Improving product's reliability by stress derating and Design Rules Check

Abstract

Electronic Systems Reliability Analysis is based on Failure Rates (FR), which is mostly predicted using physics of failures techniques. In fact, we assume that the “electronic design” is near to perfect, and that the FR is based only on the physical materials. However, based on field failures analysis, we have found that many failures actually occur due in large part to a poor design, and that this poor design is mostly caused by common mistakes. In this article, a new method will be introduced, which will provide guidelines to understanding the root cause of common design mistakes and eliminating them. These common design mistakes will then be transformed into a set of rules that we call a "rule check," which will ensure that these very same mistakes will be detected and eliminated in new designs. Thousands of electronic board failures were collected for the duration of one year. The data was analyzed statistically, and for the 100 most important failures a root cause analysis was performed. It was found that 50 out of the 100 mistakes occurred due to errors in the design and these errors were based on common mistakes. To avoid these mistakes from reoccurring, a dictionary of rules was created which we call "Good Design Reliability Practice" (GDRP). The statistics show that the field failures are categorized as: - 25% are overstressed components and need to be up rated, - 35% are mistakes which increase the power dissipation and if we correct the mistake the power will be reduced, - 10% are mistakes which looks as if they are software failures but actually are hardware failures. - 30% are pure GDRP mistakes, such as pull-up /pull-down resistors, technology match, resistance/ capacitance values, power supply match etc. As of today we have successfully collected over 200 design rules, which have been grouped into five categories. These rules are used in the scanning of new designs, and significantly help eliminate the mistakes from reoccurring. Doing so dramatically increases the product's reliability (power dissipation and stress will be reduced) and robustness (design will be free from design errors). The innovation of the method is that by analyzing field failure data analysis we can setup rules which are checked in new designs.