A survey on the dynamic characterization of A/D converters

This article presents the different ways used to characterize A/D converters through the “CanTest System” test bench and its functionalities. The different ways used to characterize A/D converters dynamically could be organized into three parts: time domain, statistical and spectral analysis. The most commonly used algorithms in time domain analysis are presented. Due to instrumentation error, some may give “inaccurate” results. We explain why and propose a new method based on a direct estimation of conversion noise to avoid this problem. The different methods and applications linked to statistical analysis are also described. We explain the three different methods used to measure the differential non-linearity (DNL) and the integral non-linearity (INL). A comparison between these three algorithms in simulation and through an experimental case shows that their behaviours differ when the histogram presents edging effects. The final study presented in this paper concerns the A/D converter performance evaluation by analysis in frequency domain. From the output power spectrum, relationships are given for the computation of all spectral parameters. We present a correction to include in the computation of spectral parameters when the analog input does not span the entire full-scale range of the A/D converter under test. When the input generator provides a sine wave with harmonics, we propose the “Dual-Tone” method in order to separate the distortion associated with the component and the one associated with the generator. In the last part of this paper, we describe the test bench giving its specifications (acquisition system, frequency range,…). Some typical measurements and comments about the interest of analysis in different modes are given also.