The maximum permissible error, or MPE, is the limit that defines how far an instrument can deviate from the reference and still be considered acceptable. It is set by the manufacturer in its specification, by a regulation, by a product standard or by the user according to their process. It does not describe the error of a specific instrument, but the limit required of it.
How it is used
In a calibration, the indication error is obtained at several points: the instrument reading minus the reference value. That error is compared with the limit at each point. If it falls within the limit, the instrument conforms at that point; if it falls outside, it does not. The limit can be expressed as an absolute value, as a percentage of reading, as a percentage of full scale or as a combination of several terms, and you should read carefully which one applies.
Reading and full scale are not the same
A limit expressed as a percentage of full scale is constant over the entire range, so at the low end it represents a large relative error. One expressed as a percentage of reading shrinks when measuring small values. This is why an instrument can comply at the top of the range and be of little use at the bottom for a given tolerance.
The role of uncertainty
The observed error is never exact, because the calibration itself has uncertainty. Near the limit, a result that appears to be inside may actually be outside, and vice versa. To decide consistently, a decision rule is defined that takes uncertainty into account, for example requiring the error plus its uncertainty to remain within the limit.
How to choose the right limit
- Start from the process or product tolerance you need to control.
- Make sure the instrument, with its uncertainty, has sufficient margin relative to it.
- Compare with the manufacturer’s specification, but do not automatically take it as the acceptance criterion.
- Document the decision rule and always apply it the same way.