Enhancement of Placement Accuracy for SMD via Development of a New Illumination System

摘要

Ever since surface-mounting technology (SMT) for printed circuit board (PCB) assembly processes has been developed, electrical products continuously tend toward the miniaturization of components, with denser packing of its boards. With the increasing necessity for reliable PCB product, there has been a considerable demand for high speed, high precision vision system to place the electric parts on PCB automatically. To recognize the electric parts with high accuracy and reliability, illumination condition is instrumental to acquisition of part images. In this paper, a versatile lighting is developed which utilizes three different types of illuminating methods: direct, indirect, and back-light illumination. The direct illumination uses to recognize the flat and specular surface of electric parts such as lead, and indirect illumination to recognize the lambertian surface or curved specular surface such as J-lead, BGA ball and gull-wing type lead. Finally, the back-light illumination is used to extract the orientation and location of the electric parts from acquired image. The recognition algorithm, since geometry and dimension vary with part, has been aslo developed for various electric parts. The recognition algorithm consists of two different image processing methods. One is binary image based algorithm to increase the speed of image processing algorithm and to recognize parts and defects, and the other is gray-level based algorithm to increase the accuracy of recognized parts pose. The performances of the developed vision system and recognition algorithm are tested on a number of samples of the electric parts used for PCB board. Experimental results reveal that the proposed system recognizes various type of electric parts with good accuracy and fast speed.