Methodology

The premise of this site is that a calculator you cannot check is not much use. A figure with no visible provenance is a figure you have to take on trust, and trust is the wrong basis for sizing a conductor. So every tool here prints the table or the formula it worked from, directly underneath the answer.

This page is the part of that promise you cannot see from a calculator page: which document each table came out of, how the arithmetic is verified, and — the part most sites leave out — which figures on this site are not transcribed from any standard at all.

Two kinds of number

Not everything here has a code citation behind it, and pretending otherwise would be the first thing to go wrong. The data this site runs on splits cleanly in two.

Transcribed code tables. Ampacity, conduit and box fill, grounding conductor sizes, motor full-load currents, standard overcurrent ratings. These are copied table by table out of a published standard, and each entry in the source carries a comment naming the table it came from — so a reviewer reading the code can find the row in the document without guessing. These are the figures the site is genuinely authoritative about, in the narrow sense that it is repeating a published number faithfully.

Trade and market conventions. Commonly sold UPS ratings, standby generator sizes, typical appliance wattages, conduit bending multipliers, nominal power factors by load type. No standards body publishes these. Some are geometry — a 30° offset multiplier is exactly 2 because of trigonometry, not because a committee agreed. Others are simply what products converge on in the catalogue, and they vary by manufacturer and by market. Where a calculator uses one, the page says so in as many words rather than dressing it up as a code figure.

The distinction matters at the moment you are challenged on a number. For the first kind, the answer is a table reference. For the second, the honest answer is "this is the size the market sells, go and confirm against the actual product" — and the pages that use these figures say exactly that.

Which edition

Standards are revised, and a calculator that does not say which revision it followed is not a reference. Here is the current position, per jurisdiction, in full.

Adoption is separate from publication, and this is the failure mode that catches people out. A jurisdiction adopts an edition on its own timetable and frequently amends it on the way through. Being current with NFPA 70 is not the same as being current with what your Authority Having Jurisdiction enforces, and this site knows nothing about the second. Check what has actually been adopted where you are working before you rely on any figure here.

How a calculator is built

Each of the 26 tools follows the same shape, and the shape is what makes the answers checkable.

• The arithmetic is a separate module. It has no knowledge of the page, the form or the browser. It takes numbers and returns numbers, which means it can be tested directly rather than through a user interface.
• Every module has a test suite that runs the published examples and the boundary cases through it — the row where a table changes band, the value that sits exactly on a threshold, the input that should be refused. A calculator whose tests do not pass does not build.
• The worked example is computed, not typed. This is the detail worth knowing about. The step-by-step example on each page is generated at build time by calling the same module the calculator in your browser calls. Nobody keyed those figures in, so they cannot drift out of agreement with the tool, and correcting a table corrects the worked example on the same build.
• The reference table shown is the data being used.The table printed under a calculator is rendered from the same source the calculation reads. It is not a decorative copy that could quietly disagree.

Rounding

Rounding is where a correct formula still produces a wrong answer, because the direction is not a matter of taste — it is set by whether the code figure is a floor or a ceiling, and the two look identical on a calculator screen.

Round up to a minimum. A conductor ampacity requirement, a required service size, a capacitor bank: the code states the least that will do, so the answer is the next size at or above it. Landing exactly on a standard size is allowed and leaves no headroom, which is worth noticing rather than celebrating.

Round down from a ceiling. A motor branch-circuit device under Table 430.52, or an air-conditioner maximum overcurrent protection figure, is the most you may fit. Rounding those up the way you would round a general load produces a device that exceeds the permitted maximum, and on HVAC equipment it voids the listing. The calculators that deal in ceilings round down, and say on the page that they are doing it.

Displayed figures are rounded for reading, never for the next step: intermediate values stay at full precision through the calculation and are rounded once, at the end. Where a result lands within a whisker of a threshold, that is generally reported rather than hidden, because a conductor that passes at 3.07% against a 3% target is a design conversation, not a pass.

What is checked before a page ships

The build refuses to complete unless a set of gates pass, so the standards below are not aspirations someone remembers to honour.

• Full type-check across every page and module, with warnings treated as failures.
• Every calculator test suite, run against the built pages as well as the modules.
• A content floor: at least a thousand words of genuine explanation inside each calculator page's written region, with the reference tables excluded from the count so a large table cannot flatter a thin page. The same check flags any sentence that appears on more than one calculator page.
• A site audit over the built HTML rather than the source: every internal link resolves, every page has a unique title, description and canonical, every form control has an accessible name, and the structured data parses.
• Colour contrast measured from the palette tokens, not eyeballed. See the accessibility statement for the numbers.

What this site does not know

A methodology page that only lists strengths is marketing. These are the real limits, and they are the reason every page carries a disclaimer.

• Your local amendments. The site models a published standard, not the version your AHJ enforces, and it has no way to know which that is.
• Your actual installation. Conditions that change an answer — ambient temperature, bundling, thermal insulation, sunlight on a rooftop raceway, the state of an existing service — are inputs you supply. Where you supply nothing, the tool assumes reference conditions and says which ones.
• Equipment listings. A manufacturer's instructions and listing conditions override a general calculation, and they are not in any table here.
• The metric tables, as set out above.
• Whether the answer is a good idea. The tools compute what the code permits. Whether it is the right design — whether that service should be 100 A because it complies or 150 A because of what the owner will add in five years — is judgement, and several pages say so at the point it matters.

Corrections

The rule is simple: if a calculator disagrees with the table printed beneath it, the table wins and the code is wrong. That is a bug, it is treated as one, and it is worth reporting even if you are not certain — a report that turns out to be a misreading costs a few minutes, and one that turns out to be real costs considerably less than the alternative.

The same goes for a transcription error against the published document. If a row here does not match your copy of the standard, your copy is the authority. Reporting routes are on the policy pages, and everything else about how the site behaves in the field is in how this site works.