Minimalist Fault Masking, Detection and Recovery Techniques for Mitigating Single Event Effects in Spaceborne Microcontrollers UCLA Computer Science Department Technical Report TR-980025
D.W. Caldwell, D. A. Rennels · 1998
This paper presents a design approach for implementing fault-tolerant embedded computing nodes in spacecraft using non-hardened, commodity microcontrollers. The a pproach u ses faulttolerance to address high rate transient errors and o ccasional l atch up s that are e xpected o f such devices in the space radiation environment. A key constraint is to minimize the amount of external logic needed to support fault-tolerance so that t he primary advantage of microcontrollers, h igh functional density, can b e maintained. Low-cost approaches which leverage features of existing commercial microcontrollers are discussed. A built-in, h igh-speed serial port i s used for voting among redundant devices and a novel wire-OR output voting scheme e xploits the bidirectional controls of I/O pins. A fault-tolerant node architecture is presented, and the effectiveness of the faulttolerance techniques is discussed. A testbed is being constructed to verify the described techniques, using triplicated microcontrollers to control a 3-axis s et of r ate gyros, thereby providing a realistic application example.