There is a strong demand for devices such as mobile phones, tablets and large screen TVs all over the globe. The business is competitive, which puts pressure on prices. At the same time, production costs are fairly high due to complex production processes. Consequently, a high yield becomes paramount for good profit margins. Multivariate data analysis (MVDA) is being employed by an increasing number of manufacturing companies to increase yield, and the electronics industry is no exception. This article provides examples of where and how real-time data analytics can be used in the electronics industry.
Multivariate data analytics helps manufacturers optimize processes and increase quality.
Increase yield to stay profitable
Continuous process monitoring is essential in all complex production processes and is especially obvious in the production of large-sized TV panels. A minor defect on the panel could mean that the whole panel must be discarded. While consumers are demanding increasingly larger TVs, they also demand perfect high-resolution at an affordable price. Smaller flat panel displays are easier to produce, but it is the large, high-end TVs that are most profitable for TV manufacturers. However, the rejection rate is higher for large panels and the yield is correspondingly lower. So how can you increase yield in these cases?
Employing real-time data analytics in complex multistep production
Irrespective of whether you are producing small or large flat panel displays, or other products that involve complex, multistep production processes, yield is an area of focus for all manufacturers. That’s one reason that, multivariate data analysis is increasingly being employed to deal with yield issues. Below is a summary of some ways that MVDA can be used for continuous process monitoring:
Detect developing problems at an early stage – The production of flat panel displays involves hundreds of steps and thousands of variables that need monitoring. Advanced systems must be in place for continuous process monitoring to minimize production errors that could lead to panels having to be reworked or discarded. Otherwise, defects could linger until the finished display is tested, leading to wasted time, money and materials. For example, the correct amount of liquid crystal is a critical parameter for panel quality. Employing real-time process monitoring makes it possible to detect developing problems at the earliest possible stage and make corrections immediately.
Predict quality – Real-time process monitoring can also be used to predict quality immediately after the completion of a specific process step. The result is improved quality control, as well as time saved on quality assessments and measurements.
Predict critical issues – In a multistep sequential process, multivariate data analysis can be used to create a top-level model from sub-models of all the processes that have been identified for critical issues. Thereby, operators can gain a deeper understanding of critical issues and the interactions of variables in the processes. Also, operators can monitor both individual processes and critical issues simultaneously using real-time data analytics.
Equipment monitoring – Equipment health monitoring is necessary for a high yield. Instead of monitoring single variables, multivariate data analysis can be used to monitor multivariable indicators to better determine the status of the equipment. This reduces the risk of quality problems or unplanned downtime due to equipment being out of service, which also helps to increase yield.
Real-time data analysis improves process monitoring
Real-time data analytics makes the difference between knowing what an optimal process should look like and being able to implement it. By starting with multivariate data analytics, manufacturers can thoroughly evaluate their processes and build a picture of the optimal process that takes into account hundreds or thousands of variables.
MVDA allows manufacturers to better understand the impact multiple variables have on each other. It can show, for example, how adjusting the temperature at a certain time in the process affects adhesion in the next step, or how air quality combined with light might affect a circuit. Then real-time analytics allows manufacturers to apply these optimal settings to continuous process manufacturing, even using multivariate process monitoring. The result means more highly tuned processes that stay under control and produce fewer defects for a better quality end product.
Want to know more?
Learn how AU Optronics, a leading manufacturer of flat panel displays, has increased its yield by 3-5% with continuous process monitoring, employing tools from the Umetrics Suite of Data Analytics Solutions.