The Mean Time Between failures (MTBF) quoted in a product datasheet is often incorrectly assumed to provide a direct indication of expected lifetime. In fact, reliability is a statistical concept – it is the probability that an individual unit of any given product, operating under specified conditions, will work correctly for a specified length of time. If a component manufacturer states the MTBF of a part as being, say, one million hours, reliability theory tells us that 9.5% of units will fail within 100,000 hours and 59.4% within 500,000 hours. Only 33.67% of units will still be operating after the stated MTBF. These predictions are valid in the flat part of the bathtub curve familiar to reliability engineers, after early (infant mortality) failures and before the rising trend associated with end-of-life failures.
For designers to assess the reliability of a complete system, they must combine the individual failure rates of all the various components. From this it becomes clear that overall reliability cannot be greater than the weakest component. Focusing effort on improving those components is an effective way to increase reliability.
We can also see that calculating reliability becomes more complicated as the number of components in the system increases. Industry-standard reliability handbooks provide invaluable help to calculate the accuracy of electronic systems, although predictions can differ from real-life performance. Experience and accumulated field data are both valuable assets in Design for Reliability.
CUI’s Tech Insights blog takes A Closer Look at MTBF, Reliability and Life Expectancy to help avoid this misunderstanding and gives a concise introduction to the subject of Design for Reliability.