A 30 A load 15 feet from a 12 V battery is ordinary: a compressor fridge, a small inverter, a winch. The wire that would be generous on a household circuit is badly undersized here. With stranded copper at 75 C, 10 AWG loses 1.12 V on that run, 9.29 % of the supply, and turns 33.5 W into heat. The smallest size that keeps the drop under 3 % is 4 AWG (0.28 V, 2.31 %).
The same 360 W at 120 V draws 3 A, and over the same 15 ft even 18 AWG drops only 0.59 %. At 24 V, 10 AWG passes; at 48 V, 16 AWG. Doubling the voltage halves the current, and a quarter of the copper gives the same percentage drop.
What the tool does
Voltage Drop Calculator takes the source voltage, the load in amps or watts and the one-way run, and returns, for every AWG and kcmil size (or every metric size from 0.75 to 300 mm²):
- the drop in volts and percent and the voltage left at the load;
- the smallest size within your target, never below a minimum you enter from your ampacity table;
- the longest run each size allows at this current, and the most current it allows over this run;
- the watts lost in the wire and, given hours per year and a price, what those losses cost.
It handles DC, single-phase and three-phase (1.732 instead of 2, line to line), copper or aluminium, solid or stranded, any conductor temperature, parallel sets, and a power factor with an optional reactance.
Computed, not copied
The resistance comes from first principles: the conductor area from the AWG geometric series (or kcmil, or the metric nominal area), the resistivity of annealed copper (100 % IACS) or aluminium conductor grade (61 % IACS) at 20 C, the temperature coefficient for the conductor temperature you choose, and 2 % for stranding. That means no code table is reproduced, and the method is on the page for anyone to check. The cost is honesty about precision: published resistance tables run a few percent higher for some sizes, and the maximum resistances in the metric conductor standard are higher than the nominal area implies. So the page takes a resistance from your own table for the size you check, and uses it instead.
What it does not do
It does not check ampacity, which is the other half of choosing a wire: the conductor must carry the current without overheating, and that comes from the table your code uses with its corrections. The size you install is the larger of the two answers. It also leaves out skin effect in large AC conductors, reactance unless you enter it, motor starting dips, and the drop across fuses, breakers and crimped lugs, which on a 12 V system can be a real share of the total. Every number is checked by hand arithmetic in its tests, not against a manufacturer's tool.
Why this one
The repository has a service load check and a PV string sizer, and both stop short of the conductor: the string sizer lists voltage drop as "not modelled". Wire sizing is the next question an electrician, a solar installer or a van builder asks, and the answer they usually get is a chart that assumed someone else's run length and acceptable drop. A walk-through of the 12 V example is on the tool's guide page. The engine is a pure function an agent can call: node platform/agents/tools.mjs call voltage-drop-calculator analyze --input '{"volts":24,"load":40}'.
Not electrical advice. The tool does arithmetic on the numbers you type; you are responsible for checking its output against your code and your equipment.