On the Statistical Properties of the Peak Detection for Time-Domain EMI Measurements

This paper presents a discussion on the inherent characteristics of the measurements performed with time-domain electromagnetic interference measurement systems in regards of the detection of the maximum emissions levels. In that sense, some relevant statistical properties of the frequency components of the maximum emissions levels in the amplitude spectrum are investigated using the extreme value theory to provide a model based on the Gumbel probability distribution and estimates for its parameters, expected value, variance, and Cramer-Rao bounds. The results suggest that using the expected maximum value of the emissions levels instead of the just the observed maximum value improves the measurement repeatability and also reduces the uncertainty in the results. This paper presents an additional insight measure that enhances our understanding of the statistical behavior of the measured EMI and of the time-domain measurement process itself.