Conductor size

The method, and where it stops

The order, and why it is this order

Sizing a conductor is a search, not a formula. There is no closed form, because the ampacity of a candidate depends on the candidate and the voltage drop depends on it again, differently. So the page walks the table from the smallest size upward and asks two questions of each one.

1. Is it big enough for the current? The required current is the load, at 125 % if it is continuous (210.19(A)(1), 215.2(A)(1)). The candidate’s allowable ampacity is computed by exactly the ladder the conductor ampacity page shows — base at the conductor’s own insulation column, ambient correction from that same column, conductor-count adjustment, then the uncorrected 110.14(C) termination cap, and the answer is the lesser. This page calls that page’s engine rather than repeating it, so the two cannot disagree.
2. Does it drop too much? V_d = K × I × L × (R·cos θ + X·sin θ) ÷ 1000, with R and X from Chapter 9 Table 9 at the raceway you chose, and divided by the number of parallel sets. I is the actual load current, not the 125 % figure: the 125 % is a sizing rule, and the drop a load causes is caused by the current it draws.

The first size that clears both is the answer. If no single conductor clears both, the search restarts at two parallel sets, then three, up to ten — and 310.10(G)(1) removes every size below 1/0 AWG from the search the moment sets exceed one.

Voltage drop is a recommendation, not a requirement

The articles are . This matters more than it sounds. A calculator that prints non-compliant against a recommendation is asserting a Code violation that does not exist, which is the same class of error as a green pass over a real one. When this page says the size was decided by voltage drop, it means you are making a design choice — a defensible one, and usually the right one, but not one the NEC obliges you to make.

Which limit governs, and how that is decided

Not by comparing the chosen conductor’s margins. Sizes come in steps, so the smallest conductor that clears a 3 % target often clears it by a wide margin, and reading that margin as “voltage drop was not close” would be wrong. The question is counterfactual: of the sizes below the answer, what did each one fail? If they all failed on current, ampacity governs. If they all failed on drop, voltage drop governs. If different sizes failed on different things, both are load-bearing and relaxing either alone buys you nothing — and the page says so.

The overcurrent device, and the trap in 240.4(D)

The device is sized from the load — the smallest standard 240.6(A) rating at or above the required current — not from the conductor. Two things can then bite.

240.4(D) is a flat ceiling on 14, 12 and 10 AWG that overrides the ampacity table: 15, 20 and 30 A for copper, 15 and 25 A for aluminum. A 12 AWG copper conductor with a 90 °C ampacity of 30 A is still limited to a 20 A device. When the load needs more than that ceiling, the conductor that satisfied ampacity and voltage drop still cannot be protected, and the page says so rather than printing a device rating the Code forbids.

240.4(B) permits the next standard rating up when the conductor’s ampacity does not correspond to a standard rating — but only where the conductors do not supply receptacle outlets and the rating does not exceed 800 A. This page cannot see your receptacles, so it reports the condition rather than assuming it.

The equipment grounding conductor

Table 250.122, indexed by the device rating — not by the phase conductor, and not by the load. Two adjustments follow from the rest of the answer.

250.122(B): where the ungrounded conductors are increased in size from the minimum with sufficient ampacity, the grounding conductor is increased proportionately by circular-mil area. That is exactly what happens when voltage drop decides the size, so on this page the EGC frequently comes out larger than the bare Table 250.122 lookup — and the caveat states the ratio and both sizes.

250.122(F): with parallel sets in separate raceways, each raceway needs a full-size grounding conductor sized from the same device rating, never one EGC divided among them. With all sets in one raceway, a single grounding conductor is permitted. The page follows whichever layout you chose.

Where the numbers come from

Every table on this page is generated from the same source the firm’s own engineers use in the Bellazipeda project-management suite, not retyped for the web. Table 310.16, Table 310.15(B)(1), Table 310.15(C)(1), 240.6(A), Table 250.122, 240.4(D) and Chapter 9 Table 9 are copied by a generator, and a drift test compares the published bytes against a fresh generation, so a stale copy and a hand-edit fail identically. The edition is the stated NEC edition. Jurisdictions adopt on their own schedule and amend freely; check yours.

One gap is worth naming, because it looks like a bug. Table 310.16 prints a 700 kcmil row and Chapter 9 Table 9 does not — its printed rows run … 600, 750, 1000 kcmil. So a 700 kcmil conductor has an ampacity and no published impedance, and this page refuses to grade its voltage drop rather than interpolating one and calling it Table 9. The search skips it and says why.