We need to install some industrial control panels that have an both a regular ground bar and an isolated ground bar that the shield wires from some 24VDC instruments' are connected to. How should the isolated ground bar be grounded.... run to building steel, drive a separate ground rod near the panel, or what?????
The panel is powered from a breaker panel by two 120VAC 15 amp circuits.
You ask a real loaded question Bill. I'll stick my neck out first. Part H of Article 250 covers DC grounding, but without a bit more info, I can't swear it does (or doesn't) relate to your system. Is it power limited or not ? In any respect, the shield ground should likely go back to the 24 volt DC source, as it is not AC system, nor equipment ground, but an RF drain. Depends again on your equipment and its sensitivity. some fire alarm equipment does not take kindly to being mixed with AC grounds, and some have to be landed only on one end so as to increase the capacitance and reject stray RF and voltage. Clear as mud ??? I do believe your 24v DC ground should be taken back to the 24 volt source is the short answer.
Quick answer: Do not install a new ground rod unless it is "connected" to the system's ground grid. You probably need to connect your isolated ground bars to the ground reference connection of your 120VAC power source.
Please keep in mind that there can never be a true isolated ground. In the case of most communications and shielding systems the intent is to provide a single point ground not an isolated one.
To create a single reference (ground) point you need to prevent all possible ground connections except at your load and your source, this is where the term isolated comes from. In a normal electrical system there are multiple ground paths and many interconnection points, for example a green grounding wire is often connected to the conduit system at each receptacle point. In isolated systems all interconnections between the green grounding wire and the conduit are prevented except at the source (a single point).
Yes, there are subtle differences between communications and instrumentation grounds but the concept of single point grounding is still applicable.
[This message has been edited by JBD (edited 11-28-2001).]
As JBD as said, a lot of grounding for sensitive equipment is about providing one, and only one, ground path, thus avoiding the possibility of a ground loop.
If you visualize a cable shield which has multiple paths back to the main ground, some of which run parallel for long lengths next to an AC power wire, then that ground loop can behave like a shorted turn of a transformer and get noise induced into it. The current flowing in the shield can then induce interference into the conductors it is supposed to be shielding.
It's a similar principle to the ground-loop hum you can get with separate hi-fi units, each linked by shielded audio cables and each also grounded through the power cords.
But as George pointed out, it's hard to be specific without some more details. For the instrumentation in question, it's important to be clear as to whether we're talking about a ground in the AC power sense, a DC sense, or a high-frequency RF sense.
For example, in some radio equipment an RF ground may be made through a small value capacitor which effectively isolates DC and 50/60Hz AC but provides a low-impedance path to high frequencies. Conversely, a 50' length of #16 ground wire in an extension cord may be very effective as a protective AC-power ground, but its inductance makes it almost useless as a ground for, say, UHF radio frequencies.
So, the shielding connected to an independent busbar may be best grounded by bonding that bar to the main AC ground or to the 24V DC supply common, depending upon the type of ground it is intended to be and the grounding arrangements of the rest of the system.
The proper and best method for your IG "Sub Feeder" is to bond it to the Grounding Electrode System of the system which powers the panel, at the same bus which the Grounded Conductor is connected to the GEC. This is the more "Designed" method, rather than the NEC required. NEC doesn't care about Design issues, rather it does care about safety. This "Forces" the NEC to say it's OK to bond the IG SubFeeder to any convenient Grounding Point [such as the Subpanel with the IG Circuit's Branch Circuit's origin]. That will work, but is not the "Intent" of an IG System.
The IG System in most Design Parameters will be derived from the "Power Source's Main Grounding Point" Simply this equals out to:
[*] For Systems directly fed from the Utility Transformer [Main Service is 208Y/120 3 ph. 4 wire for example], run the IG subfeeder to the main service equipment, and connect it to the Grounded Conductor's Bus where the Grounding Electrode Conductor is also bonded to that Bus [the Grounded Conductor].
[*] For Separately Derived Systems [example 480 VAC to 208Y120 3 ph. 4 wire customer owned Transformers], run the IG subfeeder to the Transformer, then bond it as the Grounded Conductor, Equipment Grounding Conductors, Enclosure and the Grounding Electrode Conductor are all bonded together within the Transformer. </OL>
If you drive any rods for the IG system, connect and use them as SUPPLIMENTAL items only - not the sole Grounding Method. If you bond to Building Steel, do so in a matter that limits the overall length [physical length] that IG currents need to flow, in order to "regroup" at the Gronded Conductor.
Best if you bond the IG subfeeder as shown in examples 1 and 2 above - this will leave the lowest potential between IG and EGC, or IG and Grounded Conductor - which will result in the lowest levels of Ground Loops and related noise.
We have some discussions and Schematics for IG Systems in ECN, so take a look through the various areas for more information, or feel free to ask more questions.
Scott " 35 " Thompson Just Say NO To Green Eggs And Ham!
Re: Isolated ground installation#5625 11/29/0106:58 AM11/29/0106:58 AM
Guys, You certainly are correct on your advice for AC systems, no doubt it MUST tie into the system ground, if I was unclear about that, sorry. You NEVER drive a ground rod and call it adequate. BUT... he is working on a dc instrumentation system. Every time I've had trouble with those, it was because some sucker had tied it into the ac ground system. If I read him correctly, he needs to take the 24v instrumentation ground back to ITS source, not the ac source.
Re: Isolated ground installation#5626 11/29/0111:09 AM11/29/0111:09 AM
If the DC system is grounded, then the AC and DC grounding electrodes must be tied together. If not there will potential differences between the 2 grounding systems that may lead to problems. About 15 years ago we installed a DCS control system where the manufacturer insisted on an independent grounding system for the DC ground. No connection to the AC ground was permitted, other than the earth. For the first three years every time there was a thunderstorm in the area, boards would be burnt out on the control system. We told the manufacturer that the grounding systems should be tied together to eliminate the problem. They insisted otherwise and had us repeatedly prove that the two grounding systems were independent. After the third year of burnt out boards, they let us tie the grounding systems together. The problem went away. Don(resqcapt19)
Re: Isolated ground installation#5627 11/30/0109:57 AM11/30/0109:57 AM
Thanks for all the responses so far. The panel feed is 15 amp 120VAC grounded system, which then connects to a 1000VA UPS. Each instrument has its own small power supply to keep everything isolated. The 24VDC power supplies are ungrounded (at this time). The shield/drain wires for the instrumentation cables are the only thing connected to the isolated ground bar at this time.
Did I read it correctly that some responses are recommending tieing the isolated ground bar back to the DC power supply negative lead? Any answers are appreciated. Thanks
Re: Isolated ground installation#5628 11/30/0110:10 AM11/30/0110:10 AM
I'm surprised that you are being required to tie the instrumentation cable shields to a ground bus at the supply end.
My experience has been that on shielded low-voltage loops, fed from an ungrounded supply, you would bond the shields only at the instrument end, not at both ends. This eliminates the possibility of induced ground loops in the shield.
If the engineer absolutely insists that you bond at supply end to the isolated bus, then you should carry an insulated grounding conductor of the appropriate size for the upstream overcurrent protection back from the isolated bus and tie it to the UPS system ground.