DNL and INL yield models for a current-steering D/A converter

Abstract

The key measures of the static linearity of a current-steering D/A converter are Differential Nonlinearity (DNL) and Integral Nonlinearity (INL). Several papers have been published in which various models of the INL yield as a function of the variance of the current source mismatch have been proposed. However, most of these methods neither describe the statistical behavior of the INL and DNL yield accurately nor do they take to account the effects of the segmentation on the INL and DNL yields. In this paper, the analysis of the statistical behavior of the INL yield is presented. It is demonstrated, that the segmentation of the current sources affects the statistical behavior of the INL and DNL. In addition, regression models for the DNL and INL yields are presented. These models are applicable in cases of 10 to 16 bits with 1 to 6 and 1 to 3 thermometer coded MSB bits for the INL and DNL, respectively.