Balance calibration: uncertainty, minimum weight and checks

Useful calibration connects balance performance to the masses and conditions actually used. The certificate describes metrological results; deciding whether a weighing task is suitable also requires the method’s requirements and uncertainty in use. Minimum weight cannot be obtained by counting display decimals or adding the vessel’s mass to a small quantity of sample.

1. Describe the weighing task, configuration and requirement

Specify the smallest net mass, usual masses, maximum gross load, vessels and use of tare. Record the balance, pan, active range, resolution, accessories, location and environmental conditions. Weighing by difference requires both readings and the model connecting them: uncertainty at a single point is not always enough.

Define the quantity to report, for example conventional mass under the adopted model, and the acceptable contribution of weighing to the analytical method. Distinguish the limit on indication error from the requirement for measurement uncertainty. The calibration plan should represent discrete loads, successive additions or taring, as appropriate. EURAMET cg-18 addresses this relationship. [1]

Instrument selection is covered in balance capacity, readability and environment. Here the purpose is to interpret evidence for the installed configuration, rather than choose a commercial model.

2. Assess reference weights beyond their nominal class

To determine error, use reference values with uncertainty, traceability and status appropriate to the test. Check identifiers, calibration results, allowable drift, storage, integrity and possible contamination. The class designation alone does not provide all this information. Nor is the nominal value necessarily the value to use without correction.

Suitability depends on the contribution to the particular test’s uncertainty budget: do not impose a universal fixed ratio between reference and balance. A stable weight can be used to investigate repeatability without that test also determining indication error. Clarify the purpose before demanding identical properties of every load.

Temperature, convection, air buoyancy and handling can affect the comparison. NIST describes these influences in its Good Measurement Practices: these are metrological recommendations, not pharmaceutical GMP. [2] Do not automatically transfer timing or environmental limits from mass calibration to your QC laboratory.

3. Distinguish tests and uncertainties

The indication test compares indication I with reference-load value m_ref. Here e = I − m_ref; the correction is −e. Repeatability assesses dispersion under defined conditions. Eccentricity assesses the effect of load position. None of these tests automatically replaces the others. Points, sequences and repetitions follow the applicable method and agreed scope.

The uncertainty of calibration error concerns the comparison performed. Weighing uncertainty in use also considers configuration, taring procedure, environmental changes, drift and relevant object properties. Cofrac LAB GTA 95 distinguishes these levels within its accreditation framework. [3] Uncertainty established with compact weights at the centre of the pan does not automatically cover a large vessel at another workstation.

For an individual reading, do not substitute the standard deviation of the mean for the standard deviation simply to reduce the budget. Avoid double counting contributions from the certificate, repeatability, resolution and operating conditions. For a difference, also assess correlations between readings: do not indiscriminately add two expanded uncertainties. Document how corrections are applied and residual effects treated.

4. Minimum weight: a requirement linked to use

Minimum mass depends on the criterion adopted and demonstrated performance. Appendix G of cg-18 relates relative uncertainty to the user’s requirement and net mass; tare does not increase the useful denominator. [1] Variability and risk may justify an operating margin, but that cannot replace a missing uncertainty budget.

Check applicable pharmacopoeial chapters separately. The 2025 USP <41> preview distinguishes calibration and periodic checks, with risk-based frequencies; its scope excludes manufacturing balances. [4] EDQM describes chapter 2.1.7 in the European Pharmacopoeia context. [5] This public information does not establish a universal compendial formula or tolerance for every balance: consult the current full text for your application.

Simulated case. For a balance with a 5 g vessel, assume uncertainty in use U = 0.12 mg, k = 2, valid for net masses from 20 to 200 mg under the stated conditions. Corrections are applied and residual effects included in the hypothetical budget. These are neither real data nor a pharmacopoeial test.

The predetermined rule is U/m ≤ relative requirement. For 20 mg with a 0.5% requirement, U/m = 0.6%: the criterion is not met; uncertainty of at most 0.10 mg would be allowed. For 200 mg with a 0.1% requirement, U/m = 0.06%: the criterion is met, with a 0.20 mg limit. A relatively stricter requirement can therefore be compatible with the larger mass.

Only under the stated constant-U assumption does U/requirement give thresholds of 24 mg and 120 mg. Do not extrapolate beyond the justified range or automatically call these USP minimum weights. Increasing tare does not solve the first weighing task; assess another configuration or an approved method change.

5. Original matrix: from need to evidence

A row resolves a question only if its evidence concerns the required configuration. If data are missing, record the gap and affected use: do not replace the test with a “compliant” checkbox.

6. Routine checks, interventions and return to service

Select checks able to detect relevant changes between calibrations: points, references, limits, frequency and actions depend on use, history and risk. Define beforehand how to distinguish an alert, a limit breach and an invalid test. A check at one load does not certify the entire range; a stable sequence does not by itself authorise extending the interval.

When an anomaly occurs, retain the data and assess instrument status and affected weighings. Do not repeat until a favourable result appears. Adjustment, including automatic internal adjustment, changes the response and is not equivalent to a documented comparison with uncertainty. Agree before/after data and record relevant activations, maintenance or repairs.

Review the report for identifiers, location, conditions, loads, results, uncertainties and limits; then have authorised functions assign the operating range and checks. Relocation and environmental changes require an assessment of evidence validity. Connect the decision to weighing technique and sample transfer and the uncertainty budget. Further pathways are in the Calibration & Laboratory Metrology hub.

7. Sources and limitations

Checked: 1 October 2026. The matrix and case are original GuideGxP material; simulated values are not regulatory tolerances. The public documents below were consulted, not the complete pharmacopoeial chapters.

1. EURAMET cg-18, version 4.0, November 2015 — sections 4, 5, 7 and Appendix G.
2. NIST IR 6969, GMP 10, 2019 — factors affecting weighing.
3. Cofrac LAB GTA 95, revision 02 — applicable from 30 January 2023; non-automatic weighing instruments.
4. USP <41> Balances, 2025 preview — scope and life cycle.
5. EDQM, Weighing according to the European Pharmacopoeia — announcement of 26 February 2022 on 2.1.7.