Measuring and modelling of field reliability for an electronic product

Abstract

Traditionally field reliability has been measured and predicted by the use of manufacturing data. One of the great challenges that manufacturers of electronic products face is trying to measure accurately how a product is performing in the field. Based on a literature review as well as discussions with an electronics company, it was determined that it was necessary to base field reliability monitoring on field data. The company had a previous reliability issue where a product was performing poorly in the field but by time it was identified and corrective action was taken, it had cost the company millions of dollars. For this company it was important to determine if a product’s reliability in the field was performing well or deteriorating. The relevant literature research was divided into three main fields: Introduction to reliability, field databases in use in industry and finally measuring field data. The field study focused on two products for Company X. The main tools used to measure and model the data were, Rate of occurrence of failures (ROCOF), replacement rate by month trending, replacement index (RI) modelling and control charts. Future prediction was facilitated by the use of the lognormal distribution plot. Product A had a latent defect that only exhibited in the field and wasn’t identified until nineteen months after it started shipping from manufacturing site. Product B was a very similar product to Product A however performed very well in the field. The results showed clearly the Product A did not perform as well as product B. The RI model indicated at a very early stage that there was evidence of wear-out for Product A. The p control chart would have indicated six months before the actual investigation happened that there was a problem with this product. The lognormal distribution plot was very accurate in predicting the future reliability of the product. The results show that these techniques can provide a company with a clear indication of field reliability and highlight areas of concern.