at their bank branchs. Their standards on grounding/bonding is extensive. They have us bond a TMGB to the main distribution panel. Depending how far away the MDP is away determines the wire size. In some banks this has been 2/0 wire. Then we also have to run a wire from the TMGB to building steel. Also length determining the wire size. It's a least a #6. Then we bond from the TMGB on the backboard (thats usually mounted behind the rack) to the TGB on the data rack. Also the ladder tray is bonded to the TMGB. Then we provide a ground wire from the TGB to their Cisco router in the rack. We have to use 2 hole compression lugs for every termination. We have probably done this at maybe 1000 banks. Seems overkill, but its their standards and they are paying.
Actually it is not their standard. As it follows TIA 607-B (grounding and bonding data and telecom equipment) to the letter. And that's a good thing, IMO, as both customer & installer have an objective frame of reference. But the OP mentioned only a single rack. Not the complicated infrastructure (component->rack->telecom closet->equipment room) that would necessitate structured/hierarchical bonding/grounding as described in the standard. Within a single rack, a #6 copper (or maybe #4 aluminum-as per your code) would be more than enough, as it would handle close to 200A (I don't remember the exact figure offhand).
Don't forget this is a performance or to use the proper term technical ground. Its main function is to protect equipment against electrical noise. There has been loads of research on this, as these problems are the hardest to pin down. And yes, paint can act as an insulator. All bonding surfaces should be bare. Another reason is that bonding/grounding effectivesness is related to the geometry of apparatus such as the TelecomMainGroundBar. IE thickness, dimensions, number & size of terminals, material of construction, AND contact-surface area. If you use a certain size ground lug/screw it is understood that its bonding surface (however many square millimeters it is) should be the one expected as per the standard. I agree that maybe this is all too much. On the other hand, if the budget is there, they're pretty easy to implement, and as I've said above the standard does provide with an objective reference.
