Probability Model
Probability model is a continuous distribution where all values within a specified interval are equally likely to occur, with zero probability outside that range. The rectangular distribution is the default model used when only the upper and lower limits of an error source are known, with no information about its central tendency. It is widely applied to quantization errors and manufacturer tolerance specifications.
Uncertainty Calculation
To convert the half-width of the tolerance interval to a standard uncertainty, the value is divided by the square root of three. This mathematical relationship is derived from the variance of a uniform probability density function. The resulting standard uncertainty can then be combined with other uncertainty components using the law of propagation of uncertainty.
Metrological Application
Digital resolution limits and analog-to-digital converter quantization errors are modeled using this uniform distribution. It is also used when a calibration certificate specifies a tolerance limit without a confidence level or coverage factor. This approach represents a conservative estimate, ensuring that the uncertainty is not underestimated when details about the error distribution are absent.
Statistical Limit
The assumption of equal probability across the entire interval represents a worst-case scenario that may overstate the actual uncertainty if the values cluster near the center. It ceases to apply when there is any evidence of a central tendency or if the error is known to be asymmetric. In those cases, a triangular or normal distribution must be used to provide a more realistic representation of the measurement process, ensuring that the overall uncertainty budget is neither excessively inflated nor unsafely minimized.