In direct-sampling RF designs, data converters are typically characterized by the NSD, IM3, and ACLR parameters rather than by traditional metrics like SNR and ENOB. In software-defined radio and similar narrow-band use cases, it is more important to quantify the amount of data-converter noise falling into the band(s) of interest; legacy data conversion metrics are ill-suited to do this.
This white paper first presents the mathematical relationships underlying traditional ADC parameters—SFDR, SNR, SNDR (SINAD), and ENOB—and illustrates why these metrics provide good characterization of data converters in wide-band applications such as superheterodyne receivers. It then delineates why these metrics are inappropriate for data converters that do not function over their full Nyquist bandwidth, as in direct RF sampling applications like SDR. Derivation and measurement of NSD, IM3, and ACLR are described in detail, including use of the Xilinx RF Data Converter Evaluation Tool in the measurement of RF data conversion parameters.