On-field determination of the minimum-noise filter for digital radiation spectrometers

An experimental method which permits to quickly determine the minimum-noise filter to be used in high-resolution digital spectrometry is presented. First the spectral density of noise as referred to the spectrometer input-which basically dictates the shape of the optimal filter-is determined by means of established digital-signal-processing techniques. By a least-mean-square fitting procedure the individual noise components (white, Lorentzian and high order contributions in both the series and the parallel equivalent input noises) are then disentangled out of the overall noise density. These latter are used to feed a software procedure which computes the optimal-filter weight function, having the option to introduce an adequate finite width and a flatted top around the measurement time.