ANOVA-Based Approach for DAC Diagnostics

In order to assess the performance of digital-to-analog converters (DAC) the attention is generally focused on the integral nonlinearity INL. Useful diagnostic tools to detect the causes of poor linearity are the linearity and intermodulation errors, which can be evaluated from INL measurements. Linearity and intermodulation errors highlight and quantify erroneous calibration and unwanted interactions between current sources inside the DAC hardware. Unfortunately, their estimates, particularly those related to high-order intermodulation errors are characterized by high uncertainty. On the base of their very uncertain value, it is difficult to establish if interactions represent relevant factors or not. It is shown that by means of the analysis of variance (ANOVA) the relevance of intermodulation errors can be assessed from a limited set counting a minimum of two INL measurements. ANOVA is in fact capable of distinguishing if the variance in INL measurements has to be addressed to active factors or noise, even if the effect of each factor is widespread upon different elements of the INL array and in different combinations with the other factors.