Systematic errors and measurement uncertainty: An experimental approach

The paper discusses the problem of quantifying the contribution of systematic errors on the overall measurement uncertainty. After a brief review of the recommendations of the Guide to the Expression of Uncertainty in Measurement (GUM), and of some notable alternative methods already proposed in the literature, the authors outline a fully experimental approach, which is based on the ISO 5725 ideas, circumventing the adoption of the concepts of “degree of belief” and “subjective probability” which, instead, are central both in the GUM and in the proposed alternatives. As a matter of fact, the proposed experimental approach does not alter the mathematical framework of the GUM (differently from the alternative proposals), but imposes that the uncertainty is always an experimentally verifiable, or refutable, figure. Theoretical considerations are followed by the description of an experimental scheme to study the uncertainty due to systematic errors in generic instruments and, in particular, in ADC-based instruments. A major focus is given to experimental results obtained for the systematic gain error, which, in turn, plays a significant role in the overall error contribution. The ultimate goal of the paper is to provide a basis to promote reflections upon a central problem: how much subjectivity and objectivity should be allowed in an uncertainty evaluation.

► We promote the experimental evaluation of uncertainty due to systematic errors.
► The method is alternative to non-experimental approaches, including the GUM one.
► The uncertainty due to systematic errors in ADC-based instruments is addressed.
► We present a general scheme, and use it for the uncertainty due to ADC gain error.
► The GUM type B uncertainty seems quite poor, compared to the experimental uncertainty.