Resistor Tolerance Calculator

How to use this calculator

Start in Range from nominal mode: enter the part's nominal resistance with its unit (Ω, kΩ or MΩ) and the tolerance percentage — 1 for a ±1% metal-film part, 5 for the classic gold-band carbon film. The result is the guaranteed window: nominal × (1 − tol/100) to nominal × (1 + tol/100). With the defaults (10 kΩ, ±1%) it reads 9.9 kΩ – 10.1 kΩ, and the gauge below the inputs draws that window with the nominal value at its centre. The percentage field offers the standard grades as suggestions, but any positive number works — some military and thin-film parts carry tolerances like 0.02%.

Switch to Check a measured part when you have a meter in hand: keep the nominal value and tolerance of the part you are testing, then enter what the meter reads (in its own unit — the field converts, so 10.05 kΩ is fine). The result panel adds three rows to the window: the measured value, its deviation from nominal as a percentage (e.g. +2.34%), and a PASS/FAIL verdict against the window, with the margin to the nearest limit when it passes or the overshoot when it fails. The gauge shows the measured point as a marker — green inside the shaded window, red outside it — so a reel of parts can be triaged at a glance.

Where do the two numbers come from in practice? Usually from one of the band calculators on this site: decode the colors to get the nominal value and the tolerance band's percentage, then bring both here to see the ohm window your circuit must survive. Keep the units straight when you type — a 10 kΩ part is "10" with kΩ selected, not "10000" with Ω (both work, but mixing them up is how 10 MΩ surprises happen) — and remember the window is a guarantee on the batch, not a prediction of where any one part sits: individual units distribute roughly bell-shaped inside it, so most will be closer to nominal than the limits suggest.

What it means in a real circuit

Tolerance is the fingerprint of how resistors are made. The resistive track is deposited or wound longer than needed and then trimmed — by laser, wire cutting or evaporation control — until the value lands somewhere inside a target window; no trimming process hits its target exactly, so the manufacturer sorts production into batches and stamps each batch with the worst-case deviation it contains. A ±5% mark means "every part in this reel is between 95% and 105% of nominal," full stop. The guaranteed window is therefore a manufacturing specification: it says what must be true when you buy, before temperature, age or stress enter the story.

How that window affects a circuit depends on which kind of specification the resistor serves. If it sets an absolute quantity — a current limit, a voltage drop, a timing constant with a capacitor — design for whichever extreme stresses your spec: maximum current flows at the lowest resistance in the window, minimum drop appears at the highest. If it forms a ratio with another resistor — a divider, an op-amp gain, a feedback network — the individual tolerances combine, and in the worst case roughly add in percentage: two ±5% legs can push a nominally 1/2 divider to about 0.475 or 0.525. If two parts must match each other (ladders, bridges, differential pairs), even tight individual tolerances are not enough — independent ±1% parts can differ from each other by ~2%, which is what matched arrays and sorting exist to fix.

The economics follow the same shape. Tolerance costs money because it is yield: a 0.1% batch discards or downgrades everything that misses, so tighter grades are pricier and thinner in stock, and ±1% metal film won the market as the point where cost, availability and performance meet for most designs. And tolerance is only one axis of accuracy — temperature coefficient tells you how much the value moves per degree after purchase, so a "±1%" part with a 100 ppm/°C TCR can drift several times its tolerance across a hot-to-cold cycle. Buy tolerance for the manufacturing spread your spec needs, and TCR for the thermal range it must survive; this calculator handles the first of those two numbers.

Definitions

Tolerance is the maximum permitted deviation of a part's resistance from its nominal value, expressed as a percentage: it defines the guaranteed window [R·(1−tol/100), R·(1+tol/100)]. The nominal value is the marked or decoded target (digits × multiplier from the bands, or the code on an SMD part). Deviation of a measured part is (R_measured − R_nominal)/R_nominal as a percentage — positive when high, negative when low. The temperature coefficient (TCR), in ppm/°C, is a separate specification describing how the value drifts with temperature (ΔR/R ≈ TCR·ΔT/10⁶), and it is not captured by tolerance at all.

Design tips

Budget tolerances against the spec type, not the loudest number in the schematic. For absolute specs, compute the circuit at both window extremes and keep whichever fails first: a current limiter sized for nominal R will exceed its target current when the part sits at the low end (I = V/R grows as R shrinks), so check I_max at R_min first. For ratio specs, use the sum of the individual percentages as your worst case — two ±1% parts give ~±2% on the ratio — and either buy tighter parts or add a reference to break the dependency. Where matching matters (R-2R ladders, bridge circuits), specify matched pairs/arrays with a stated mismatch tolerance rather than relying on two independent tight parts to land close together.

Keep a margin so that even the worst case passes: if your spec allows 20 mA maximum and nominal design gives 19 mA, verify it still holds at R_min (and with supply sag/spike, which this calculator does not model). And when debugging a circuit that "works on one board but not the next," tolerance spread is a first suspect — measure the actual parts in the failing board before changing anything; a divider leg sitting at the edge of its 5% band can shift a threshold by more than any other component in the chain.

Common mistakes

The most common is designing to nominal and discovering the spec only holds for half the batch: a "1 kΩ" divider leg can be anywhere from 950 Ω to 1050 Ω, and a circuit tuned at exactly 1 kΩ has no margin on either side. Its twin is confusing tolerance with TCR — both are percentages of the value, but one is a one-time manufacturing spread (this calculator's subject) and the other is continuous thermal drift; a part can be well inside its tolerance window at 25 °C and outside it at 85 °C if its TCR is large. A subtler trap in check mode is measuring live: self-heating from the meter's own test current, or the circuit's operating temperature, shifts the reading, so verify suspicious FAIL verdicts on a cold, de-energized part before condemning it.

Also avoid assuming two parts with the same nominal value match each other — they can differ by up to twice the tolerance — and do not read a tight measured value as proof of a tight part: a 5% part that happens to measure within 0.2% of nominal is still a 5% part, thermally and over its lifetime. Finally, watch unit slips when switching between the Ω/kΩ/MΩ fields; they are converted correctly by the calculator, but a "10" typed under MΩ instead of kΩ moves the window by a factor of 1000.

Component-selection advice

Pick the tolerance from the tightest spec the resistor serves, not from habit: pull-ups and LED limiters with margin run happily on ±5%; dividers, gain sets and references want ±1% or better; metrology paths want 0.1–0.5% thin film. Within a grade, construction matters as much as the number — metal film beats carbon film on TCR (typically ±20–100 ppm/°C versus several hundred) and noise, which is why "±1%" is not one product but a family. If you need a value between stocked grades, the cheap route is sorting: buy 1% parts in quantity, measure them with a good meter, and bin the ones inside your tighter window — you are trading yield for precision, exactly as the manufacturers do, just at bench scale.

Limitations of this calculation

The window arithmetic is exact — but it describes only the manufacturing guarantee at the moment of purchase. It says nothing about where a specific unit sits inside the window (the distribution is roughly Gaussian, so most parts cluster near nominal), how the value will drift with temperature (TCR, a separate specification), or how it ages through thermal cycling and high-voltage stress. At very low resistances the absolute window in milliohms can be smaller than the lead and contact resistance of your measurement setup, which is why sub-ohm parts specify four-terminal (Kelvin) measurement; at very high resistances, leakage across the meter's input dominates instead. And tolerance grades are symmetric by standard — this calculator applies that symmetry — so it cannot represent one-sided or custom specifications some military parts carry.

Example use cases

Worst-case check on an LED limiter. A red LED (2.0 V forward) from a 5 V rail at a target of 20 mA wants R = (5−2)/0.02 = 150 Ω; the nearest E24 value is exactly 150 Ω, and a ±5% part gives a window of 142.5 – 157.5 Ω. At the low end the current rises to (5−2)/142.5 ≈ 21.1 mA, at the high end it falls to ≈ 19.0 mA — both inside a "20 mA ±10%" spec, so the selection holds without derating the LED. Had the target been 25 mA maximum, the same arithmetic would have forced a larger nominal value (or a tighter part) before anything was soldered.

Triaging a reel in check mode. A bin of "470 Ω ±5%" parts (window 446.5 – 493.5 Ω) is being reused in a precision divider that actually needs ±2%. Measuring one part reads 481 Ω: deviation +2.34% from nominal — inside the 5% window, but outside the 2% requirement (±9.4 Ω around 470), so it fails your spec even though it passes its own. The verdict row and the margin number make that distinction explicit part by part, which is faster and less error-prone than doing it in your head across a hundred measurements.