Designed fatigue experiments to improve the reliability of liquid crystal displays

The objective of this paper is to show how reliability improvement may be achieved using statistically designed experiments. Environmental stress testing has been performed on liquid crystal displays (LCDs) to identify the parameter settings that allow an increase in their lifetime. During their operating life, the LCDs are connected to electronic devices and are inserted into cases. They are used to display measurements to aircraft pilots during commercial flights. Typically the failure time was 4000 hours of flight which was considered not to be long enough. The design of experiments provided an opportunity to check whether the most likely hypothesis was correct. During the tests, degradation rates were used as a response. The effects of three factors were studied using two overlapped 2/sup 2/ orthogonal designs. A powerful spotlight located at 2.5 meters from the tested LCDs, was used to model the degradation rate. One of the factors was found to reduce dramatically the degradation rate, allowing a very substantial increase in the product lifetimes. More generally, we propose to use the slope of the degradation function as a response in designs of experiments for reliability. The advantages of doing this are that the responses are normally distributed and when the data are right censored, it does not cause as many problems as when failure times are recorded. The drawback is that the product degradation needs to be periodically monitored until an accurate estimate of the degradation rate is available, moreover the degradation function may not be linear so that a data transformation is required.