Accurate testing of ADC's spectral performance using imprecise sinusoidal excitations

Abstract

Analog to digital converter (ADC) is the world's largest volume mixed-signal circuit. It is also a key building block in nearly all system on chip (SoC) solutions involving analog and mixed-signal functionalities. ADC testing is also crucial for built-in-self-test (BIST) solutions of AMS testing in SoC technology which is identified by the ITRS as one of four most daunting SoC challenges. ADC spectral testing is of critical importance to a large class of integrated circuits and is particularly challenging for high speed and/or high resolutions circuits. In this paper we use spectrally related excitations (SRE) to accurately test the spectral performance of ADCs. Unlike standard approaches, the SRE approach uses low-cost imprecise sine signals as input to the ADC and uses the spectral relationship between multiple input signals to separate distortion inherent in the ADC from that in the input. Efficient DSP algorithms are used to determine the true spectral performance of the ADC. This approach works in both production test and BIST environments. Simulation results show two sine waves with < 60 dB purity can be used to accurately test spectral performance of high resolution ADCs with SFDR in excess of 100 dB. The low-cost SRE signals can be readily generated with simple RC filters with lax band edge requirements. Extensive simulation shows that the algorithm is robust to filter errors, to nonstationary in the test environment, and to measurement noise.