The reliability of electronic equipment has received increasing attention during the past twenty years, as apparatus has grown more complicated and has been applied to a great variety of important tasks. Despite its current interest to the practising engineer, the study, design and evaluation of reliable systems as a professional activity has gained only limited recognition. One reason for this is undoubtedly the wide range of topics which concern the reliability engineer. These include mathematical subjects such as probability theory and statistics, practical subjects such as the characteristics of electronic components and devices and the design and construction of electronic equipment, and some physics and chemistry in the study of failure and corrosion. He may also be involved in the environmental testing of components and apparatus under a wide range of conditions. Thus an electronic engineer who is concerned with equipment reliability will encounter a variety of mathematical and scientific problems in addition to those arising from his own particular branch of engineering.
Since the basic definition of reliability is expressed in terms of probability, the concept of probability, its prediction and measurement are of prime importance. Thus we must first of all define what is meant by probability and give rules for manipulating probabilities, and for calculating probability in various situations.
In this chapter we investigate the reliability of complete systems in terms of the reliabilities or the failure rates of the components or units used to construct them. The first requirement is thus to divide the system into units, each small enough to enable its reliability to be assessed directly. In general this requires subdivision down to component level so that we may use reliability data gathered from many different systems which incorporate the same components. Although we regard as components individual items such as a resistor, a transistor or a capacitor, the introduction of integrated circuits of increasing complexity has involved some alteration to the definition of a component.
The reliability calculations given in the previous chapter are based upon the assumption that the failure rates of the components do not vary with time. Although this assumption leads to the simplest prediction techniques it is permissible only if it represents fairly accurately the measured behaviour of components in a working system. In general terms the assumption may not hold for mechanical components, but is a good approximation for electronic components, particularly since the introduction of transistors in place of valves.
In the design of electronic equipment, reliability is one of many requirements, such as weight, size, cost, power input and performance which must be given adequate attention in arriving at the most effective product. Like most other characteristics, reliability is a feature which must be given due attention at all phases of the design, and cannot satisfactorily be added on after the prototype stage has been reached. It is, however, useful to consider the problem of designing reliable equipment under four headings, system design, circuit design, choice of components and mechanical design, including packaging and assembly methods.
