Modified Shewhart Charts for High Yield Processes

Abstract

The conventional Shewhart p or np chart is not effective for monitoring a high yield process, a process in which the defect level is close to zero. An improved Shewhart np chart for monitoring high yield processes is proposed. A review of control charts for monitoring high yield processes is first given. The run length performance of the proposed Shewhart chart is then compared with other high yield control charts. A simple procedure for designing the chart for processes subjected to sampling or 100% continuous inspection is provided and this allows the chart to be implemented easily on the factory floor. The practical aspects of implementation of the Shewhart chart are discussed. An application of the Shewhart chart based on a real data set is demonstrated.

Keywords:

• Average run length
• binomial counts
• parts-per-million non-conforming items
• supplementary runs rules
• statistical process control

Acknowledgements

The research was completed when Dr Chang was affiliated with the University of Macau. At present, Dr Chang is attached to GlaxoSmithKline, Singapore as a Senior Statistician.