A Fault-Tolerant BIST Design of MEMS Infrared Thermopile Sensor

In this paper, a novel framework of fault-tolerant BIST (Built-In Self-Test) for MEMS (Micro-Electro-Mechanical-System) infrared thermopile sensor is proposed. In the proposed method, the BIST and fault-tolerant are both realized by BP (Back Propagation) neural network. During the BIST process, the test stimuli of the thermopile are provided by a built-in thermal resistance, and the test responses of the thermopile are analyzed to judge whether the device fails or not. The fault tolerance is realized by dividing the entire thermopile into four similar sections, and their responses are further analyzed to decide whether the faulty sensor can still be used with loss of accuracy or not. The simulation results show that the test accuracy of the proposed scheme is beyond 94%, with the fault tolerant scheme, the yield can be improved from 91.95% to 93.98%.