Improvement of multiexponential signal decomposition by the preprocessing

Abstract

Dielectric absorption of capacitors is a source of possible errors in the analog preprocessing blocks and integrating analog to digital converters. Parasitic relaxation components in the capacitor's model could be estimated by the decomposition of the charging multiexponential current. Paper deals with the possible preprocessing methods, selection of the optimal sampling conditions with the aim to enhance the robustness of the signal decomposition algorithms on the superimposed thermal and quantization noise.