Ultrafast stimulus error removal algorithm for ADC linearity test

Abstract

Linearity test of an analog-to-digital converter (ADC) can be very challenging because it requires a signal generator substantially more linear than the ADC under test. For high performance ADCs, the overall manufacturing cost could be dominated by the long test time and the high-precision test instruments. This paper introduces the ultrafast stimulus error removal and segmented model identification of linearity errors (USER-SMILE) method for high resolution ADC linearity test, allowing the stimulus signal's linearity requirement to be significantly relaxed and the test time to be reduced by orders of magnitude compared to the state-of-art histogram method. The USER-SMILE algorithm uses two nonlinear but functionally related input signals as ADC excitations and uses a stimulus error removal technique to recover test accuracy. The USER-SMILE algorithm also uses the ultrafast segmented model identification of linearity errors (uSMILE) approach to dramatically reduce test time while achieving test accuracy and coverage superior to the histogram method. The USER-SMILE algorithm is validated by extensive simulation with different types of ADCs, different resolution levels, and different types of input signals including nonlinear ramps, nonlinear sine waves and even random input signals. Statistical simulation results show that for a 16-bit SAR ADC, with two 1 hit/code nonlinear ramp signals, the INL test error is within +/- 0.4LSB.