Characterisation of frequency instability and frequency offset using instruments with incomplete data sheets

Abstract

This paper describes how to deal with electronic instruments, commonly used in frequency calibrations, whose data sheets do not include complete information regarding the sources of error (uncertainties), which affect the accuracy and stability of the frequency under test. Considering a time interval counter as the measurement unit, and a GPS receiver as the traceable standard, the purpose is twofold. First, contributions to the Type B uncertainty are calculated under the assumption of uniformly distributed errors. An expression is used for frequencies under test whose values are close to 1 Hz. Secondly, short-term instability is studied using non-classical statistics, which have been previously tested using simulated data. Then, a white noise test is performed based on the calculation of the classical to non-classical variance ratio, and supported by the direct measurement of the slopes in the graphs which depicts the variances (the Allan deviation and the modified Allan deviation) vs. the measurement time.