The Detached Garage with a Bonded Neutral

Why this matters

A detached garage fed by a 4-wire feeder is supposed to keep its neutral and its equipment ground apart, joined only back at the house's service. Leave the bonding screw in the garage subpanel and the garage still works perfectly: lights steady, breakers hold, every receptacle tests correct. What changes is invisible. Part of every amp the garage uses comes home on the equipment grounding conductor and on the metal water line instead of the neutral. This composite case was found because a customer mentioned a tingle in passing, and it was proved and closed with a clamp meter.

The call

The ticket was to move the ceiling receptacle for a new garage door opener in a detached garage about 110 ft from the house. While the tech was on the ladder, the homeowner mentioned that a week earlier, rinsing a part at the utility sink with the space heater running, he had felt "a little buzz" when he put a wet hand on the steel workbench. Once, and not since.

The garage: a 60 A, 2-pole breaker in the house panel feeding a 4-wire feeder in PVC (two 6 AWG copper hots, a 6 AWG copper neutral, a 10 AWG copper equipment grounding conductor, which is the 250.122 size for a 60 A device), a 12-space subpanel with a separate ground bar, a ground rod outside, and a copper water line from the house to the garage sink, bonded at both buildings.

A tingle report on grounded metal is treated as a fault until proved otherwise. The tech asked the homeowner not to use the bench or the sink with wet hands until the visit was finished, finished the opener receptacle with its circuit off and proved dead, and then went looking.

The reading that said everything was fine

The first measurement was the obvious one: steel bench to the ground hole of the bench receptacle, high-impedance meter, space heater running on high on the bench circuit. It read 0.0 V. Bench to the sink faucet: 0.0 V.

Both readings were true and neither meant anything. The bench, the receptacle ground, the faucet and the garage panel enclosure were all bonded to each other. If the whole garage grounding system had been lifted above the house's by a volt, or by fifty, a meter with both leads on that system would still read zero. A meter reading voltage between two points on the same suspect system cannot see that system's voltage.

A reference outside the problem

The tech drove a 2 ft test stake into damp soil about 25 ft from the garage, away from the ground rod and the water line, and ran a long test lead from it to the meter. Bench to stake:

• Heater on: 0.8 V.
• Heater off: 0.3 V.

Small, and it moved with the garage's load. A garage whose grounding system carries no current would read the same with the heater on or off, because nothing in the garage's own circuits would be lifting it.

The size was honestly small. Whether 0.8 V, with a wet hand on cold steel and wet shoes on a slab, is what the homeowner felt is uncertain, and the tech said so rather than claiming the buzz was explained. What the dependence on load did prove was that garage load current was flowing on something it should not be.

The second tell, at the opener receptacle

The opener receptacle was on its own circuit with nothing plugged in. With the heater running on the bench circuit, the tech read neutral to ground at that receptacle: 0.0 V.

At a correctly wired detached building that reading should not be zero. The garage neutral bar sits above the house neutral bar by the feeder neutral's voltage drop, while the garage ground bar, carrying no current, stays at the house's ground potential. Expected drop: NEC Chapter 9, Table 8 gives 0.491 ohm per 1,000 ft for 6 AWG stranded copper, so 110 ft is 0.110 x 0.491 = 0.054 ohm, and 12.5 A of heater current across it is 0.054 x 12.5 = 0.68 V. A neutral-to-ground of 0.0 V means neutral and ground are joined somewhere downstream of the house. It is the same signature as a bootleg receptacle, a building's worth bigger. See related: How to Check Receptacle Wiring with a Meter.

Following the current with a clamp

 HOUSE SERVICE PANEL               GARAGE SUBPANEL
 +----------------+  feeder N      +----------------+
 | N bar          |----------------| N bar          |
 |   | main bond  |  feeder EGC    |   | bond screw |
 | G bar          |----------------| G bar          |
 +---+------------+  (110 ft)      +---+------------+
     | water pipe bond                 | water pipe
     |                                 | bond
 ====+====== copper water line ========+====

The bond screw ties the garage N bar to the garage G bar. From there, garage neutral current has three paths back to the house's main bond: the feeder neutral, the feeder equipment grounding conductor, and the water line through both bonding jumpers.

The clamp readings were taken with the heater on, at about 12.5 A on one leg and nothing else running, so the garage's whole neutral current was that heater's return. Clamping inside the garage subpanel meant an open, energized enclosure: qualified-person work under 29 CFR 1910.333 with the arc-rated PPE the shop's NFPA 70E risk assessment assigns to that panel, jaws around one conductor at a time, hands clear of the lugs.

• Feeder neutral: 7.6 A.
• Feeder equipment grounding conductor: 2.9 A.
• Water line, clamped on the exposed copper at the garage entry: 2.0 A.

7.6 + 2.9 + 2.0 = 12.5 A, the heater's current, all of it accounted for, and 2.9 + 2.0 = 4.9 A of it, about 39 percent, on conductors that are not supposed to carry current at all.

What was in the garage panel

With the 60 A feeder breaker at the house off and locked, the garage panel was proved dead live-dead-live, the tester proven on a known live source before and after, per NFPA 70E Article 120 (120.5 in the 2021 edition, renumbered in 2024), with work practices under 29 CFR 1910.333(b)(2). The panel's deadfront came off.

The neutral bar carried the green bonding screw the panel ships with, threaded through the bar into the enclosure. The feeder equipment grounding conductor, all the branch grounds, the ground rod's conductor and the water line bond required by 250.104 were correctly on the separate ground bar, which is bonded to the enclosure. So the panel's two bars, correctly separated by whoever wired it, had been joined by one screw through the can. It is a common form of this defect, because the screw is factory-installed and a panel used as service equipment needs it.

NEC 250.32(B)(1), NEC 2023 (2020 where the section moved), as adopted, requires in substance that a building supplied by a feeder with an equipment grounding conductor not connect the grounded conductor to that equipment grounding conductor or to the grounding electrode, with an exception for some existing installations fed without an equipment grounding conductor; check the current text. NEC 250.142(B) carries the same prohibition for the load side of the service disconnect generally, and 250.6 is the objectionable-current section that explains why.

Why the bond mattered more than its volts

The 0.8 V was not the danger. Three things were:

• It hid a failing neutral. With the bond in, a loose or burning feeder neutral produces no flicker and no complaint, because the equipment grounding conductor and the water line simply take more of the current. The neutral can fail completely and nobody finds out until something else does.
• It put current on the water line. A plumber who cuts that copper line, or a homeowner who replaces it with plastic, opens a current-carrying conductor. If it happens to be the last good path after a neutral has failed, the cut is where the return current arcs and the garage side of the pipe rises toward line voltage.
• It defeats the separation the equipment ground depends on. The equipment grounding conductor exists to carry fault current only, so its voltage at the far end means something. Once it carries load current, it does not.

The fix

With the garage panel still proved dead, the bonding screw came out and was taped to the inside of the door with a note, since it belongs to the panel's listing. Then the isolation check, which has one trap: with the feeder neutral connected, an ohmmeter from the garage neutral bar to the enclosure reads near zero through the house's main bonding jumper 110 ft away. So the feeder neutral was lifted at the garage first: neutral bar to enclosure read OL, and the neutral was re-landed and torqued to the value marked on the panel with a calibrated tool per NEC 110.14(D). Every branch white was confirmed on the neutral bar and every bare or green on the ground bar.

Verification by clamp

The deadfront went back on, everyone stood clear, and the feeder breaker at the house went back on with the tech to the side of the panel. The deadfront then came off again for the clamp readings, under the same qualified-person controls and PPE as the first set. Heater on, same 12.5 A load:

• Feeder neutral: 12.5 A.
• Feeder equipment grounding conductor: under 0.05 A, the meter's floor.
• Water line at the garage entry: under 0.1 A.
• Neutral to ground at the opener receptacle, heater running on the bench circuit: 0.7 V, matching the 0.68 V computed for the feeder neutral.
• Bench to the test stake: 0.3 V with the heater on and 0.3 V with it off.

The bench no longer moved with the garage's load, the opener receptacle now showed the feeder drop it should, and all of the return current was on the neutral. The 0.3 V that remained was the house's own neutral-to-earth voltage, which belongs to the utility system and every house has some of it. The ticket recorded the before and after clamp readings side by side, and the homeowner was told the buzz could not be pinned on the screw with certainty, while the screw was a real defect worth the visit either way.

References

• NEC 2023 (2020 where the section moved), as adopted: 250.6, 250.32, 250.104, 250.122, 250.142(B), 110.14(D); Chapter 9 Table 8
• NFPA 70E, Article 120 (120.5 in the 2021 edition, renumbered in 2024), current edition as your safety program adopts it
• 29 CFR 1910.333(b)(2), deenergized parts
• See related: Equipment Grounding Conductor Sizing (NEC 250.122)
• See related: How to Check Receptacle Wiring with a Meter