New approach to linearity testing of A/D converters

Abstract

A new method is developed for integral-non-linearity (INL) testing of analogue-to-digital converters (ADC). While providing better accuracy and repeatability, the new method uses only a fraction of the test time used by either the servo-loop method or the histogram method, which are the traditional test methods currently used in the industry. The proposed new method in INL testing is an application of digital signal processing (DSP), namely the fast Fourier transform (FFT) and inverse fast Fourier transform (IFFT) methods, for filtering out the INL information from the digital data generated by an ADC under test. The approach takes advantage of the fact that the INL is directly related to the harmonic distortions of an ADC, and creates a method to extract the INL information from the frequency components that are INL related. By doing this, the environmental noises, such as test system noise and some device noise, are not taken into account when the INL is calculated. This is different from the servo-loop method and the histogram method, in which the environmental noise is averaged out by taking a large number of samples per code. By avoiding the need of having to take a large amount of data to eliminate or reduce the effect of the environmental noise, the method is able to reduce the test time and improve the repeatability dramatically. A mathematical proof is also given in this paper to show the relationship of the INL in the time-domain to the harmonic distortions in the frequency-domain for a given ADC. Simulation results and comparisons with the traditional test methods are also provided in the paper.