I'm always getting calls from people who want to plug thier new computer in, and can't decide into what.
I've called the maker's tech support many times, with mixed results.

Most want grounding, some will ask for a GFI to replace a 2-wire receptacle, a few want a dedicated circuit,very few want an Iso-ground. On the bizzar side,one chain i service insists on iso-grounds to the closest sub-panel ( I really think that guy just likes to walk around saying "ISO-Grounds") , and i did have an individual insist on a seperate ground rod some years ago.

Enter the UPS & or Power conditioner man, and you can apparently throw out that last paragraph.

It's my sense that there is a little less hoo-haa on the subject, now that computers are more refined, and cost less than they used to.

My Q is how do the rest of you guys in the field deal with this???

Sparky,

You've touched on a subject that I have some passion for, many ideas, quite a bit of designing around, studied and given design consultations to other ECs on.

I must warn you first off that my replies will definitely be long, in order to fully explain stuff.

With this in mind, would you and everyone else in this forum mind if I participate in this thread

Let me know, and if yes, I'll begin replies.


Scott "S.E.T."

Scott, go for it !
please keep my 3-sylable limit in mind...

Just a thought to add, grounding seems to be one of the hardest concepts in this trade.

I see NEC articles hashed out , many ending at the "theory" angle.

It appears as if we as a trade lack certain fundamentals pertinent to grounding.

Sparky,

I think that We, "as a trade", I mean the guys in the trenches may have a better grasp on the fundamentals of gounding than the "Theorists" that specify IG and mean a grounding rod alone and no bonding of the neutral in a 3phase Y transformer. They must teach that in MIS school, I have no Idea where they get that from.

Scott, maybe you, as an 'EE type' can give us some insight into this rationale?

Bill

I'm interested in what Scott has to say.

It is my experiance that 99.9999% of electronic equipment does not need isolated grounding.

75% of the isolated ground receptacles & circuits that I've seen are a code violation. The worst being the one's that involve a ground rod that is not bonded to the electrical systems ground.

The best book I've ever read on grounding is "Soares Book on Grounding."
Tom

I think that the reason there are so many problems with "grounding" is the choice of the word itself. 99% of what we call "grounding" is really bonding. By using the word "grounding" many people in the trade automatically start to think about a connection to earth. The word grounding should be reserved for the grounding electode system and grounding electrode conductor, everything beyond that is bonding.
Don(resqcapt19)

Good points, i got Soares a while back, it is a very understandable book and relavant to specific articles within 250
I'm a highlighter guy, so it now looks like my kids got to it...

Maybe some of the confusion is generated by the "theorists" , or maybe the terminology is simply flawed..????

[This message has been edited by sparky (edited 02-10-2001).]

Got a troubleshooting call myself from a local pharmacist. One of their computers is crashing and the 'puter guru insists that it must be a power quality problem.

I don't have an oscilloscope, so bear with me here.

The voltages fluctuate normally between 119 and 121 volts, but occasionally will drop suddenly to 114 and build back up. I realize that the voltages are within "normal" ranges, but does the sudden drop point to anything? Nothing that we know of seems to correlate with the voltage drop by a machine turning on automatically, etc.

Checking the current was more bizarre... it fluctuated rapidly between 3 and 15 amps with every imaginable number randomly appearing between. The computers, routers, hubs and printers were on but idle at this time.

All other circuits in the 120/208V 3 phase panel showed normal fluctuations in voltage and current. The voltage was steady on the questionable circuit on the line side of the breaker, but fluctuated as described above at the receptacles.

Would ISO grounds, isolating the neutrals from 4-wire "networks" to each leg having it's own neutral, and/or UPS (or line conditioners) help in this circumstance?

I'm moving my questions over to "Power Quality Primer"... I should have stopped there first!

Sparky,

Sorry it took so long to get back in touch with you on this thread - you've undoubtedly seen what's happening in other posts/forums

In order to keep the amount of info I'd like to throw in here, I'll spread my message across a few replies, to topics that arise.

Here's some basic stuff to get things started:

The usage of IG circuits can help a lot, or they can help create problems. The key point being the way things are installed and terminated.

The most crucial piece of equipment that would warrant an IG circuit in almost all installations would be the server. Work stations in stand alone buildings would most likely be fine without an IG circuit. When there are multi tenant buildings with various types of power consumption, the work stations could benefit from IG circuits.

There are a few methods for IG branch circuits that would be considered the "preferred" way to wire them. These are:

1: Dedicate only one IG ground conductor per circuit. Don't share the IG with any other circuits in a multiwire branch [1 IG per 20 amp circuit],

2: Dedicate the grounded conductor to only the LAN equipment on multiwire IG circuits [a 3 phase 4 wire multiwire branch would only run workstations, no convenience / G.P. receptacles or other types of loads connected to the multiwire circuit],

3: Dedicated circuits with Dedicated grounded conductors [2 wire circuit with IG conductor] for server equipment, or distribution frames,

4: Specialty equipment - consider using circuitry example #3 above.

If there is no way to get back to the grounding electrode for the system that supplies the LAN's power, IG circuits might be of no help - but might actually create problems [this depends on the level of noise and TVSS distributed on the grounding system].

Placing an IG bus in the sub panel that the IG circuits come from [the bus is Isolated like the grounded conductor's bus is]. Feed the IG bus with a subfeed ground that terminates to the grounding electrode system. It can be of minimum size [per largest circuit], or can be larger. We usually install a #4 cu IG subfeeder.

Don't connect the IG feeder to a stand-alone ground rod, or other means that are not bonded to the grounding electrode system. This not only has a poor ground fault connection, but also generates ground loops and noise. This is worse than no IG at all!

When bonding MDF racks and related equipment, bond it to the grounding electrode also. Bonding to a stand alone driven rod generates ground loops. Bonding to a water pipe at any distance away from where it might be bonded to the grounding electrode will generate noise.

For equipment connected to the IG circuits:

1: Supply at least 2 Dedicated IG circuits for server equipment. Each server and it's related / locally connected peripherals [routers, hubs / MAUs, external MODEMs, monitors] should be on the same circuit, unless there is a request for one peripheral to be on a separate circuit. The other circuit would be for future connected equipment. Verify load KVA on all connected items [most servers should be rated at 750 VA]. If a UPS is used with server equipment, it needs to be rated for the KVA load. If there is a local printer at the server / MDF / IDF, put it on a separate IG circuit from the server [this helps out on the voltage fluctuations].

2: For work stations, limit the maximum of work stations that have local laser printers to 3 per circuit, unless load KVA allows more, in that case 4 might be the Max. Keep all equipment per workstation on the same circuit. If using P.O.C. surge protection, use only for that workstation's equipment [this can be nearly impossible, as clients will plug things into any outlet available!]. Including a 4 plex convenience receptacle for G.P. use might help keep people from plugging unrelated equipment into the IG circuits.
Also, you might need to triple check / triple explain to the LAN equipment vendors that they should only use the IG circuits you have painstakingly labored to install [I have to do this over and over and over!!! and still find things plugged into the convenience receptacles].

3: Place LAN printers on Dedicated IG circuits.

4: Only connect LAN copier / printers to IG Dedicated IG circuits. If not connected to the LAN, no need to IG it's circuit.

5: If you do a lot of bank branch installations [like I do], give the ATM at least 2 Dedicated IG circuits.
Give the TCD machines a Dedicated IG circuit for each machine [unless equipment vendor agrees to allow up to 3 per circuit].
Give the TCD server a Dedicated IG circuit if you can [it doesn't normally need one since it is an inexpensive piece of equipment, however, if the location has excessive noise and/or poor power quality, the server might reboot at random, which can be a nuisance].
Install the ATM and TCD machine's circuitry as the LAN server's circuit would be - dedicated grounded conductor and dedicated IG conductor per 20 amp dedicated circuit, running only that piece of equipment.
If the CCTV system is one similar to the "Accu-Trak" system, consider dedicated IG circuit for it.
When connecting the static build up ground connection for the ATM and MDF room floors, bond that to either a driven rod or water line. It's purpose is to drain static [ESD] only.
Supply the Tel equipment controller with a dedicated IG circuit.
Supply the Telco a minimum #6 cu ground, bonded to the grounding electrode system, for their MPOE. Supply them an IG receptacle if the MPOE is at a remote location from the Tel equipment / MDF [not in the same closet].

When working with shielded network cables [coax, triax, 10 base-2, 10 base 5, IBM type 2, etc.], instead of UTP [CAT 5, etc.], be sure there is only one common point that they get ground bonded at. Otherwise there will be loop current flowing on the shields.


In a nutshell, the PC, or workstation needs to be grounded [solidly] not only to conform with the NEC, but also to drive any received RFI to ground level at the machine it's self. Since the power supply's secondary eventually is center tapped and connected to the frame and chassis inside the machine - and the center tap is used for the "Common" of the +5/0/-5 VDC power [this is for the logic gates], it helps to eliminate any interference that could cause data loss [parity checksums incorrect, oscillation of the local bus that's out of sync with the CPU/bus controller, etc.]
Poor power quality, being in the area of sustained sags and surges in the line voltage might cause the power supply to initiate a series of resets to the CPU [reboot]. The level and duration of this depends on how good the power supply is. Cheaper ones will reboot at the drop of a hat, better ones will accept quite a lot of "Line Trash" and voltage variations before resetting the CPU.
BTW and FYI: This is also the way the systems is restarted when the "reset" button is pressed. It applies a logic 0 to the CPU's power OK signal address, which causes the CPU to hold it's self in a reboot state until the signal becomes high again [logic 1].

Thought this would be interesting


If your client really wants IG circuits and is willing to pay for correctly installed IG systems, by all means go for it. Just make sure the person you are discussing this with understands what is really going on, not just walking around quoting "Buzz Words" to appear like they know what they're talking about.

Be a good consultant to your client. Checking here for some input is a great thing to do! Supplement this by contacting equipment suppliers, vendors and manufacturers for their input. That's where you will get the environment specs and load KVA specs needed to design your circuits.

An extra benefit to IG circuitry is the inherent TVSS suppression. This is one of the reasons that 2 wire circuits with dedicated IG conductors are preferred on certain loads, along with why a dedicated IG conductor is preferred per circuit on multiwires and the whole connected load scheme.


How's this for a "short" message??

Hope this sheds light on some basics.

Scott "S.E.T."

Thanks Scott;
I had a funny feeling about those "stand alone" ground rods. I can see how it would cause problems. I'm not sure where that came from, i'm going to say out of the trade, as we all are heavily programmed to bond all applicable together to the Earth's
core...
could you expand on the distant h2o pipe problem??