Using fixed and adaptive multivariate SPC charts for online SMD assembly monitoring

This paper explores different alternatives for online monitoring of the assembly of surface mounted devices (SMD). The paper compares three widely used multivariate control charts: Hotelling, multivariate exponentially weighted moving average (MEWMA), and multivariate cumulative sum (MCUSUM). Two different scenarios are analyzed: one in which the sampling interval is fixed, and another in which the sampling interval is variable. The results presented in this paper are part of the development of an integrated quality environment for SMD assembly. For the first scenario, where fixed sampling intervals were used, MEWMA outperformed Hotelling and MCUSUM. That is, MEWMA was faster to detect shifts in the process mean. For the second scenario, where variable sampling intervals were used, only Hotelling was tested. The results showed that Hotelling with variable sampling intervals performed better than MEWMA with fixed sampling intervals.