Why Accuracy at Full Scale Can Mislead an Instrument Comparison
Direct Answer & Executive Summary: A quote can look technically stronger than it is when its stated accuracy is read outside the range and conditions that produced it.
Last updated: 2026-09-17
The short answer
A stated accuracy value is only useful when its reference basis, operating range, test condition, temperature band, and time since calibration are known. A small percentage of full scale can become a large proportion of the actual reading when the process normally operates at the lower end of the instrument range.
Start with the duty point, not the brochure headline
When comparing two instruments, write down the normal reading, the lowest usable reading, the highest credible reading, and the decision tolerance. These four values make an accuracy statement meaningful. A device can be excellent at one part of its span and a poor economic fit at another.
Ask the supplier to separate the terms
Do not accept a single accuracy line without context. Ask whether the number is stated as a percentage of reading, a percentage of full scale, a count of digits, or a combination. Then ask for the temperature condition, averaging method, bandwidth or response setting, and calibration interval used for the claim.
Compare error at the actual operating point
Turn each claim into an absolute error at the readings that matter to your process. This avoids selecting an unnecessarily expensive model just because its headline percentage is smaller, or accepting a low-cost model whose usable error grows at the low end of your requirement.
Put the acceptance rule in the RFQ
The requirement should name the operating band and acceptance method, not merely repeat a catalog specification. For example, define the readings to be verified, the ambient condition, the traceability expectation, and whether the error budget includes the sensor, lead, fixture, or signal conditioner.
Practical takeaway
Accuracy is a system decision. The most defensible selection is the one that meets the measurement decision at the real duty point with a stated verification method. It is not automatically the instrument with the smallest number on the front page of a datasheet.
