The reliability of small digital controllers

Abstract

Increasing use is being made of small digital controllers in Industry and Commerce. The failure of such controllers is important since it may cause either plant to become unsafe or the interruption of production. Fault tolerant techniques are discussed for improving the reliability of digital controllers with special reference to the development of a hybrid electromechanical gas governor, whose electronic controller is an example of a small digital controller. Three microprocessors are used in a two out of three majority voting configuration and the memory is Hamming code protected. Redundancy techniques are used to protect against faults in other parts of the controller and it will tolerate most classes of transient fault. When comparing designs or attempting to meet reliability criteria, it is necessary to predict the reliability of a system and its individual components. Several sources of failure rate prediction are compared and the wide variation in the failure rates of integrated circuits is highlighted. The comparison concludes by recommending which reliability data source is likely to be most accurate for each type of component. The gas governor is an example of a repairable system and analysis is developed for predicting the improvement in reliability for repairable redundant systems and for determining the optimum maintenance and repair times for equipment. The testing of redundant systems is difficult because of their complexity, and under certain circumstances the redundancy can mask design faults. Testing methods using complex test equipment are described, as well as the testing of the experimental controller. A review is included of other fault-tolerant systems. Although the work on large computers is not directly applicable to small controllers, many of the techniques can be used.