The risk/benefit situation of SG and IG wiring methods has many variables, all depending upon the specific application and situation.
There's plenty of confusion regarding the use, performance, and NEC requirements of IG (isolated ground) wiring. Can it be used anywhere, or is it restricted to electronic load equipment? Does it really reduce electrical noise? What does the Code really say about IG wiring? Let's attempt to eliminate this confusion and provide guidance in complying with the NEC.
IG-style direct connection
This means of IG connection is relatively new to the NEC and is somewhat controversial because making an NEC-acceptable electro-mechanical interface to electronic load equipment is somewhat elusive. This topic is covered in the Exception in Sec. 250-75. Basically, the intent of this Exception is to give the same opportunity of using IG-style wiring on direct-connected electronic load equipment as cord, plug, and receptacle-interfaced versions have had for many NEC code cycles. There was no "push" based on any demonstrated need for the code-making process to permit this IG-style wiring method.
A typical direct connected IG-style wiring method is based on a metallic conduit or raceway that's suitable for use as an equipment grounding conductor path. Examples would include electrical metallic tubing (EMT), intermediate metal conduit (IMC), and rigid metal conduit (RMC).
The insulating means used between the branch circuit's metal conduit/raceway and the electronic load equipment metal enclosure is undefined in the NEC, except for the requirement that it be listed by a nationally recognized testing laboratory (NRTL). This has been discussed and clarified to mean that the insulating fitting need not be specifically listed for use as a direct connection IG insulating fitting; it can be any listed nonmetallic electrical fitting that would do the job in a manner acceptable to the authority having jurisdiction.
It would seem, then, that listed electrical fittings used with rigid PVC electrical conduit would be acceptable, while plastic plumbing fittings would not.
Where can you use the IG connection?
There has been a lot of confusion over where the IG-style connection can and cannot be used in an electrical system installed under the requirements of the NEC. This is clarified in Fig. 1, where the acceptable location is shown in contrast to all of the unacceptable ones. Note that there is only one location that is acceptable: at the outlet end of the branch circuit used to connect to electronic load equipment.
This means that any attempt to place an IG-style connection (receptacle or direct connection) onto any other point on the wiring system, or to serve non-electronic equipment with it, is an NEC violation of either Sec. 250-74, Exception No. 4 for receptacles, or Sec. 250-75, Exception for direct connections.
The above is important, since plastic bushings have been seen installed in metal enclosure knockouts and punchouts, where 4-in. rigid steel conduits of multiple paralleled feeders serving high-ampacity switchboards terminate. Once, this author even saw such bushings installed at both termination points of a 3-in. EMT 480/277V feeder. These are hazardous practices not permitted by the NEC. Unfortunately, until the IG methodology is fully understood by all, these practices will continue to pop up.
High- and low-impedance loads and IG circuits
Loads that are affected by common-mode currents and voltages that are propagated over the AC power branch circuit fall into high- and low-impedance categories. High-impedance loads are susceptible to electrical noise in the form of voltage. Low-impedance loads are susceptible to electrical noise current.
The IG wiring method changes the lower impedance of the typical SG (solid grounding) branch circuit into a much higher impedance. Hence, it attenuates noise current, but permits an increase in noise voltage on the path.
Looking into the load-end of an SG branch circuit, you "see" a relatively low impedance that permits a fairly heavy noise current to be propagated along the path, but with minimum voltage drop. Conversely, looking into the load-end of an IG circuit, you "see" the reverse condition: maximum voltage drop and minimized current flow.
From the above, you can infer there's a compatibility consideration between the victim load's impedance and its susceptibility or immunity to electrical noise in the common-mode, and whether an SG or IG wiring method should be used. The general "rule" to improve electronic load immunity or to reduce susceptibility to common-mode electrical noise is use inverse connections (e.g., connect a low-impedance load to a high common-mode impedance branch circuit, and a high-impedance load to a low common-mode impedance branch circuit). Also, woe unto those who invert it.
Length of the IG circuit is key
The length of the IG circuit from the solidly grounded AC source to its outlet end affects the impedance of its path from one end to the other. This, in turn, affects the amount of electrical noise current (as opposed to electrical noise voltage) the IG path can propagate along its route. It stands to reason, then, that whatever effect the IG wiring method is going to have with a given load, it will be proportional to the length of the IG circuit. A similar effect exists on the SG circuit, but in inverse fashion.
From the above, it's apparent that placing an IG receptacle 6 in. away from the branch circuit panelboard, and having the panelboard a few feet from its serving solidly grounded transformer that forms the separately derived AC system, is not going to have much effect at all, one way or the other. There simply isn't enough IG length to the circuit to develop any useful common-mode impedance on the equipment grounding conductor path for the arrangement to do any good (or bad). However, if the IG branch circuit is 100 ft or 200 ft long, then the effects (again, good or bad) would be appropriately produced.
This should tell you something about paying extra cost to install IG receptacles on circuits that are only a few ft (or in.) long. In particular, there's no useful gain in installing IG receptacles onto the same case/enclosure that contains a solidly grounded power conditioning device such as a transformer used with a UPS, a voltage regulator, or similar item of self-contained power conditioning equipment.
Long IG circuits and increased noise problems
Is there such a thing as too much of a good thing? Always, and in the case of the IG wiring method, this is again true. For example, if the site has a lot of potential difference between points of grounding (as over large area buildings), an IG wiring method can act to ensure that the AC power source and the electronic load get conductively connected across that potential. The result is increased common-mode current trying to flow on the IG path and a much greater common-mode voltage being developed across it. This can cause more interference with the victim load, if it has a communication or signaling circuit attached to it and is ground-referenced at some further distance away.
IG circuits and lightning problems
IG wiring methods bring with them increased susceptibility to lightning-related problems. Generally, this hasn't been appreciated by those advocating the use of the IG circuit with the expected results. This can be readily explained.
When lightning strikes the face of the earth, there's a step-potential situation created known to result in kV/meter across the soil. Similarly, this kind of potential difference can be created by lightning currents vertically, horizontally, and on any diagonal in a facility of given size, particularly multistoried and large area buildings. Hence, there can be a very great potential difference, for example, between a service entry and the main building's grounding point, and any remote point in the building when a lightning strike occurs. This means that if the victim load equipment is located a long distance from the service entry, its local grounding conditions will be at a significantly different potential from what the conditions at the service equipment will be. If this path and its potential difference is then bridged by the connection of an IG equipment grounding conductor, there will be, by necessity, a traveling voltage and current wave front barreling down the IG conductor from one end to the other, in a futile attempt to permit current flow and to equalize potential between the two points. This is called a lightning surge current.
Let's assume that the service entry is the originating point for the above surge current and the victim electronic load is the terminating point for it. If there are data, signaling, or communications cables attached to the load equipment, the lightning surge current will try to flow on these circuits in a continuing attempt to find "ground." Typically, this is quite damaging to these circuits and, since the surge current is in the common-mode, no amount of line-to-line and line-to-ground surge protection equipment installed on the AC power circuit will protect the load and its attached circuits.
Also, if there is grounded metal nearby to the victim load, and if the surge current has sufficient potential driving it (due to impedance mismatch and related voltage reflections at the load-ground interface point), there can be a lightning side-flash of up to 6 ft horizontally through the air between the energized victim load and whatever grounded materials or other equipment is nearby. Side-flash phenomena is generally described in NFPA-780-1991, The National Lightning Protection Code.
So, if an SG circuit is used in the above case instead of the IG style, the victim load will be largely protected, since it will have become ground-referenced to its local ground conditions and will be generally unable to develop a side-flash potential problem. The same holds true for an IG circuit that is physically short, as when its AC system is not the remotely located service entry (and that is at a greatly different potential), but is instead a dry-type transformer (or similar source) installed as a solidly grounded and separately derived AC system in the same location as the load being served on the IG path.
Summary
There are no clear winners and losers between the IG and SG wiring methods since, as we've tried to explain, the risk/benefit situation is fraught with variables of a wide description. Therefore, no blanket recommendation or statement can be made in relation to the SG or IG methods as regards to predictable benefits that either can provide. The results are related to the site's conditions, the length and routing of the circuit to be made IG or SG, and the high- and low-impedance nature of the victim loads themselves. Please understand that the effects of the SG wiring method versus the IG wiring method on electrical noise are only useful with common-mode noise problems, not transverse-mode ones.
Finally, what's clear is that it's very difficult and expensive to retrofit an SG design into an IG design, if you find out you need one. However, retrofitting an IG design into an SG design is relatively inexpensive, since the IG conductor easily can be either disconnected or reconnected as an additional SG conductor at any time. This translates into a recommendation that electronic loads should be provided with AC branch circuit wiring that's IG in nature, but with the clear idea that if conditions warrant it, the circuit will be converted to the SG type (and back again at a later time) as may be needed.
RELATED ARTICLE: IG WIRING AND THE NEC
The following are excerpts from the 1996 NEC that apply to IG wiring.
IG wiring with receptacles. NEC Sec. 250-74, Connecting Receptacle Grounding "Terminal to Box. An equipment bonding jumper shall be used to connect the grounding terminal of a grounding-type receptacle to a grounded box."
"Exception No. 4: Where required for the reduction of electrical noise (electromagnetic interference) on the grounding circuit, a receptacle in which the grounding terminal is purposely insulated from the receptacle mounting means shall be permitted. The receptacle grounding terminal shall be grounded by an insulated equipment grounding conductor run with the circuit conductors. This grounding conductor shall be permitted to pass through one or more panelboards without connection to the panelboard grounding terminal as permitted in Section 384-20, Exception so as to terminate within the same building or structure directly at an equipment grounding conductor terminal of the applicable derived system or service."
"(FPN): Use of an isolated equipment grounding conductor does not relieve the requirement for grounding the raceway system and outlet box."
IG wiring on direct connected circuits. NEC Sec. 250-75, Bonding Other Enclosures. "Metal raceways, cable trays, cable armor, cable sheath, enclosures, frames, fittings, and other metal noncurrent-carrying parts that are to serve as grounding conductors with or without the use of supplementary equipment grounding conductors shall be effectively bonded where necessary to ensure electrical continuity and the capacity to conduct safely any fault current likely to be imposed on them. Any nonconductive paint, enamel, or similar coating shall be removed at threads, contact points, and contact surfaces or be connected by means of fittings so designed as to make such removal unnecessary."
"Exception: Where required for the reduction of electrical noise (electromagnetic interference) on the grounding circuit, an equipment enclosure supplied by a branch circuit shall be permitted to be isolated from a raceway containing circuits supplying only that equipment by one or more listed nonmetallic raceway fittings located at the point of attachment of the raceway to the equipment enclosure. The metal raceway shall comply with provisions of this article and shall be supplemented by an internal insulated equipment grounding conductor installed in accordance with Section 250-74, Exception No. 4 to ground the equipment enclosure."
"(FPN): Use of an isolated equipment grounding conductor does not relieve the requirement for grounding the raceway system."