How our calculations are made and checked

One engine for every number

Each calculator is a small, self-contained module with typed inputs and a single compute function. The same function produces the live result, the worked steps under it, the examples and tables on the page, the print sheet and the text you copy. Because nothing is typed in twice, a table can never disagree with the calculator next to it.

Units and precision

• Calculations run in full double precision. Values are rounded only for display, never between steps.
• Displayed results usually show three or four significant figures, with SI prefixes for electronics (4.7 kΩ, 2.55 mA).
• Imperial and metric inputs are converted exactly (1 ft = 0.3048 m; 1 mm² = 1,973.5 circular mils).
• Where a code method is defined in its own units (for example NEC voltage drop in ohms per 1,000 ft), we compute in those units so our numbers match the published method.

Rounding and borderline results

Sizes are always rounded up to the next standard conductor, raceway or device size. When a result lands on or very near a boundary (for example a proportional calculation that comes out a fraction of a percent over a standard size), the calculator shows a warning that explains the edge and which way we resolved it, instead of silently picking one.

Ambient-temperature and bundling factors are applied to the conductor’s own insulation rating (for example the 90 °C column for THHN or XHHW-2), and the result is then limited to the temperature rating of the terminations, as 110.14(C) allows. That is the method the 2026 edition of NEC Annex D Example D7 uses. Older printed examples that apply the factor to the 75 °C column can arrive at a larger conductor; where the two methods differ, the result says so.

Code editions

Code-based calculators let you choose the edition: NEC 2017, 2020, 2023 or 2026; IEC 60364; or BS 7671:2018+A4:2026 (our BS 7671 table data was compiled against A2 and A3 and is being re-checked against A4; any difference is fixed and logged on the corrections page). Server-rendered pages default to NEC 2023, the edition adopted by the most US states, and your browser can switch to the edition in force where you work. When rules differ between editions (for example the 2026 NEC moved load calculations from Article 220 to Article 120), results say so.

Where the data comes from

Standard values such as conductor resistance, ampacity, raceway areas and box volumes are used as facts, with the table or section they come from cited next to the result. We do not reproduce the text or layout of copyrighted codes. Physical data comes from manufacturer publications where possible, and every data row in the engine records its source and the date it was checked. See sources.

How we test

• Golden test cases. Each calculator is checked against worked examples from published sources: code annexes, manufacturer guides, textbooks. Only the numbers and the citation are used.
• Property tests. Behaviour that must always hold, such as “a longer run never gives less voltage drop” and “switching units never changes the answer”.
• Independent cross-checks. Sizing tools are compared against established public calculators over a grid of cases, and every difference is explained or fixed.
• Release gate. If any test fails, the site is not deployed.

The test cases are published at test cases so you can check them yourself.

What the calculators don’t do

They don’t replace the code, an engineer’s judgement or your inspector. Local amendments, equipment listings and installation instructions can all override a general rule. Each calculator states its assumptions under the result.