A procedure for highly reproducible measurements of ADC spectral parameters

Abstract

The evaluation of spectral parameters characterizing analog-to-digital converters (ADCs) is addressed by applying a single- or dual-tone generator to the device input and by properly processing its output data stream. The coherent sampling condition, highly recommended by the IEEE standards 1057 and 1241 which list the most effective ADC testing procedures, is usually difficult to achieve, and sometimes even unfeasible. In fact, it requires a fine synchronization between the input and the sampling signals frequencies and it cannot be achieved when spurious tones are present in the ADC output spectrum. Data windowing is usually employed to reduce the associated spectral leakage phenomenon. However, IEEE standards do not provide clear criteria for choosing the window to be used for testing a given b-bit converter. Therefore, a reduced measurement reproducibility can result. The European draft standard Dynad suggests the employment of one out of seven windows in accordance to the ADC resolution. However, each proposed sequence covers only a limited converter resolution range. In this paper, an ADC testing procedure is described, suitable to yield highly repeatable and reproducible measurements, also when noncoherent sampling applies. To this aim, the use of a class of windows is proposed, that uniquely applies to ADCs with arbitrarily high resolution. Finally, experimental results that validate its effectiveness are presented.