Infrared solder joint inspection on surface mount printed circuit boards

Abstract

This paper reported some experimental work and the result of applying an infrared inspection system developed in the Machine Vision Laboratory, NJIT to inspect an experimental surface mount solder joint circuit board designed and manufactured by the Machine Vision Group, AT&T Bell Laboratories, Princeton, NJ. The infrared inspection system which consists of a CCD infrared camera, a Datacube image processing system, and a SUN Station is introduced. New features selected for the infrared inspection, the statistical pattern recognition algorithm utilized and the related software are presented as well. It has been demonstrated that infrared imaging technique can identify solder joints of surface mount circuit boards according to their solder volume. The correct classification rate was found to be in the range of 89% to 100%. To our knowledge, the infrared inspection of surface mount PCBs has not been reported in the vision literature.