Track 2 Session 10
9:10 to 10:10 a.m. Friday March 30, 2012
Predicting Field Failure Rate From Production Process Yield Data Using Real Life LCD TV Production Data
The cost of building a sufficient quantity of LCD TV test units to make statistically valid reliability predictions is continually increasing as design and manufacturing failure rates continually reduce to such low levels that 90% confidence predictions require extremely large sample sizes tested over long periods. This makes cost-effective reliability testing very difficult. The “Reliability Solutions” model is based on combining LCD TV test and manufacturing data to provide a method to predict field failure rates per month based on monthly data that gives a true estimate of failure rates per monthly vintage of production while combining with volumes produced. This presentation shows a method that correlates closely to actual field data, enabling the producer to set continual improvement targets and predict the effects on reducing field failure levels as targets are achieved. Ongoing reliability test data is also used to make the prediction model fully robust and combine reliability test data with respective manufacturing data drawn from the same vintage of product. Manufacturers can now set KPIs (key performance indicators) with confidence that achieving them can and will produce a product that has a field failure rate that is in line with their customer service targets while managing lower cost investment in accelerated life testing needs.
Key Words: Failure Rate Prediction, Yield, Early Life