In the context of increasing automation and labor shortages within the microelectronics manufacturing sector, enhancing the reliability and efficiency of quality control systems has become critical. This project presents a data-driven analysis of a printed circuit board assembly (PCBA) line, focusing on the solder paste inspection (SPI) process that follows solder paste printing. SPI systems perform high-resolution measurements of individual solder paste deposits and flag potential defects when measurements fall outside predefined specification limits. Despite their technical sophistication, these systems still rely heavily on operator judgment for defect validation.
SPI machines can produce a significant number of false positives, which directly impacts operator workload and overall line efficiency. As PCB designs become more complex and densely populated, the number of solder paste measurements per board can reach into the tens of thousands. To address this, we apply statistical outlier analysis to effectively filter and reduce the data set available for process engineers supporting the line, allowing for focused investigation of anomalous solder paste deposits. This method enables the identification of potential true defects that may have been missed by both the SPI system and the operator, ultimately supporting more reliable quality control and building a stronger foundation for automated, data-driven process improvements.
Furthermore, we used JMP visualization capabilities and Python integration to extract image of the potential defect generated by the SPI and generate automated report that are interactive.
