Post-correction of analog-to-digital converters

This thesis deals with digital post-correction ofanalog-to-digital converters (ADCs). The performance ofpractical ADCs is deteriorated by nonidealities and flaws inthe converter. Methods for mitigating these errors by applyingdigital signal processing to the output of the converter havebeen proposed in the past. These methods are often referred toas postcorrection methods. This work is mainly concerned withpost-correction methods based on look-up tables. Practical ADCs often exhibit dynamical error behavior,meaning that the error is dependent on the dynamics of theinput signal. In this thesis an extension of previouslyproposed post-correction methods is proposed. The method usesthe present sample in conjunction with a number of past samplesto form the table index. In order to reduce the number of indexbits, and thereby the size of the table, the method comprises a\u91bit mask\u92, which selects a subset of the availablebits to be used in the index. Evaluations using experimentalADC data show that the proposed method improves the converterperformance, but also that the choice of index bits has asignificant impact on the outcome of the correction. Theincorporation of a bit mask enables an analysis of the effectof different bit masks. The analysis results in a framework forcomparing different correction tables. The framework is then applied in an optimization problem.The goal is to find the best allocation of a fixed number ofindex bits. Two different criterions are applied: minimizationof the total harmonic distortion and maximization of thesignal-to-noise and distortion ratio. The results of theoptimization, performed with experimental data, show that theoptimal bit allocation is different depending on whichcriterion is used. Moreover, the performance of a correctionscheme deteriorates only slowly with decreasing table size, ifappropriate index bits are selected.NR 2014080