Metal conduit as equipment ground between meter disconnects

Short answer: yes. Wrench-tight metal conduit and the enclosures it links carry ground-fault current between your meter trough and each tenant disconnect, with no pulled wire at all. The multimeter beep is not the proof; tight joints are. Two jobs stay outside the conduit's reach: the tenant water pipe at the far end of the building needs its own bond jumper, and the utility may still want a conductor in the meter loop. NEC Table 250.122 sizes every wire you add, from the breaker ahead of the run.

Can metal conduit serve as the equipment ground between a meter and a disconnect?

Yes — the National Electrical Code lists rigid metal conduit, intermediate metal conduit and EMT among its permitted equipment grounding conductors, so the metal path from your meter trough through each tenant disconnect carries the fault with no green wire in it. Bolted steel does that job for the few cycles a breaker needs to clear. NEC 250.96(A) extends the duty to the boxes: metal parts intended to serve as equipment grounding conductors — raceways, cables, equipment and enclosures — must be bonded together so they can safely conduct any fault current likely to be imposed on them. The trough, the meter cans and every disconnect enclosure become parts of the circuit, the way a panelboard enclosure can act as an equipment bonding jumper on its own.

For the fundamentals — why fault current has to return over metal, and why soil never does the job — the grounding and bonding guide covers the whole picture.

What makes the raceway count as a real ground path?

Every coupling, hub and locknut stays wrench-tight, every wall stays metal, and NEC 250.12 calls for paint and enamel to come off the threads and contact surfaces at grounding and bonding connections. A fitting designed to cut through the coating needs no stripping; everything else gets clean bare metal. Your 1980 bank fails here more often than anywhere else: decades of repaint and corrosion leave joints that beep on a meter and choke a fault.

Three fittings decide each box:

• Threaded hubs and standard locknuts on smooth, bare knockouts carry the bond once a wrench seats them.
• Concentric or punched knockouts leave slivers and air gaps, so fit a listed bonding bushing with a jumper instead.
• Bonding locknuts and bonding bushings replace the plain kind wherever the wall is painted.

Which hub passes at service equipment, and what the inspector looks for at each one, is its own question; the listed conduit hubs for bonding article walks through it.

Does a multimeter continuity reading prove the path is good?

No — the beep proves only that metal touches metal at 9 V, while a real path has to stay near zero ohms while carrying hundreds of amperes until the feeder breaker trips. NEC 250.4(A)(5) names the job in code language: an effective ground-fault current path, and the earth is not considered part of it.

The arithmetic shows why a beep is not enough. A corroded locknut that reads 2 Ω end to end limits a 240 V fault to 240 ÷ 2 = 120 A, below the 300-600 A magnetic pickup of a typical thermal-magnetic 60 A breaker. The fault then cooks the joint while the handle sits still. A clean 25 ft run of 1 in EMT contributes a few tens of milliohms, so anything near a whole ohm is a joint problem, not geometry. Keep your meter check — end to end, covers in place — and treat 1 Ω on a short run as a joint to open and remake.

Do you need a separate ground wire in the PVC run to the sub-panel?

Yes — PVC offers no metal path, so the feeder carries a wire equipment grounding conductor sized from the breaker ahead of it. NEC Table 250.122 gives the size; the table below shows the copper and aluminum rows. In a 1980 meter bank this is usually what you find already: a green wire from the disconnect enclosure running through PVC to your tenant sub-panels.

NEC Table 250.122, copper and aluminum columns:

The wire does one job unbroken: land it on the ground bar at the sub-panel, never on the neutral bar, and size it from the breaker ahead, not from the connected load. Where a run mixes steel conduit with a PVC section, the wire bridges the plastic part while the raceway still counts on the metal part. The EGC from service disconnect to sub-panel article pairs the wire with the bonding jumper at the panel.

What does the utility want inside the meter loop?

Ask the serving utility before you count on the raceway alone through the metering section: many require a grounding conductor pulled into the meter loop, sized from the largest service conductor. The metering is their property, their shop pulls meters for test, and their rule sits on top of the code as an addition to your path, not a rival to it.

The code's own work at the service stays as built: the neutral bonds to ground once, at the main switchgear, and every tenant disconnect downstream keeps neutral and ground separate. Why the neutral bond sits only at the service disconnect explains the split; where the grounding electrode conductor goes with multiple service disconnects covers the electrode side of a bank like this one.

Does the tenant water pipe need its own bond jumper?

Yes — the NEC requires metal water piping in the building to be bonded to the grounding system, and the jumper has to bridge everything that breaks the metal: water meters, dielectric unions, plastic filters. One uncut conductor crosses every insulating gap.

On the service side the work is usually done: the buried main counts as a grounding electrode when 10 ft [3 m] or more of it lies in direct contact with the earth, and the bond from that pipe to the electrode system lives in the meter room, right next to the jumper to the water line. How the grounding electrode conductor lands on the service enclosure shows that landing detail.

Everything past that bond is interior pipe: the risers and the copper runs past the tenant water meters at the far end of your building. Left unbonded, a riser sits at a different voltage from the equipment beside it during a fault or a storm, which is the touch-potential shock that stings a wet hand. Bonding makes pipe and enclosure rise together, so nothing bridges the difference through a person; a lightning strike itself is a surge problem, not a bond problem. Clamp one end of the jumper to bare pipe, the other to the disconnect enclosure or another point on the effective path.

What size jumper bonds interior water pipe at a tenant unit?

Size the jumper from the overcurrent device ahead of the bond point — the same row of NEC Table 250.122 that sized the feeder wire, taken from the aluminum column if you pull aluminum (8 AWG on the row used below).

Worked example: one tenant unit. Assume a 60 A tenant disconnect, 120/240 V single-phase, 25 ft of 1 in EMT from the meter trough, a copper cold-water line 6 ft from the enclosure, a water meter and a dielectric union between the two, and a whole fault loop near 0.1 Ω.

• Raceway path: no wire needed between trough and disconnect; the steel adds a few tens of milliohms over 25 ft, and wrench-tight couplings add a few milliohms each.
• Wire instead: the table row above gives 10 AWG copper for this feeder.
• Water-pipe jumper: same row, about 1.0 Ω per 1000 ft at 20 °C, so 6 ft is 6 × 1.0 ÷ 1000 = 0.006 Ω (6 mΩ) — noise next to the path.
• Fault check: at 0.1 Ω the 240 V fault drives 240 ÷ 0.1 = 2,400 A, about 40 times the handle rating and far above the magnetic pickup, so the breaker clears in a few cycles and the pipe never carries the fault.

The inputs you replace are the breaker rating, the system voltage and the jumper length. A 100 A feeder moves the jumper one row up; a 30 A feeder stays on the same row; pipe diameter and clamping point change nothing. The bond to the electrode portion in the meter room is a different conductor, sized from the service conductors rather than from this table.

How do you prove the path on a 1980 building before you sign it off?

Walk the run and put one number on it: end to end, trough to last disconnect, the reading should sit well under 1 Ω on a short run. The metal of 25 ft of 1 in EMT contributes a few tens of milliohms, so a whole ohm is a joint, not geometry. Lock out the feeder and confirm the absence of voltage before any cover comes off; live lugs sit behind these troughs.

• End to end: milliohms pass; a whole ohm means a loose locknut, a corroded coupling or a painted hub, so open and remake that joint.
• Inside each box: bushings seated, locknuts wrench-tight, no tape hiding a plastic-coated fitting.
• The retrofit trap: one PVC repair or a liquidtight whip spliced into the steel run breaks the raceway path; pull a wire through that section, sized from the table for the feeder ahead.
• Concentric knockouts: each gets a listed bonding bushing and jumper, the same detail that decides MC-cable terminations on concentric knockouts.
• Water side: trace each tenant riser and list every meter, union and filter; the jumper bridges them in one run.
• Meter room: confirm the pipe bond to the electrode system is present and tight, and get the utility's answer on its own conductor.

When your survey finds a corroded trough or a wall you cannot trust, replace the box instead of fighting it: welded 304 stainless gives bare metal inside and out and puts every hub on clean steel, delivered in 10 business days from your drawing approval. The custom stainless enclosures page takes your size and your cutouts, hub locations included.

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.

Questions