Bond the incoming green equipment grounding conductor to the enclosure body and the ground bus, and leave the plant grid cable alone. The grid is a bonding network that holds steel, poles and boxes at one touch potential; it has no low-impedance connection back to the transformer, so it cannot deliver the current a breaker needs to trip. Work the numbers for your feeder — say 220 VAC on a 20 A breaker with a 60 ft [18 m] run: your EGC loop delivers about 1,100 A and trips magnetically, while the grid loop delivers 22 A and the breaker rides through the fault. Bond the EGC at the entry, size it from the breaker ahead of the circuit, and keep the insulated bus as a clean terminal only when a standard EGC also runs the loop.
Does the incoming EGC have to bond to the enclosure body?
Yes: bond the incoming green EGC to the enclosure, because NEC 250.4(A)(5) defines the fault return as the conductor run with the circuit conductors and excludes the earth from that path entirely.
A fault at one of the power supply's line terminals has to reach the source through metal: phase to chassis, chassis to bus, bus to enclosure, then along the EGC back to the transformer. NEC 250.96(A) requires metal parts that carry fault current to be bonded together so they can conduct it safely. The fundamentals sit in the enclosure grounding and bonding guide.
The back plate is in that loop too, because the supplies sit on it. Bond the plate to the enclosure and to the bus with jumpers sized like the EGC. The insulated footing under the bus changes none of this; it only decides which terminal is the clean one. Land the EGC on a lug at or beside the gland plate and take a bonding jumper from the bus to the enclosure stud, as the grounding conductor entry page shows.
One separation still matters. The grounded circuit conductor lands on its own terminal, never on the case: NEC 250.6 aims to keep neutral current off metal parts, and the subpanel neutral and ground article walks that wiring.
What is a plant ground grid?
A plant grid is a buried network of conductors and driven electrodes that bonds steelwork, tanks, racks and enclosure bodies to one touch potential, typically reading 1-5 ohms to earth where the design follows substation practice (IEEE 80).
A chemical plant runs bare copper through trenches and up the poles, bonds every tank shell, rack, fence post and enclosure body to it, and drives electrodes at intervals. The jobs are touch and step voltage, lightning and static: everything you can touch rises together during a fault or a strike, so the voltage across you stays small. It is a bonding network, not a circuit conductor, and nothing in it returns to the transformer winding except the soil itself.
The resistance figures sharpen the difference. A single driven rod in average soil reads up to 25 ohms, the benchmark the NEC applies when deciding on a second electrode. Both the grid figure and the rod figure describe one connection to earth. A fault loop needs two such connections plus the soil path between, and in the worked example below that loop comes to 10 ohms.
Why can't the plant grid carry fault current back to the source?
Because the earth loop back to the transformer is long and resistive: at 10 ohms of loop resistance, a 220 V fault pushes only 22 A, which a 20 A breaker reads as 1.1 times its rating and rides through on the thermal element.
Three impedances stack in that loop. First comes the soil path between your grid connection and the source electrodes. Second comes electrode contact resistance, and even a good 5 ohm grid connection dwarfs the 0.2 ohm of a copper loop. Third comes inductance: loop reactance grows with the spacing between the go and return paths. Conductors run together in a conduit show about 0.05 ohm per 1,000 ft at 60 Hz in standard conduit reactance tables. Sixty feet of that adds 0.003 ohm. A grid loop tens of feet wide carries several times that reactance, on top of the soil resistance.
Bonding and fault clearing are different jobs. The grid holds your box, the next pole and the tank farm at one potential; the code still demands the conductor run with the circuit to clear a fault, and the earth is no part of that effective ground-fault current path. The same question reaches motor control centers, and the page on whether an MCC bonding conductor needs a source return path works it in the same terms.
How much fault current does each return path deliver?
Take a realistic case: a 220 VAC feeder on a 20 A breaker, a 60 ft [18 m] run of 10 AWG copper EGC paired with the phase conductor, and a plant grid bonded at 5 ohms locally plus 5 ohms at the source electrode.
The EGC loop. 10 AWG copper measures 1.21 ohms per 1,000 ft at 75 °C (167 °F) in standard conductor resistance tables. Sixty feet of go-and-return is 2 x 60 x 1.21 / 1,000 = 0.145 ohm. Add 0.05 ohm for terminations and reactance and the loop is about 0.2 ohm. Fault current is 220 V / 0.2 ohm = 1,100 A, which is 55 times the breaker rating. The magnetic element of a typical thermal-magnetic breaker trips between 5 and 10 times rating, so this fault clears in about one cycle, 16.7 ms at 60 Hz.
The grid loop. 5 ohms + 5 ohms = 10 ohms, and 220 V / 10 ohms = 22 A, just 1.1 times rating. That sits below the magnetic trip, and the thermal element needs minutes at that multiple, so the breaker stays closed while the fault burns in the box.
Touch voltage. With the grid as the only return, the enclosure sits at 22 A x 5 ohms = 110 V above remote earth, and it stays there until something burns open.
Both paths together. 0.2 ohm in parallel with 10 ohms is 0.196 ohm, so the loop carries 220 V / 0.196 ohm = 1,122 A. The EGC takes 1,122 x 10 / 10.2 = 1,100 A, about 98 percent, and the grid takes 22 A, about 2 percent.
| Return path | Loop impedance | Fault current | 20 A breaker | Outcome at the box |
|---|---|---|---|---|
| EGC run with the circuit | 0.2 ohm | 1,100 A | Instant trip, about one cycle | Touch exposure ends in one cycle |
| Plant grid only | 10 ohms | 22 A | Rides through for minutes | 110 V above earth, sustained |
| EGC and grid in parallel | 0.196 ohm | 1,122 A | Instant trip | Grid path shunts 2 percent |
Do two ground return paths in parallel weaken the breaker trip?
No: parallel return paths only add current, so in the worked example the breaker sees 1,122 A instead of 1,100 A and still trips magnetically, with the grid carrying about 22 A, roughly 2 percent of the total.
A breaker responds to the current through its own contacts; it cannot tell which return path delivered which share. Parallel impedances raise the total, and both paths here leave the box through the same feeder, so neither bypasses the trip. The instinct to keep returns separate belongs to noise control and circulating currents, not to fault clearing. It is the same logic that lets one raceway carry a single EGC for several circuits. A parallel run then gets an EGC in each raceway, every one sized from the same breaker (NEC Table 250.122). More paths make the trip faster, not slower.
What about the insulated ground bus bar?
The insulated bus is an isolated-ground layout: the green feeder wire acts as an isolated equipment grounding conductor, and the NEC permits that only when a standard EGC also runs from source to load, bonding every metal part along the way.
NEC 250.96(A) still governs your panel: the enclosure and the back plate are metal parts that must be bonded, so a lone insulated wire cannot be the whole answer. Two arrangements work. Bond the green wire to the enclosure and give up the isolation, which is the fix when the feeder carries one conductor. Or pull a second insulated conductor so the panel runs a standard EGC and an isolated one side by side, each sized like the EGC (10 AWG in the example), with the insulated run kept whole from source to bus.
The bus's insulated footing then does its real job: it keeps supply leakage and switching noise off the electronics' reference while the fault path stays intact. The supplies' protective-earth terminals land on the bus either way. Whether the DC output common joins the enclosure is a control-noise decision, made once, at one point in the panel.
How do you bond the EGC inside a stainless enclosure?
Run a copper jumper from the bus to the enclosure stud and the back plate, sized like the EGC — 10 AWG in the worked example, above the 12 AWG minimum NEC Table 250.122 sets behind a 20 A device.
De-energize the feeder and verify absence of voltage at the terminals before you open the gland plate; after that it is ordinary hardware work. The step-by-step version of bonding the bus to the box is in the bonding an enclosure to a ground bus article, and the short version runs:
- Land the feeder EGC lug on the bus bar at the lug's marked torque.
- Jumper the bus to the enclosure stud and to the back plate.
- Keep the contact surfaces bare: stainless arrives uncoated, and NEC 250.12 still requires paint, lacquer and enamel removed from threads and contact surfaces, so skip touch-up paint on a bond land.
- Bond the door with its own jumper when it carries equipment; the enclosure door bonding jumper article sizes that bond.
- Where screened cable enters, land the screen's earth path at the gland, as the cable gland earthing article explains.
| Breaker ahead of the circuit | Minimum copper EGC, NEC Table 250.122 |
|---|---|
| 20 A | 12 AWG |
| 60 A | 10 AWG |
| 100 A | 8 AWG |
| 200 A | 6 AWG |
What changes on the enclosure order?
Order bonding in from the start: mark every gland cutout with its cable outside diameter, place the bond stud beside the entry, and keep field drilling to zero, because every later hole puts the manufacturer-rated NEMA 4X / IP66 rating of the box at risk. Our enclosures are rated by the manufacturer; not UL listed.
We build custom stainless enclosures to your drawing: you set the material, the width, height and depth, and every cutout. Delivered in 10 business days, counted from your approval of the drawing; the custom enclosure builder takes the dimensions and cutouts directly. In a chemical plant outdoors, 316L earns its keep: both 304 and 316L reach chloride stress-corrosion cracking near 60 °C (140 °F), but the molybdenum in 316L resists the pitting that chloride deposits start, even at room temperature.
Put the gland positions, the bond-stud location and the back-plate standoffs on the drawing; the cutouts, glands and windows guide covers the entry hardware. If a hole does get cut in the field, the article on the rating a panel keeps after cutting tells you what survives.
So the decision comes out three ways. A feeder with one green wire: bond it to the enclosure and the bus, and keep the grid cable where it is. A panel that needs a clean reference for its electronics: pull a second insulated conductor and run it beside the standard EGC. A new box: put the bond stud and the entries on the drawing, so the rating is never touched in the field.
Catalog dimensions to compare
Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 12 × 10 × 6 in | 304 stainless | 12 x 10 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 20 × 16 × 8 in | 304 stainless | 20 x 16 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 24 × 20 × 8 in | 304 stainless | 24 x 20 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
How do I size the equipment grounding conductor for the feeder?
NEC Table 250.122 sizes the EGC from the rating of the overcurrent device ahead of the circuit, not from the load: 12 AWG copper behind a 20 A breaker, 10 AWG behind 60 A, 8 AWG behind 100 A and 6 AWG behind 200 A. The worked example runs 10 AWG, one size above the 12 AWG minimum, for margin on a 220 V feeder.
Does a ground rod or the plant grid count as the EGC?
Under NEC 250.54 a rod beside an enclosure is an auxiliary electrode: bond it to the EGC if you like, but it never substitutes for the conductor run with the circuit. A single rod in average soil reads up to 25 ohms, so a two-electrode earth loop leaves about 4 A on 220 V, far below any trip.
What color does an isolated ground wire have to be?
NEC 250.119 requires insulated EGCs of 6 AWG and smaller to be green or green with one or more yellow stripes. Pull the isolated conductor in that color, label both grounds at each end, and land the standard EGC on the enclosure while the clean bus keeps its own insulated run.
Should I disconnect the plant grid cable from the enclosure?
Leave it. It holds the enclosure, the poles and the cable trays at one touch potential through faults and lightning, and in the worked example it shunts only 22 A of a 1,122 A fault, about 2 percent. Removing it would not speed the breaker; it would remove the equipotential control.





