Minimum-noise filter for baseline estimation in radiation detection systems

Baseline height estimation in nuclear pulse spectrometry can be optimized by using a proper weight-function. The minimum-noise function for baseline estimation in presence of series and parallel white noises, constrained to a finite duration, is theoretically derived in this paper. Baseline restoration can be subsequently performed by subtracting the minimum-noise baseline estimation to the output signal pulse. The overall signal-to-noise performance obtained using the minimum-noise baseline estimation for pulse height correction is compared to that obtained with routinely used Base Line Restorers. An improvement of up to 16% in the Equivalent Noise Charge, with respect to classic baseline restoration, can be theoretically achieved in case of a simple triangular pulse shaping. The minimum-noise baseline weight-function can be synthesized in practice by means of a digital processing unit.