Sec. 250-24(a)(5) is one of the more fundamental Code rules, and perhaps one of the most misunderstood at the same time. Often referred to as the "single-point grounding rule," this is the prohibition against making more than one connection between a grounded circuit conductor (could be a neutral, but not necessarily) and an equipment grounding conductor:
Load-Side Grounding Connections. A grounding connection shall not be made to any grounded circuit conductor on the load side of the service disconnecting means except as otherwise permitted in this article.
Most electrical people know about this prohibition, on some level, but if the hits on our Web site over the last year are any guide, many of them don't understand it. This rule isn't in the Code for some arcane theoretical purpose. It's in the Code to address serious safety concerns.
The EC&M Panel's analysis. The best way to understand this rule is to look at what happens if you don't comply with it. Suppose, for example, you wire a 100A subpanel at the end of a 100-ft feeder comprised of No. 3 copper THHN run in 11/4-in. EMT. Assume the wiring system is 480Y/277V, with the usual main bonding jumper at the service where this feeder originates. We'll say the line-to-neutral load runs about 60A.
Now suppose there's a bonding jumper in the subpanel, in violation of Sec. 250-24(a)(5). That means there's a solid, deliberate connection between the equipment grounding system and the grounded electrical system at the subpanel. There's a similar connection at the service, over the main bonding jumper installed in accordance with Sec. 250-28. The grounded conductor is just that, a conductor. The equipment grounding path between the two locations is also a conductor, albeit not the kind we buy on a reel at a supply house. Two conductors joined at each end form a parallel current path, and each leg carries current in inverse proportion to its impedance.
In the example we've been working with, the conductor impedance (from Chapter 9, Table 9) runs about 0.024 ohms. The EMT impedance (from industry sources) runs about 0.063 ohms for the same length. Impedances in parallel lower the total circuit impedance, in this case (based on 1/Z1`1/Z241/ZT) to 0.018 ohms. When you apply the 60A load across this impedance, the voltage drop comes out (based on E4IZ) to just over one volt.
This probably doesn't seem impressive, but now look at what's happening to the current flow. The impedance ratio of the two parallel paths (raceway and neutral conductor) in this case stands about 2half-to-1, so about 18A ends up on the raceway, all the time. This is a dangerous condition. If someone loosens a set screw or tries to work on that run, they'll end up drawing an arc from supposedly dead metal.
There's another, even more serious problem. Equipment grounding connections just aren't evaluated as connections that should routinely carry current. For example, many of you have tripped over the different bus termination rules for branch circuit neutral terminations versus equipment grounding terminations of similar sizes. For the neutral connections, only one wire per terminal (usually), but you could put up to three (in some cases) equipment grounding conductors of the same size in the same hole.
That's because the neutral connections must meet a heat-cycling test that equipment grounding conductors don't have to pass. Why? Because equipment grounding conductors aren't supposed to be routinely carrying current and going through heat cycling in the field. Although raceway fittings have to meet fault-current testing, those fittings certainly aren't tested as elements of a load conductor.
When do you suppose the manufacturer of that friendly piece of EMT over there ever imagined they were making load conductors? When might the testing laboratory ever have imagined the same thing? If you improperly reground grounded circuit conductors downstream, however, that's just what you've created. Bad idea.