5 mistakes to avoid when installing a panel sub-feed

5 mistakes to avoid when installing a panel sub-feed, the field-ready guide for working electricians.

Sub-feeding a panel from another panel is routine work, but it's also where shortcuts cause callbacks, failed inspections, and the occasional fire. The NEC has tightened up grounding and overcurrent rules over the last few cycles, and a panel that passed in 2014 won't pass today. Here are five mistakes that show up on inspection reports again and again.

1. Bonding the neutral at the sub-panel

This is the classic. The main bonding jumper belongs at the service disconnect, period. Once you leave that point, the grounded (neutral) and grounding conductors must stay separate. NEC 250.24(A)(5) is explicit: no grounded conductor connection on the load side of the service disconnecting means, except as permitted in 250.30 and 250.32.

At the sub-panel, pull the green bonding screw. Pull the bonding strap. Float the neutral bar on its insulated standoffs and land grounds on a separate ground bar bonded to the can. If you energize a sub-panel with the neutral bonded, you've put parallel current paths on every metal raceway, every ground wire, and every box in the system.

If you can read voltage between the neutral bar and the ground bar at a sub-panel with a meter, something is wrong upstream. They should read zero or near zero, and the bars should never be tied together at that location.

2. Running three wires instead of four

Three-wire feeds to a separate structure or sub-panel were allowed for decades under the old 250.32 exception. That exception is gone. Since the 2008 cycle, you must pull four conductors to a sub-panel: two ungrounded, one grounded (neutral), and one equipment grounding conductor. NEC 250.32(B)(1) is the current rule.

This trips up guys working on older homes. You find an existing three-wire feed to a detached garage, the homeowner wants a panel swap, and now you're on the hook to pull a fourth conductor or run a separate EGC. Don't reuse the old feeder unless it actually has four conductors and the EGC is sized correctly per Table 250.122.

  • Two hots, sized per the load calculation
  • One neutral, full-size if it carries unbalanced current
  • One equipment grounding conductor, sized per 250.122
  • All four in the same raceway or cable assembly

3. Skipping the OCPD at the supply end

Feeder taps and sub-panel feeds need overcurrent protection sized to the feeder ampacity, located at the point where the conductor receives its supply. NEC 240.21(B) covers tap rules, and 408.36 requires a panelboard to be protected on the supply side by an OCPD not exceeding the panelboard rating.

The mistake is feeding a 100A sub-panel from a 200A breaker because that's what was available in the main. The conductor and the panel must be protected at their rating. If the sub-panel is rated 100A, the breaker feeding it lands at 100A or less, sized to the conductor ampacity per 310.16 with all the derate factors applied.

The 10-foot and 25-foot tap rules in 240.21(B) exist for a reason, but they have specific conditions: terminations, conductor sizing, and physical protection. Don't invoke a tap rule unless you've actually met every condition in writing.

4. Undersizing the equipment grounding conductor

The EGC sizes off the upstream OCPD, not off the conductor. That's a common reversal. NEC Table 250.122 gives the minimum EGC size based on the rating of the automatic overcurrent device in the circuit ahead of the equipment.

If you upsize the ungrounded conductors for voltage drop, 250.122(B) requires you to upsize the EGC proportionally. A lot of long sub-feeds get caught here. You bumped the hots from #2 to 1/0 to handle a 150-foot run, but you left the EGC at the table minimum for a 100A breaker. That's a violation.

Run the math. Cross-sectional area of the upsized ungrounded conductor divided by the cross-sectional area of the original conductor, then multiply by the table-minimum EGC area. Round up to the next standard size.

5. Filling the panel without a load calculation

A 100A sub-panel is not a 100A sub-panel just because it has a 100A main. If the calculated load on that sub-panel exceeds the feeder ampacity, you're out of compliance the moment the inspector runs the numbers. Article 220 governs this, and 220.40 requires the feeder to carry the calculated load.

For a residential sub-panel feeding general lighting, small appliance circuits, and a couple of dedicated loads, run the standard calculation in 220.42 through 220.55. For a commercial or industrial sub-feed, the optional methods in Part IV may apply, but only if the conditions match.

  1. Add up all continuous loads at 125 percent
  2. Add non-continuous loads at 100 percent
  3. Apply demand factors per Article 220
  4. Compare against the feeder ampacity after derate
  5. Compare against the panel bus rating

A panel schedule taped inside the dead front, filled in legibly with circuit numbers and load values, makes the inspector's job easy and gives the next electrician something to work from. That's not a code requirement in every jurisdiction, but it should be your standard practice.

Final field check before energizing

Walk it before you close the door. Neutral floated, bonding screw out, EGC landed on the ground bar, four conductors in the feeder, breaker sized to the panel rating, torque values hit on every lug per the manufacturer's listing and 110.14(D). Megger the feeder if the run is long or the conditions were rough during pull.

The five mistakes above account for the majority of sub-panel red tags. None of them are hard to avoid if you slow down at the rough and the trim. Pull the code book when you're not sure, not after the inspector writes it up.

Get instant NEC code answers on the job

Join 16,400+ electricians using Ask BONBON for free, fast NEC lookups.

Try Ask BONBON Now