Feeder and branch circuit design and sizing per NEC

Service, feeder, branch: where each one sits

Power moves through a building in three named stages, and the NEC defines each one in Article 100 by where it starts and stops. The service is the conductors and equipment that bring utility power to the premises, ending at the service disconnect. The feeder is everything between that service, or another source like a transformer or generator, and the final overcurrent device. The branch circuit is the last leg, from that final overcurrent device to the outlets and the load.

The dividing line is always an overcurrent device. Upstream of the last breaker or fuse is feeder. Downstream of it is branch circuit. A conductor leaving the service that feeds a downstream panel is a feeder. The conductor leaving a breaker in that downstream panel to a receptacle is a branch circuit. Same pipe, same building, different rules.

The rules differ because you know different things at each stage. On a branch circuit the overcurrent device sets the size first, and the conductor follows it. On a feeder you work the other way: you calculate the load, then size the conductor and the overcurrent device to carry it. The load that drives the feeder comes out of an Article 220 calculation, and the conductor's heat rating comes off the ampacity tables, both of which have their own guides. This one ties them together.

What is a branch circuit?

A branch circuit is the circuit conductors between the final overcurrent device protecting the circuit and the outlets or the load it serves. That is the Article 100 definition, and the key word is final. Once a conductor passes the last breaker or fuse ahead of the load, it is a branch circuit, and the rating of that overcurrent device names the circuit.

The rating sets everything. A 20 A breaker makes a 20 A branch circuit, and the conductor has to carry at least 20 A after derating and at the termination temperature. The standard residential and light-commercial branches run 15 A on 14 AWG copper, 20 A on 12 AWG, 30 A on 10 AWG, and up from there, with the conductor matched to the breaker, not the other way around.

Where rookies get this backwards is on a long run. They size the conductor to the 20 A breaker, pull 12 AWG, and the equipment at the end runs low because nobody checked the voltage drop over the distance. The breaker rating is the floor for the conductor, not the whole answer. Length can force you larger, and that decision belongs in the calculation before the wire is on the reel.

The branch circuit types

The NEC sorts branch circuits by what they feed, and the type changes the rules you apply. A general-purpose branch circuit serves two or more receptacles or outlets, the ordinary lighting and convenience circuits. An appliance branch circuit serves one or more outlets for appliances and carries no permanently connected lighting that is not part of the appliance. An individual branch circuit serves a single piece of utilization equipment, like a dedicated circuit to a furnace or a water heater.

The individual branch is the one to reach for when the equipment is large, continuous, or sensitive to a tripped neighbor. A dedicated circuit means the load has the whole breaker to itself, so it cannot be knocked offline by something else on the same conductor, and you size it to that one load.

The multiwire branch circuit is its own animal, common enough and misunderstood enough that it gets its own section below. The rest break down cleanly by what is on the other end: shared general-use circuits, appliance circuits, and dedicated individual circuits. Match the type to the load, then size the conductor and the breaker to the type's rules.

The multiwire branch circuit and the shared neutral

A multiwire branch circuit uses two or three ungrounded conductors that share one grounded neutral, with the hots on different phases so the neutral carries only the unbalanced current, not the sum. On a single-phase 120/240 V system the two hots are 180 degrees apart and the neutral carries the difference. On a three-phase wye the three hots are 120 degrees apart. Done right, an MWBC saves a neutral and a pull.

Done wrong, it kills someone. If the two hots land on the same phase instead of opposite ones, the neutral carries the sum of both currents instead of the difference, and an undersized shared neutral overheats. That is the failure that hides until the insulation cooks. Land each hot on a different phase, and on a panel that means adjacent breaker positions, not two breakers on the same bus leg.

The NEC requires a multiwire branch circuit to have a means that disconnects all the ungrounded conductors at once where the circuit originates, commonly cited at 210.4(B). A two-pole common-trip breaker is the clean way to do it. A pair of single-pole breakers with a listed handle tie also satisfies the simultaneous-disconnect requirement. The reason is the neutral: open one hot and leave the other live, and a worker on what looks like a dead circuit meets current returning on the shared neutral.

What is the 125 percent rule?

The 125 percent rule sizes the conductor and the overcurrent device for a continuous load at 125 percent of that load, plus 100 percent of any non-continuous load on the same circuit. A continuous load is one where the maximum current runs for three hours or more. The rule applies to both branch circuits and feeders, and it is the single most-missed step in field sizing.

The reason is heat. A breaker in an ordinary enclosure is rated for 80 percent continuous duty, so a load that sits at full current for hours has to be held to 80 percent of the device rating, which is the same thing as sizing the device at 125 percent of the load. Sizing at 100 percent is allowed only where the assembly, breaker and enclosure together, is listed for 100 percent continuous operation, and most are not.

Run the math on both the conductor and the overcurrent device. If a feeder carries 100 A continuous and 60 A non-continuous, the minimum is 125 percent of 100 plus 60, which is 185 A, for the conductor ampacity and for the overcurrent device. Skip the adder and you have undersized both. The NEC carries this rule for branch circuits in 210.19 and 210.20 and for feeders in 215.2 and 215.3; confirm the section against the adopted edition. The load-calculation guide works the demand side of the same problem.

The feeder, sized to the load it actually carries

A feeder is the conductor set between the service, or another source, and the final branch-circuit overcurrent device. It feeds a downstream panel, a sub-panel, or a single piece of large equipment. Unlike a branch circuit, where the breaker sets the size, the feeder starts with the load, and the load comes from a calculation, not from adding up every nameplate in the building.

That calculation is the Article 220 load calc, and it returns a demand load, the diversified current the feeder will really see, which is smaller than the connected sum because no building runs everything at once. The feeder is sized to that demand, then bumped for the 125 percent continuous adder, then carried into the conductor and overcurrent device selection.

Get the load wrong and everything downstream is wrong. Size a feeder to the connected sum and you pay for copper nobody needs. Size it to the running load and forget the continuous adder and you trip breakers on a hot afternoon. The feeder is where the load calculation turns into a conductor, so the number that feeds it has to be the demand load, sized correctly, with the continuous loads already grossed up.

How do you size a feeder?

Size a feeder in a fixed order, and the order keeps you out of trouble. First, get the calculated demand load in amps from the Article 220 calculation. Second, apply the 125 percent adder to the continuous portion: minimum ampacity equals non-continuous load plus 125 percent of continuous load. Third, pick a conductor whose ampacity at the termination temperature, after any derating for ambient and fill, meets that minimum. Fourth, size the overcurrent device to protect the conductor. Fifth, size the equipment grounding conductor to the overcurrent device. Last, check voltage drop over the routed length and upsize if the run is long.

Each step has its own guide because each step is its own trap. The demand load comes from the load-calculation work. The conductor ampacity and the termination cap come from the ampacity and derating work. What this section adds is the sequence: load first, conductor second, overcurrent device third, ground fourth, voltage drop as the check that can override the conductor on a long run.

The mistake is doing the steps out of order or skipping the check. A feeder sized to ampacity alone and never checked for voltage drop passes inspection and still starves the equipment at the end. A feeder sized for drop but with the wrong demand load is oversized and overpriced. Run the order every time.

The conductor, the ampacity table, and the termination

The conductor has to carry the sized current without overheating, and the number you size to is not the headline ampacity on the wire. It is the ampacity at the temperature the terminations are rated for. Most equipment terminations are rated 75 degrees C, and the NEC limits you to the lowest-rated termination in the circuit, commonly cited at 110.14(C). A 90 degree C conductor on a 75 degree C lug gives you the 75 degree C ampacity, no more.

So the sequence is: take the base ampacity from the table for the conductor, correct it for ambient temperature, adjust it for the number of current-carrying conductors in the raceway, then cap the result at the termination column. The capped, derated number is what has to meet the sized current. That whole process has its own guide, and it is worth reading before you commit a conductor size, because the derating on a hot, crowded run can move the answer by a size or two.

Keep one thing straight here. You can start a calculation from the 90 degree C column to get the most out of the derating, but the final allowed value still caps at the termination temperature. The high column is a head start, not the finish line. The ampacity and derating guide walks the columns; this guide just points you there and moves on.

Matching the overcurrent device to the conductor

The overcurrent device protects the conductor, so its rating cannot exceed the conductor's ampacity, with one practical exception. When the conductor's ampacity does not land on a standard breaker or fuse size, the NEC permits the next standard size up, commonly cited at 240.4(B), as long as the device is 800 A or less and the circuit is not a multi-outlet branch feeding portable plug-in loads. The standard sizes themselves live in 240.6: 15, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 125, 150, 175, 200, and up.

Work an example. A conductor derates to 185 A. The next standard size up is 200 A, and a 200 A device is allowed to protect it under the next-size-up rule. But a conductor that derates to exactly 200 A and the device at 200 A is a clean match with no rounding. Above 800 A the next-size-up rule does not apply, and the conductor ampacity has to equal or exceed the device rating.

On a branch circuit the logic runs the other way, because the device rating is fixed first. A 20 A branch needs a conductor good for at least 20 A after derating. The conductor follows the breaker. On a feeder the conductor follows the load, and the breaker follows the conductor. Same parts, opposite starting points, and mixing up which one you are sizing is how conductors end up under-protected.

Voltage drop on a long feeder

Ampacity sizes a feeder for heat. Voltage drop sizes it for distance, and on a long run distance wins. A feeder that the load current alone would carry on a smaller conductor can still leave the panel at the far end low, because the resistance over the length eats voltage the equipment needed.

Many designs target about 3 percent on the feeder and 5 percent total across feeder plus branch, but those figures are recommendations in NEC informational notes, not enforceable limits. The project specification and the equipment's listed voltage tolerance control the real number. A feeder that runs a few hundred feet at a meaningful load is a voltage-drop problem first, and the fix is usually a larger conductor, a shorter route, or a panel moved closer to the load.

When you upsize a feeder for voltage drop, the equipment grounding conductor has to grow with it, in proportion to the phase conductors. That step gets skipped constantly. The voltage-drop guide carries the formula, the resistance values, and a worked example; the point here is that the run length can override the conductor that ampacity and the load calc would have allowed.

Sizing the feeder and branch neutral

The neutral carries the unbalanced current, the difference between the phase loads, not the full phase current, so a feeder neutral can sometimes be smaller than the hots. The NEC allows a reduced neutral sized to the maximum unbalanced load on the feeder, with limits, and on a balanced three-phase wye the neutral can run smaller than the phase conductors. That is the textbook case.

The textbook case quietly breaks on non-linear loads. Switch-mode power supplies, LED drivers, variable-frequency drives, and the gear in a server room draw current in pulses, and the third harmonic and its multiples do not cancel at the neutral the way the fundamental does. They add. On a three-phase wye feeding heavy electronic load, the neutral can carry more current than any single phase, which is the opposite of what the balanced model predicts.

So for harmonic-rich loads you do not reduce the neutral, and many data and lighting designs run a full-size or even doubled neutral, the 200 percent neutral, as a design choice. The NEC does not impose a flat 200 percent across the board; it prohibits applying the usual neutral demand reduction to the non-linear portion of the load, commonly cited at 220.61, and it leaves the oversize to the engineer and the equipment. The rule of thumb: linear load, the neutral can shrink; non-linear load, the neutral stays full or grows.

Sizing the equipment grounding conductor

The equipment grounding conductor is sized to the overcurrent device, not to the load and not to the phase conductors. You take the rating of the breaker or fuse protecting the circuit and read the equipment grounding conductor off NEC Table 250.122. A 100 A device wants a 8 AWG copper ground, a 200 A device wants 6 AWG copper, and the table runs up from there, with aluminum a size larger than copper for the same rating.

The ground exists to carry fault current long enough to trip the device. If a hot faults to a metal enclosure, the fault current has to return through that conductor fast enough and large enough to open the breaker before the enclosure becomes live. Undersize it and the fault path is too high in impedance, the breaker is slow to trip or does not, and the metal stays energized.

There is one case where the ground grows beyond the table value. When you upsize the phase conductors above the minimum for any reason, including voltage drop, the equipment grounding conductor increases in proportion, commonly cited at 250.122(B), by the same ratio of circular-mil area. Upsize the feeder from 3/0 to 250 kcmil to hold a voltage-drop target and the ground that suited the 3/0 is now undersized. Write the ground change in the same note as the upsize so the next person sees they belong together.

Feeding a sub-panel: four wires, no bond

A feeder to a sub-panel is a four-wire feeder, and the detail that trips people is the neutral-to-ground bond. At the service, the grounded neutral and the equipment grounding system are bonded together, once, at the main. At a sub-panel fed from that service, they must stay separated. Four conductors run to the sub: the ungrounded phases, an isolated neutral, and a separate equipment grounding conductor.

The neutral bar at the sub-panel floats, isolated from the enclosure, and the ground bar bonds to the enclosure. Leave the factory bonding screw or strap in place at the sub-panel and you have created a second neutral-to-ground bond, which puts normal neutral current onto the grounding conductors and the metal. That is a real shock and fire path, and it is one of the most common things an inspector catches at a sub-panel.

The exception is a separately derived system, a transformer or a generator with its own neutral, where a new neutral-to-ground bond is established at the source. That is a different situation with its own rules. For an ordinary sub-panel fed from the building service, the rule is simple and absolute: four wires, neutral isolated, ground bonded, no second bond.

Feeder taps without an overcurrent device at the tap

Normally a conductor is protected at its supply end by an overcurrent device sized to it. A feeder tap is the exception: a smaller conductor tapped off a larger feeder without its own overcurrent device at the point of the tap, allowed under specific rules commonly grouped at 240.21(B). The tap rules exist because forcing a breaker at every tap point is impractical, so the NEC trades the missing protection for tight limits on length and size.

The two field workhorses are the 10-foot and 25-foot tap rules. Under the 10-foot rule the tap conductor can be short and relatively small, with its ampacity at least the load served and at least one-tenth the rating of the feeder overcurrent device, terminating in a single overcurrent device and protected from damage. Under the 25-foot rule the tap conductor has to be larger, with ampacity at least one-third the feeder overcurrent device rating, and it also terminates in a single overcurrent device.

The trap is the size minimums, not the lengths. People remember the foot count and forget that the tap conductor still has to meet the one-tenth or one-third ampacity floor, and that it has to land in a single device that protects it going forward. A tap that is the right length but too small for the rule is a violation that an inspector will find, and a fire that a thin conductor on a big feeder can feed. The tap rules have their own guide; size to the rule, not the memory of it.

Motor branches and motor feeders run by their own rules

Motors do not follow the ordinary 125 percent continuous rule, because the starting inrush and the running heat are different problems. A motor branch circuit conductor is sized at 125 percent of the motor's full-load current, the FLC, taken from the NEC motor tables, not from the nameplate. The short-circuit and ground-fault protection ahead of it is sized much larger, often well above the conductor ampacity, because it has to let the inrush through without tripping while still clearing a fault.

That split, conductor at 125 percent of FLC but overcurrent device sized for inrush, is what makes motor circuits look wrong to someone used to ordinary loads. The big breaker is not protecting the conductor against overload. The overload protection is a separate device, the motor's overloads, sized to the running current. The breaker or fuse handles the short circuit and ground fault only.

A motor feeder, one feeding several motors, is sized at 125 percent of the largest motor's full-load current plus the sum of the full-load currents of the rest. You gross up one motor, the biggest, and add the others at 100 percent. Motor circuits live under their own article with their own tables, so size them from that work, not from the general feeder rules in this guide.

Demand load, not connected load

A feeder is sized to the demand load, not the connected load, and confusing the two is the most expensive sizing error there is. The connected load is the sum of every nameplate the feeder could serve. The demand load is what it will actually carry at once, which is smaller, because the NEC demand factors recognize that a building does not run every receptacle, every range, and every motor at full tilt simultaneously.

The demand factors live in Article 220, and they cut the neutral load, the appliance load, the range and dryer loads, and more, each by its own schedule. The result is a calculated demand current that is the real basis for the feeder. Size to the connected sum instead and you buy conductor and gear for a load that will never appear.

The order matters: apply the demand factors first, to get the calculated load, then apply the 125 percent continuous adder to the continuous portion of that calculated load. The demand factors shrink the number; the continuous adder grows part of it back. Run them in the wrong order or skip one and the feeder is wrong in one direction or the other. The load-calculation guide works the demand side in full.

The raceway and the panel the feeder lands in

Once the conductors are sized, two more things have to keep up: the raceway that carries them and the panel they land in. The raceway has to hold the conductors within the NEC fill limits, and the fill drives back into the sizing, because the number of current-carrying conductors in a raceway is what triggers the bundling adjustment that derates the ampacity. Fill and ampacity are the same conversation from two ends. The conduit-fill work has its own guide.

The panel has to be rated for the feeder. A panelboard carries a busbar rating and a main rating, and the feeder overcurrent device and the conductors landing on the main lugs cannot exceed what the bus and the lugs are listed for. A 200 A feeder lands on a panel rated at least 200 A, with a main breaker or main lugs and overcurrent protection sized to match. Land a feeder on a panel whose bus is too small and you have a listed assembly running past its rating, which is a violation and a heat problem.

Check the panel's series rating and its interrupting rating too, against the available fault current at that point. The feeder conductor and its overcurrent device are only half the install. The gear they connect to has to be rated for both the load and the fault, and the panelboard work covers that side.

Field example: feeder to a 200 A sub-panel

A 208Y/120 V three-phase feeder runs to a sub-panel. The Article 220 calculation returns a demand load of 160 A, of which 100 A is continuous and 60 A is non-continuous. The run is 180 ft one way through conduit. Work the order.

Apply the 125 percent rule: 125 percent of 100 A continuous is 125 A, plus 60 A non-continuous, gives 185 A minimum for the conductor and the overcurrent device. Pick the conductor at the 75 degree C termination column: 3/0 copper is good for 200 A at 75 degrees C, above the 185 A minimum, assuming the ambient and fill do not derate it below that. Size the overcurrent device: 185 A is not a standard size, so the next standard size up is 200 A, and a 200 A device protects the 3/0 conductor cleanly. Size the equipment grounding conductor to the 200 A device: 6 AWG copper from Table 250.122.

Now the voltage-drop check, using the 160 A load current over the 180 ft run on 3/0 copper at about 0.0766 ohms per 1000 ft. The three-phase formula gives 1.732 times 160 times 180 times 0.0766, divided by 1000, about 3.8 V, or 1.8 percent of 208 V. That leaves room under a 3 percent feeder target for the branch circuits downstream. If the run had been 350 ft, the drop would have pushed past target and forced the conductor up a size, and the 6 AWG ground would have had to grow with it.

The branch circuits you wire most

A handful of branch circuits show up on nearly every job, and each carries a rule worth knowing cold. The kitchen small-appliance circuits are a pair of 20 A circuits required to serve the counter receptacles, kept off the lighting so a toaster and a mixer do not dim the room. Bathroom and laundry receptacles get their own 20 A circuits for the same reason.

Dedicated individual circuits go to the loads that should not share: the furnace, the water heater, the dishwasher, the EV charger, the sump pump. A dedicated circuit means a fault or a trip on something else cannot take the equipment offline, and it lets you size the conductor and breaker to the one load. EV charging in particular is a continuous load with a long run, so it gets the 125 percent adder and a voltage-drop check together.

Then there is the protection layer. GFCI protection guards against shock where water and ground are near, the kitchen, bath, garage, outdoors, and basements. AFCI protection guards against arcing faults in the wiring of living spaces. Both are required by area and circuit type, both are enforced, and both have expanded across recent code cycles, so confirm the requirement against the adopted edition. The GFCI and AFCI rules have their own guide; the point here is that the protection is part of the branch circuit's design, not an add-on.

Large feeders, busway, and the data center case

On big distribution the feeder stops being a few conductors in a pipe and becomes parallel conductor sets or busway. Where a single conductor cannot carry the load, the NEC permits paralleling conductors in each phase, with strict rules: same length, same material, same size, same insulation, same termination, so the current divides evenly. Get one parallel leg longer than the others and it carries less current while the rest carry more, and the math that says they share equally stops being true.

Busway, a bus duct, replaces conductors with rigid bus bars in an enclosure for high-current distribution, and it is sized and tapped by its own rules. It is common in data centers and large plants where the feeder current runs into the thousands of amps and pulling that much conductor is impractical.

The data center is also where the neutral and the harmonics from the neutral section come due. Rows of switch-mode power supplies are about as non-linear as a load gets, so the neutral stays full or doubled, the conductors run hot from continuous duty, and the demand load is close to the connected load because the equipment really does run around the clock. The diversity that shrinks a feeder in an office building mostly is not there. Size these to run continuously at high load, because they will.

Common mistakes

• Sizing a feeder or branch without the 125 percent adder on the continuous load.
• Matching the conductor to the wrong number, so the overcurrent device over-protects or under-protects it.
• Ignoring the termination temperature and sizing off the 90 degree C column the lug will not honor.
• Skipping the voltage-drop check on a long feeder and starving the equipment at the end.
• Reducing the neutral on a harmonic or non-linear load, where it should stay full or grow.
• Sizing the equipment grounding conductor wrong, or not growing it when the phases are upsized.
• Sizing a feeder to the connected load instead of the calculated demand load.
• Leaving the neutral-to-ground bond in a sub-panel, putting neutral current on the grounding system.
• Remembering a tap's length but forgetting its one-tenth or one-third ampacity minimum.

Field checklist

What to document

A feeder or branch circuit can be sized perfectly and still leave you nothing to show for it if the math never made it into the record. When the equipment at the end runs low or the inspector asks why the conductor is the size it is, the record is the answer. Capture the load, the sizing math, and every choice that drove the conductor up or held it where it is.

For each circuit, write down what kind it is, the number it was sized by, and the key rule that governed it. Tie the conductor, the overcurrent device, and the ground together in the same record, because they were sized as a set and they have to stay a set. If you upsized for voltage drop, record what drove the size and the matching change to the ground.

Standards and references

The NEC, NFPA 70, is the framework, and it splits feeders and branch circuits across separate articles. Branch circuits live in Article 210, including the conductor and overcurrent sizing at 210.19 and 210.20 and the multiwire rules at 210.4. Feeders live in Article 215, with the conductor and overcurrent sizing at 215.2 and 215.3. The load calculation that sizes the feeder comes from Article 220.

Overcurrent protection, the conductor-to-device match and the next-size-up rule, sits in Article 240, with the standard device sizes at 240.6, the next-size-up allowance at 240.4(B), and the feeder tap rules at 240.21. The equipment grounding conductor is sized from Table 250.122, including the proportional increase at 250.122(B). The termination-temperature limit is at 110.14(C), and conductor ampacity comes off the tables in Article 310. The voltage-drop figures, 3 percent and 5 percent, are recommendations in informational notes, not mandates.

Article and section numbers move between code cycles, and the percentages and rules can shift with them, so confirm every reference against the edition the jurisdiction has actually adopted and any local amendments before you cite it on a submittal. Equipment listings under UL and the manufacturer's instructions can impose tighter limits than the NEC, and where they do, the listing governs.

Units, terms, and the words that get crossed

Feeder and branch circuit work runs on a small vocabulary that is easy to cross, especially the load terms and the conductor terms.

Current is in amps, and the load that drives sizing is the calculated demand load, not the connected load. Conductor size is AWG for smaller conductors and kcmil, thousand circular mils, for larger ones, with mm squared on metric drawings. The overcurrent device, the breaker or fuse, is the OCPD. The equipment grounding conductor is the EGC. Keep the grounded conductor, the neutral, separate in your head from the grounding conductor, the ground; they sound alike and they are not the same wire.

Branch circuit

Conductors from the final overcurrent device to the outlets or load

Feeder

Conductors from the service or source to the final branch-circuit overcurrent device

OCPD

Overcurrent protective device, the breaker or fuse that protects the conductor

Continuous load

A load at maximum current for three hours or more, sized at 125 percent

Demand load

The diversified load after Article 220 demand factors, what the feeder really carries

EGC

Equipment grounding conductor, sized to the OCPD from Table 250.122

MWBC

Multiwire branch circuit, two or three hots sharing one neutral, with simultaneous disconnect

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is an electrical load calculation?
• What is conductor ampacity?
• What voltage drop is acceptable?
• What size wire for a 48A charger?
• What does it mean to parallel conductors?