Performance comparison of near-optimum digital pulse processors in high resolution X-ray spectroscopy

Abstract

Digital filtering has a high potential in nuclear spectroscopy due to its adaptivity, easiness of calibration etc., but also because it can synthesise filters close to the theoretical optimum. This works is a cross-comparison of a traditional analog processing system and two near-optimum digital processing methods for what concerns energy resolution. The samples are obtained by digitising the signal pulse as it appears at the output of a proper analog conditioning circuit. Both digital techniques take advantage of the possibility of synthesising the best possible weighting function with respect to the actual noise conditions. A noticeable improvement of more than 10% in energy resolution has been achieved with both systems with respect to state-of-the-art systems based on analog circuitry. In particular, the two digital processors are shown to be the best choice respectively for on-line use with critical ballistic deficit conditions and for very high resolution spectroscopy systems ultimately limited by 1/f noise.