Grounding to telecom bar

[malachi constant]

Forgive my terms here, as grounding and bonding is not my area of expertise.

Situation is a new office building with 480V service, occupants moved in a few months ago. I'm the engineer of record on this project. Electrodes are properly bonded at service main ground bar, and at step down transformers (separately derived systems). Bonding conductors all appear properly sized and connected. There is a parallel telecommunications grounding system and the technology designer says it is as near as they can tell properly grounded per their TIA 607-based details.

The issue is: some AV equipment is having issues, and in troubleshooting the issues it was noted that when AV head-end equipment is plugged into a UPS power outlet (fed from a central UPS) it experiences the issue. (Something to do with devices out in the conference rooms rebooting, or not rebooting on command, or something like that - I don't know the details.) When the AV head end is plugged into a normal power outlet the issue goes away. The thinking is that the head end equipment within the data room is on one ground plane (the UPS plane), and the equipment out in the conference room is on another plane (the utility plane), and the resistance/voltage difference between them is causing the issue.

The main telecommunications room has a central UPS installed within it, 480V in 208V out, so is a separately derived system. We have observed that the UPS bonds to structural steel, and in the same room the telecommunications ground bar bonds to structural steel (the bonds are inches apart on the steel). There is not a direct connection between the UPS 208V panel ground bar, and the telecommunications ground bar in the same room. Further troubleshooting tells us although all normal panels/outlets seem to have a strong ground bond, the UPS outlets do not. Specifically there is 0.1 ohm measured between neutral and ground at all normal outlets, but at all UPS outlets that N-G measurement is over 1 ohm.

The technology designer recommends bonding the telecom ground bar to the ground bar within the UPS power panel, per TIA 607 6.5.1. "When an Electrical Distribution Panel (EDP) is in the same room as the TMGB/TGB that EDP's equipment grounding bus or the panelboard enclosure shall be bonded to the TMGB/TGB using bonding conductors sized..." In my limited past history with grounding and bonding, my understanding of the theory was you earthed the heck out of the system at the main ground, and then everything tied back into that main ground bar more or less in series. Like you bond the outlet to the ground bus in the panel, and the ground bus in the panel is bonded back to the main - but you avoid tying that panel bus to anything else (such as other panel busses) so as to avoid ground loops or something. So I am wary of tying panel ground busses into the technology ground bus. But even more so I am guessing my understanding of grounding code/theory is flawed and this is the right thing to do. Any insight / questions / concerns the community has is appreciated.

 

[zbang]

The ground path(s) seem like a red herring to me, I'd be really looking at the UPS output first*- voltage, it's own grounding, etc (even frequency but that probably doesn't matter much).

*maybe 10 years ago, I found a place where the installing electrician had wired a single phase center-tapped UPS output as 208v, not 240v, so one side was 120v to the neutral and the other was 87(!!!); some equipment would operate on that, some wouldn't. There's a post about this in the distant past.

 

[suemarkp]

You need a meter that can read into the milliohm range to check bonding. Something doesn't seem right in the UPS if there is 1 ohm between neutral and ground. Did someone use isolated ground receptacles or wire it like that and not bond it properly? I would use that same meter to check the bonding between the UPS ground and the UPS panelboard equipment ground and the telecom ground. These should all be under 100 milliohms (0.1 ohm). You can't measure down to this level with a typical meter. It might read 00.0 or 00.1 ohms, but most meters have 3 digits of slop in their last digit for accuracy. You need something that at least reads to the 10s of milliohms or under. You also need to compensate for lead length if they need to be long to reach between these things. Most bonding meters use Kelvin clips or have a lead resistance zeroing function.

We always had problem with video or RS232 between buildings because the grounds were only referenced through the earth. Video and RS232 use unbalanced lines where the shield is basically in parallel with the ground path. Stray voltages traveling along this can mess things up. Between buildings, we would convert video or RS232 to fiber since the grounds were difficult to connect.

It is also possible that the video system doesn't like the waveform from the UPS. If you could temporarily connect all items of the video system to only the UPS and not involve any other ground paths or power systems, that would be one way to eliminate that as a problem.