Short answer: yes. Bond the panelboard equipment ground bar to the electrical room ground bar, and the jumper carries fault current alongside the feeder EGC without overloading either path. Keep two things fixed while you do it. The feeder EGC sized from NEC Table 250.122 (10 AWG copper on a 60 A breaker) travels with the feeder. The grounded (neutral) conductor bonds to the enclosure only at the service disconnect. The bond you add is a parallel fault path, and the code expects fault current on metallic paths in parallel. Worked out below, a 4,000 A fault return splits about 3,200 A on the EGC and 800 A on the building steel.
Can you bond a panelboard ground bar to the electrical room ground bar?
You can bond it, with a copper jumper no smaller than the largest feeder EGC it parallels. The result is a parallel fault path the code allows rather than a violation you have to defend.
Fault current already fans out across every metal route back to the source: the raceway, the panelboard can, the building steel, any tray or structure bonded along the way. NEC 250.96(A) has you bond those metal parts together so they can safely conduct the fault current likely to be imposed on them, which is exactly what this jumper does. Most of the current takes the lowest-impedance route, the EGC that runs beside its own phase conductors. The remainder divides across the parallel metal, and no single conductor has to carry it all.
If the panelboard can is already bonded to the room bar and the feeder carries its EGC, the jumper completes the mesh between them. It also earns its keep on lightning: when a strike lifts the room bar on the counterpoise, the jumper holds the panelboard's ground near the same potential and protects insulation. The enclosure grounding and bonding guide covers these roles — fault path, electrode and bonding jumper — in more detail.
What does a ground-fault return path have to be?
NEC 250.4(A)(5) calls the return route an effective ground-fault current path: permanent, low-impedance, and able to carry the fault current long enough for the overcurrent device to open, with the earth expressly excluded from it.
In plain terms that means metal all the way, tight at every joint, and sized for the current it will carry. NEC Table 250.122 sets the feeder EGC's size from the rating of the overcurrent device ahead of it. NEC 250.96(A) sets the bonding duty on raceways, cables, equipment and enclosures. The earth's exclusion answers half the question: a counterpoise or a driven rod serves lightning and potential equalization, while fault clearing runs on copper and steel. An entry fitting earns its place in this path only when it is bonded, which is the subject of cable gland earthing and the fault path.
What happens to fault current when metal paths run in parallel?
Current divides between two parallel return paths in inverse proportion to their impedances. At a 4,000 A return with a 0.15 Ω EGC path and a 0.60 Ω steel path, that is 3,200 A on the EGC and 800 A on the steel.
Here is the arithmetic for a 150 ft [46 m] feeder. Standard wire tables put 10 AWG copper at about 1.0 Ω per 1,000 ft at 20 °C, so the EGC return path is 1.0 × 150 / 1,000 = 0.15 Ω. Take 0.60 Ω as the assumed impedance of the bolted steel loop. That value is a stand-in, and a measured loop for your building replaces it.
- EGC branch: 4,000 A × 0.60 / (0.15 + 0.60) = 3,200 A
- Steel branch: 4,000 A × 0.15 / (0.15 + 0.60) = 800 A
The EGC takes 80% because it rides in the same raceway as its phases, which keeps the loop area and the impedance small. The steel loop spans whole bays, and inductance pushes its impedance up. Warm the copper to 75 °C and resistance rises about 0.39% per K, moving 0.15 Ω to roughly 0.18 Ω and the split to about 3,080 A and 920 A. The decision survives the temperature correction unchanged.
Does bonding to building steel replace the feeder EGC?
It cannot: NEC Article 250 states that the structural metal frame of a building shall not be used as the equipment grounding conductor. The feeder EGC sized from Table 250.122 stays with the feeder, and the steel bond supplements it.
That keeps the fault path where the code can size and inspect it. Rows from NEC Table 250.122 for common feeder sizes:
| Overcurrent device ahead of the feeder | Copper EGC | Aluminum EGC |
|---|---|---|
| 20 A | 12 AWG | 10 AWG |
| 60 A | 10 AWG | 8 AWG |
| 100 A | 8 AWG | 6 AWG |
| 200 A | 6 AWG | 4 AWG |
| 400 A | 3 AWG | 1 AWG |
| 600 A | 1 AWG | 1/0 AWG |
Insulated EGCs of 6 AWG and smaller run green or green with one or more yellow stripes (NEC 250.119), so the feeder EGC announces itself at every box. A phase set of 6 AWG copper, rated 65 A at 75 °C in the NEC ampacity table, pairs with that EGC on the example feeder. A driven rod joins as an auxiliary electrode under NEC 250.54: you may connect it to the EGC, and it supplements the fault path rather than replacing it. The steel frame follows the same logic — bonded to, and not used as the conductor.
When does a parallel ground path turn into a problem?
It goes wrong when load current, not fault current, rides the steel. A stray neutral-to-ground bond downstream can push 10 A of imbalance through beam joints all day. At one 0.05 Ω loose joint, that sheds 5 W of heat there.
That heat is current squared times resistance: 10 A × 10 A × 0.05 Ω = 5 W, continuously, at a connection nobody inspects. Fault current on the steel behaves differently: it flows for a moment, the breaker clears it, and the path cools. Load current on steel keeps flowing, finds every loose washer and painted surface on the way home, and heats them for years. NEC 250.6 exists for exactly this. It aims to keep neutral current off metal parts, so neutrals land on an isolated terminal and grounds on the ground bus.
So the split is simple. Parallel fault paths: yes, by design. A second neutral-to-ground point: no, anywhere downstream of the service. Subpanel neutral and ground bonding: keep them apart walks through the isolation, and why a neutral-to-ground bond in a subpanel burns equipment traces the damage at the joints.
How do you make the bond at the panel?
Run a copper jumper from the panelboard ground bar to the room bar, sized no smaller than the largest feeder EGC it parallels. Crimp it in compression lugs and land it on bare metal at both ends.
- Lock out the feeder and prove the absence of voltage at both bars before a wrench turns.
- File or brush each contact point to bare metal; NEC 250.12 takes paint, lacquer and enamel off threads and contact surfaces at grounding and bonding connections.
- Crimp a compression lug on each end, bolt the jumper to both bars, and pull it to the torque the lug or panel label prints.
- Mark the jumper green where it is insulated and 6 AWG or smaller, per NEC 250.119.
- Log the bond on the panel schedule with its size and both landing points.
Bond the can too. The panelboard's ground bar bonds to its enclosure through the factory bonding screw or strap. An auxiliary cabinet joins the same way through its grounding point. Where the raceway itself is the fault path, the can still bonds to the raceway's fittings; bonding a device to a metal box when EMT is the EGC shows that arrangement. For where the bar sits, ground bar placement: enclosure wall vs mounting plate compares the two mounting points, and grounding conductor entry: connector and bond sorts the entry hardware.
What do you specify on the drawing?
Show the bond as a green copper jumper sized to the largest feeder EGC it parallels. Name the room bar as the landing point, and note on the schedule that the neutral bonds only at the service disconnect.
The decision by case:
| Your case | Bond the ground bar to the room bar or steel? | Neutral to the can? |
|---|---|---|
| Service disconnect | Yes — bars, can and electrodes all tie together here | Yes — the one point where they join |
| Downstream panelboard, feeder with EGC | Yes — a jumper matching the feeder EGC | Isolated; remove the factory bonding screw or strap |
| Feeder where the metal raceway is the EGC | Yes — the jumper parallels the raceway; bond the can to the raceway fittings | Isolated |
| Panelboard beside a counterpoise or ring ground | Yes — the room bar ties it in; the EGC still clears every fault | Isolated |
One rule covers every row. The EGC clears the fault, the steel path shares it, and the earth carries neither. The same layout covers the ground bar in an auxiliary cabinet; our stainless enclosures are rated NEMA 4X / IP66 by the manufacturer and not UL listed, and the bond you draw is what makes the cabinet part of the fault path. Panelboard enclosure rating in a pump room: what to pick sorts the rating by environment, and custom stainless enclosures is where you set material, size and cutouts.
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
What size EGC does a 60 A panelboard feeder need?
NEC Table 250.122 gives 10 AWG copper or 8 AWG aluminum for a 60 A overcurrent device. Run it with the feeder phases, land it on the panelboard ground bar, and keep insulated conductors 6 AWG and smaller green or green with a yellow stripe. A jumper of the same copper size to the room bar carries its share of the fault comfortably.
Can the earth or a counterpoise clear a ground fault instead of the EGC?
An electrode at 25 Ω to earth — the resistance at which the code accepts a single grounding electrode — lets a 120 V phase-to-ground fault drive only 120 / 25 = 4.8 A, far below the trip of the example breaker. The feeder EGC returns fault current over metal; the counterpoise and room bars handle lightning equalization and potential control, not fault clearing.
How much of the fault current will the building steel carry?
Set it by impedance. With the worked example's 0.15 Ω EGC loop against a 0.60 Ω steel loop, steel takes 800 A of a 4,000 A return, or 20%. Tighten the steel loop to 0.30 Ω and its share climbs to about 1,333 A, roughly 33%, which is why every joint in that path wants bare metal and real torque.
Can this extra bond upset ground-fault protection at the service?
Your bond cannot upset ground-fault protection: the jumper carries fault current only, and between faults it sees milliamperes of insulation leakage, so the sensor at the service measures no imbalance. Nuisance tripping comes from a second neutral-to-ground bond, which lets load current return on metal paths and unbalances the sensor; that is why the neutral bonds at the service alone.
