comm system bonding

[copper chopper]

in a new medical building there are IT rooms on all 4 floors, can I run a wire to the nearest point in each room to building steel for bonding the IT equipment?

 

[__dan]

in a new medical building there are IT rooms on all 4 floors, can I run a wire to the nearest point in each room to building steel for bonding the IT equipment?

I see it as a design philosophy question. Off the top of my head I might be looking to run a 3/0 common riser to the IT areas and add irreversible (hydraulic crimp) taps to either a common grounding busbar in each IT space or taps to each piece of grounded equipment. One end of the 3/0 would land on an earth grounding electrode (something chosen to be both a good ground reference point and clean of circulating currents that may be present elsewhere on the grounding system). The 3/0 would have to be bonded to the earth ground reference electrode and the source panelboard or source transformer EGC common busbar, then taps to the equipment should be beyond reproach.

Once you have a dedicated common 3/0 landed on the grounding electrode and the source(s) EGC busbar, you can bond that to that local building steel, racks, rack mounted equipment.

There are multiple objectives to satisfy:
IT equipment manufacturer's requirements for connection to "Earth" grounding in addition to providing an EGC.
Bonding interconnection of the grounding of multiple seperately derived source systems, all sources referenced to the same ground level.
Providing a clean ground reference when used for signalling or providing a dedicated path for noise that may otherwise travel on the signal paths.
Making the extra effort to comply with the NEC so if trouble does happen, at least the install meets code.

I could look for the code reference but it should say something like 'the frame of the building is not to be used as a grounding conductor'.

It's a good bet the extra wiring is not on the drawings and a bit above the NEC minimums (if minimum was the objective). I would probably see it as an adder.

Looping the 3/0 may be a bit better than a linear end to end run when the arrangement provides the opportunity to do so.

 

[suemarkp]

Is there any EGC that doesn't get you to earth at some point? Would think that would be fine for dissipating static and meeting a manufacturer rule that requires a connection to earth.

Any unbalanced signal that connects from one computer to another (e.g. USB, RS232, etc) should have all computers as close as possible and preferably powered from the same source transformer and panelboard. If there are sub supplies in each rack, you may want to bond them together, but closeness solves a lot of problems (the EGS feeding each sub supply should be bonded to the output side so all of these should be at the same "ground" potential). Ethernet signals are balanced and don't care about grounds. For anything long, fiber or an unbalanced/balanced converter are better solutions than trying to get the grounds the same.

Racks should have a bonding stud. This one is the most difficult unless the you have a lug or screw hole on a PDU in the rack to connect to. Otherwise, you need some lone wire going back to a panel or conduit, or other item you know is bonded to the power source. If the rack needs to move, this is a pain... I really hate those lugs, and prefer to bond the power source to the rack rails with special bonding screws that are thread forming and bit through the paint of the rack mount item. Then I'll tie the bonded rail to the rack bonding screw.

 

[__dan]

Is there any EGC that doesn't get you to earth at some point? Would think that would be fine for dissipating static and meeting a manufacturer rule that requires a connection to earth.

Any unbalanced signal that connects from one computer to another (e.g. USB, RS232, etc) should have all computers as close as possible and preferably powered from the same source transformer and panelboard. If there are sub supplies in each rack, you may want to bond them together, but closeness solves a lot of problems (the EGS feeding each sub supply should be bonded to the output side so all of these should be at the same "ground" potential). Ethernet signals are balanced and don't care about grounds. For anything long, fiber or an unbalanced/balanced converter are better solutions than trying to get the grounds the same.

Racks should have a bonding stud. This one is the most difficult unless the you have a lug or screw hole on a PDU in the rack to connect to. Otherwise, you need some lone wire going back to a panel or conduit, or other item you know is bonded to the power source. If the rack needs to move, this is a pain... I really hate those lugs, and prefer to bond the power source to the rack rails with special bonding screws that are thread forming and bit through the paint of the rack mount item. Then I'll tie the bonded rail to the rack bonding screw.

The EGC's are typically noisy and can be expected to be noisy, especially in a large facility like a hospital or data center. Amp clamping large feeder EGC's can show a good sized number for current flow under normal conditions in large facilities, both normally and from things that are defective or misconnected. If you amp clamp the feeder EGC and see a current flow on it, 250.6 applies, the arrangement must prevent objectionable current flow. It is not an Earth referenced connection if it is noisy, as may be required by the IT manufacturer or for the equipment to function normally.

EGC's and GEC's have two different purposes. During a fault the EGC has the fault current on it and transient noise. Large UPS's can have the capacitor filter sections referenced to ground and that connection can cause a reactive current flow on the EGC path as the reactive component is not perfectly balanced. Those problems can be compounded by distance between multiple separately derived systems referenced to different Earth grounding points.

If that grounding path noise scrambles some data, causes a server reboot / glitch, causes the network / processors to default to a lower speed or perform erratically, that effect can be both very costly operation wise and very difficult to track down and mitigate. If the problems can be forseen and mitigated at the design and construction stage, the cost is reasonable and low.

What I tried to convey would be required under 250.30. Yes there is an exception for typical rack mounted power supplies, but the exception creates a minimum. Rather than required, it would be reasonable to build in some redundancy or durability that would otherwise be required by 250.30 A 3 or 4. Grounding to the local steel frame will not satisfy 250.118. You can bond to the frame but it is not a grounding conductor.

 

[kwired]

The EGC's are typically noisy and can be expected to be noisy, especially in a large facility like a hospital or data center. Amp clamping large feeder EGC's can show a good sized number for current flow under normal conditions in large facilities, both normally and from things that are defective or misconnected. If you amp clamp the feeder EGC and see a current flow on it, 250.6 applies, the arrangement must prevent objectionable current flow. It is not an Earth referenced connection if it is noisy, as may be required by the IT manufacturer or for the equipment to function normally.

EGC's and GEC's have two different purposes. During a fault the EGC has the fault current on it and transient noise. Large UPS's can have the capacitor filter sections referenced to ground and that connection can cause a reactive current flow on the EGC path as the reactive component is not perfectly balanced. Those problems can be compounded by distance between multiple separately derived systems referenced to different Earth grounding points.

If that grounding path noise scrambles some data, causes a server reboot / glitch, causes the network / processors to default to a lower speed or perform erratically, that effect can be both very costly operation wise and very difficult to track down and mitigate. If the problems can be forseen and mitigated at the design and construction stage, the cost is reasonable and low.

What I tried to convey would be required under 250.30. Yes there is an exception for typical rack mounted power supplies, but the exception creates a minimum. Rather than required, it would be reasonable to build in some redundancy or durability that would otherwise be required by 250.30 A 3 or 4. Grounding to the local steel frame will not satisfy 250.118. You can bond to the frame but it is not a grounding conductor.

Unless you have a true isolated equipment grounding conductor why should the one in question be expected to not have any of the same "noise" on it? The true isolated grounding conductor must still be connected to the equipment grounding conductor at the supply end, but can bypass bonding it multiple times to raceways and enclosures which theoretically reduces some 'noise'.

The regular equipment grounding conductor may have (and usually does have) multiple places of contact with building steel, or earthed components or metal raceways and enclosures that do have contact with same building steel or other earthed components.

 

[__dan]

Unless you have a true isolated equipment grounding conductor why should the one in question be expected to not have any of the same "noise" on it? The true isolated grounding conductor must still be connected to the equipment grounding conductor at the supply end, but can bypass bonding it multiple times to raceways and enclosures which theoretically reduces some 'noise'.

The regular equipment grounding conductor may have (and usually does have) multiple places of contact with building steel, or earthed components or metal raceways and enclosures that do have contact with same building steel or other earthed components.

I would not call it an isolated ground, which imo, is mostly a myth, but can be built and maintained.

I would also not call it a redundant ground. To my eye it is two seperate and dedicated grounding systems or paths. Two different purposes, different paths, and for required GEC's, different rules. When they say 'ground it' usually means an equipment ground for fault current carrying, breaker tripping, and safety from stray voltages. The EGC does eventually connect to the earth and the NEC states the same saying 'grounded' is a connection to the earth.

When the IT equipment manufacturer says "Earth ground required" , imo, they are intending to require something beyond that. The Earth is a 0 volt reference point such that any connection to the earth will all be at the same 0 volt with no current flow between multiple paths. The earth would be like a giant zero impedance busbar such that one connection point would be no different than another, which is not practical or reality for many sites.

In a large facility with high current loads, multiple large services, large UPS's, connecting one source grounding busbar to another with a wire will have a measurable current flow on it. The EGC's cannot generally be expected to be cleanly zero volt since they are provided to actually carry current for normal and abnormal circumstances. Grounding is a working system, not something that sits there and does nothing. The EGC's will not have stray voltage on them but will have anomalous stray circulating currents on them.

The requirement is an adder, imo, due to the sensitivity of the equipment to minute electrical noise, which is forseeable and expected on the facility EGC paths. They are saying, provide a ground path that has no noise on it and is the same as and connected to the earth, for the purpose of not being blamed or liable when the very sensitive, expensive, and essential critical equipment fails to perform as specified. When problems do arise, logic and the process of elimination will place the grounding system on the list of possible causes. It will be checked and will be easy to demonstrate compliance with the requirements if you can point to a dedicated and compliant system.

It is a cheap way to avoid either future liability or future problems. I get paid to run wire and in this case, more is better. I would point to 250.30 (A) 3 or 4 and say, well by exception this may not be required, but if you require it, this (extra work) meets code.

 

[kwired]

I would not call it an isolated ground, which imo, is mostly a myth, but can be built and maintained.

I would also not call it a redundant ground. To my eye it is two seperate and dedicated grounding systems or paths. Two different purposes, different paths, and for required GEC's, different rules. When they say 'ground it' usually means an equipment ground for fault current carrying, breaker tripping, and safety from stray voltages. The EGC does eventually connect to the earth and the NEC states the same saying 'grounded' is a connection to the earth.

When the IT equipment manufacturer says "Earth ground required" , imo, they are intending to require something beyond that. The Earth is a 0 volt reference point such that any connection to the earth will all be at the same 0 volt with no current flow between multiple paths. The earth would be like a giant zero impedance busbar such that one connection point would be no different than another, which is not practical or reality for many sites.

In a large facility with high current loads, multiple large services, large UPS's, connecting one source grounding busbar to another with a wire will have a measurable current flow on it. The EGC's cannot generally be expected to be cleanly zero volt since they are provided to actually carry current for normal and abnormal circumstances. Grounding is a working system, not something that sits there and does nothing. The EGC's will not have stray voltage on them but will have anomalous stray circulating currents on them.

The requirement is an adder, imo, due to the sensitivity of the equipment to minute electrical noise, which is forseeable and expected on the facility EGC paths. They are saying, provide a ground path that has no noise on it and is the same as and connected to the earth, for the purpose of not being blamed or liable when the very sensitive, expensive, and essential critical equipment fails to perform as specified. When problems do arise, logic and the process of elimination will place the grounding system on the list of possible causes. It will be checked and will be easy to demonstrate compliance with the requirements if you can point to a dedicated and compliant system.

It is a cheap way to avoid either future liability or future problems. I get paid to run wire and in this case, more is better. I would point to 250.30 (A) 3 or 4 and say, well by exception this may not be required, but if you require it, this (extra work) meets code.

I agree.

The term isolated ground is likely somewhat misleading to some. Calling it a dedicated ground maybe is a little more correct, but I can still see similar misconceptions, so I don't really know what it should be called.

 

[infinity]

Is any of this stuff defined or require by the NEC?

 

[kwired]

Is any of this stuff defined or require by the NEC?

Required no, when someone tries to implement such a thing there are some rules though - like it must still be connected at very least to the grounded service conductor at the service equipment in some way or another and not just run said conductor to a ground rod.

 

[__dan]

Is any of this stuff defined or require by the NEC?

I looked for some references and as far as I could tell the NEC is only defined and required for premises wiring. Every piece of IT equipment has separately derived internal (DC) power supplies, but if they are self contained in the manufacturer's enclosure, they are not under the NEC, except for 110.3 (B).

At the design and construction stage the premises grounding has a design lifetime of 25 or more years. It is reasonable to provide a system that meets code for the broadest possible applicable loads. Doing this right on the first try is the lowest cost system. Anticipating the customers needs over the next 25 years, it is forseeable that the customer's utilization equipment will include systems that fall under the requirements of 250.20 (D) and 250.30 (A) 3 or 4. The minimum would be to expect multiple customer rack mounted UPS's and the system I proposed would be compliant.

Even if you put a large UPS in the basement, the IT maunfacturer may ignore it and plug their own rack UPS's and rack mounted lead acid batteries into it. It's wrong, but I've seen it multiple times with different manufacturers. It's forseeable trouble when you get hundreds of sensitive devices of all types, all grounded, but plugged into multiple power sources grounded at different myriad points in the building, grounded at the service, grounded at the basement UPS, grounded at remote subpanels and interconnected with com cabling. Connect all those source EGC busbars with a wire and there will be current flow on it. From the service, through the basement, through the building, the conductor will intercept and provide a path for an aggregate of problems in the building.

Required compliance with 250.30 (A) 3 or 4 is forseeable because the customer may add equipment requiring a UPS or transformer, now or in the future. Grounding the other local equipment to the same grounding system is good practice but not required for the bulk of the remaining loads, unless the manufacturer says "earth ground required" (a certainty). It is saleable work, an asset.

In a hospital you would have no choice but to request the manufacturer's grounding requirements and the reply will range from circular gobbledygook (comply with the NEC, when the NEC says may be grounded but not required) to multipage specs and difficult to comply with requirements, UFERs, dedicated isolated systems, weird stuff, especially with the utility phone service entrance, processors, network switches, medical equipment. The manufacturers requirements could be anything, which puts you back into 110.3 B