Skip to Main Content
A mixed-signal micro-controller architecture is used to efficiently implement a novel non-binary capacitor array calibration algorithm for a successive approximation charge-redistribution data converter. The non-binary capacitor array calibration algorithm is based on a simple perceptron learning rule, originally developed for Artificial Intelligence applications. The digital capacitor weights are calibrated with up to 22-bit accuracy that is better than the noise level in the system. The resolution of the data converter is no longer limited by the capacitor array matching. The mixed-signal micro-controller is portable, scalable, and seamlessly handles the converter's analog and digital datapaths.