
Two instruments can have valid calibration records and still report different readings. Compare what they measured, how they were used and the evidence for each result before deciding that either instrument has failed.
First make sure the comparison is meaningful
Both instruments must address the same quantity under comparable conditions. Two temperature probes placed at different depths may encounter a real temperature gradient. Two calipers contacting different positions on a shaft may reveal form variation. A difference between readings is not automatically an instrument error.
- Define the feature, position or sampling point.
- Use consistent units and an appropriate common method.
- Control relevant temperature, alignment, contact force and timing.
- Record sensor, fixture, range and accessory configurations.
- Check whether either instrument has been dropped, overloaded or changed since calibration.
Review corrections before comparing indications
Read the certificate's sign convention. If error is indication minus reference, the corresponding correction has the opposite sign. Apply a correction only where the method and calibration evidence support its use.
| Item | Instrument A | Instrument B |
|---|---|---|
| Indication | 100.020 mm | 100.010 mm |
| Supported correction for this example | -0.020 mm | -0.010 mm |
| Corrected result | 100.000 mm | 100.000 mm |
The raw readings differ by 0.010 mm; the corrected results agree in this illustration. This does not prove that either result is exact. Uncertainty, stability and conditions of use still matter. Do not copy these corrections to another instrument or interpolate between calibration points without an appropriate basis.
Check repeatability without hiding the original result
Preserve the first readings. Repeat the comparison using a defined sequence and record individual values, rather than reporting only a favorable average. Repositioning the workpiece or exchanging operators may help distinguish setup effects from instrument effects, but document each change so that separate tests are not mixed together.
When one instrument varies widely and the other is stable, investigate seating, contact, resolution, operator technique and instrument condition. When both are stable but differ consistently, review corrections, configuration and whether they are measuring the same feature.
Use uncertainty for the comparison
Read the uncertainty and coverage information associated with each relevant calibration result. Then consider additional effects introduced in use. A calibration certificate's uncertainty is not automatically the complete uncertainty of a production measurement.
Do not declare agreement solely because displayed digits match, or failure solely because they differ. A formal compatibility assessment needs an appropriate uncertainty model, including any shared references or correlated effects. A product acceptance decision also needs the specified requirement and decision rule.
Choose and record the next action
| Finding | Next action |
|---|---|
| Different positions or conditions | Repeat with a controlled, comparable setup. |
| Correction or units applied incorrectly | Correct the calculation and preserve the original record. |
| Unstable or unexplained readings | Restrict use where necessary and investigate before relying on the result. |
| Evidence suggests a change since calibration | Follow the equipment-control process for checks, recalibration and impact review. |
The reviewer should record the comparison, evidence considered, limitations and release or restriction decision. A fresh calibration may help resolve an instrument question, but it cannot reconstruct missing information about the original measurement setup.
Related investigations
- General investigation of an unexpected result
- How to read certificate uncertainty
- Reviewing previous measurements after failed calibration