How-to · scales

Coffee scale calibration: how to check a scale with known weights

A brewing scale can show tenths of a gram and still be wrong by more than that. The only way to know is to weigh something whose mass you already know and compare. Below: what to use as a reference, a five-step check adapted from the tests US weights-and-measures inspectors run, and a calculator that sets your error against NIST Handbook 44 tolerances.

Updated 2026-11-02 · sources listed at the end · no product recommendations

Resolution is not accuracy

Handbook 44, the NIST handbook that sets US requirements for commercial weighing devices (the 2026 edition was adopted by the 109th National Conference on Weights and Measures), separates two numbers. The scale division, d, is the step the display counts in. The verification scale division, e, is a value the manufacturer specifies, and it is what sets a scale's accuracy class and tolerances. For most commercial scales, including Classes III and IIII, Handbook 44 requires e to equal d.

A digital reading also rounds. NIST's explainer gives the example of a display showing 123 kg: the true value lies between 122.50 and 123.49 kg, which is ±0.5 d before any other error.

Maker specifications show the gap. HARIO's manual for its V60 Drip Scale lists a 0.1 g display step from 2 to 200 g but an accuracy of ±0.2 g in that range, then 0.5 g steps with ±1 g from 200 to 500 g and 1 g steps with ±3 g from 500 to 2,000 g, with accuracy assured at 20 °C ± 5 °C.

How accurate does a brewing scale need to be?

The SCA's Golden Cup reference is 55 g/L ± 10%. On a 15 g dose, 10% is 1.5 g, so a scale that is off by a few tenths of a gram stays well inside that band. Cupping is stricter: the SCA protocol asks for 8.25 g per 150 ml, with the dose adjusted to within ±0.25 g. Small doses are where an error matters most, which is why the check below starts with a small known weight.

What to use as a known mass

The weight limits are maximum permissible errors from Table 1 of OIML R 111-1. Even an M1 100 g weight is within 5 mg, a twentieth of a 0.1 g display step. Coins are rougher references: the US Mint publishes the weight of coins in production but no tolerance, and a coin from your pocket has not been checked against anything. A stack of 20 nickels is still useful as a quick screen; just don't read a 0.1 g difference into it. The cent is also on its way out: the last penny for general circulation was struck on November 12, 2025.

How good does the reference need to be? Handbook 44 asks inspectors to use field standards whose error is no more than one-third of the smallest tolerance being applied, and OIML R 111-1 sets the same one-third limit for a weight's calibration uncertainty against its permissible error.

Check your scale in five steps

These steps borrow the structure of the Handbook 44 field tests (zero, increasing load, decreasing load and a shift test). At home they tell you whether to trust the scale, not whether it would pass inspection.

1. Set up. Use the scale at room temperature on a stable, horizontal surface free of vibration (Escali's instructions). NIST's precision-scale guidance adds: keep it away from vents, windows, doors and fans, and keep it level. HARIO warns against magnetic fields and notes that a hot item can shift a reading by ±2 g right after it is placed.
2. Zero. With the platform empty, the display should read 0. Handbook 44 says the zero balance should not change by more than the smallest applicable tolerance; re-check zero after each load comes off.
3. Load up. Place the reference in the center and note the reading. Add references in steps toward the scale's capacity if you have them: Handbook 44 calls for test weights equal to 100% of capacity on scales up to 150 kg, and says a scale should be tested to capacity at its first verification.
4. Load down. Take the weights off in the same steps and note each reading. Handbook 44 includes a decreasing-load test alongside the increasing one.
5. Corners. For scales of 500 kg or less, the Handbook 44 shift test places a load of about one-third of capacity at the center of each quadrant of the platform in turn. On a 2,000 g scale that is about 667 g. Each reading should be within tolerance of the known mass.

Handbook 44 tolerances as a benchmark

Handbook 44 tolerances are counted in divisions of e and grow in steps as the test load grows. NIST's explainer tabulates the maintenance tolerance (the limit for scales in service):

The acceptance tolerance, used for new or newly repaired scales, is half of that: NIST's 2024 precision-scale overview sets an acceptance tolerance of 0.5 e against a maintenance tolerance of 1.0 e. Which class applies depends on e and the number of divisions n (capacity ÷ e). Handbook 44's Table 3 puts Class III at e from 0.1 g to 2 g with n from 100 to 10,000, and Class II at e of 100 mg or more with n from 5,000 to 100,000. A 2,000 g scale counting in 0.1 g has 20,000 divisions, beyond Class III's limit, so the calculator benchmarks it as Class II.

None of this binds a kitchen scale. Wikipedia's weighing-scale article notes that scales meant for non-trade use must be labelled "Not Legal for Trade," and a brewing scale is not tested to these limits. The tables simply give a sourced yardstick for "close enough."

If the scale is off

• Repeat the check after the scale has sat at room temperature; Wikipedia's weighing-scale article lists temperature change and circuit drift among causes of mis-calibration.
• Calibrate where you brew. Spring and strain-gauge scales respond to weight, gravity varies by over 0.5% across Earth's surface, and Wikipedia notes that a commerce-legal spring scale must be calibrated where it is used.
• Follow the maker's routine if there is one. Acaia's, for example, runs from weigh-only mode: press T rapidly 6–7 times until the display shows CAL, then place the weight in the middle. It accepts F1 or F2 weights of 100, 300, 500, 1,000 or 1,500 g, and Acaia says frequent calibration is needed with regular use.
• If the manual has no calibration procedure (neither the HARIO V60 Drip Scale nor the Escali Primo instructions include one), compare the result with the stated accuracy and contact the maker if it is outside.

For the basics of how load cells, tare and flow rate work, see coffee scales explained. A checked scale makes the numbers in the extraction calculator and the tablespoon-to-gram conversions worth trusting.