A NOVEL METHOD FOR LOCATING SOLDER JOINTS BASED ON MODIFIED BINARY POTENTIAL FUNCTION

Abstract

The solder joint location on inductors is an extremely important aspect of the industrial process; in particular, the visual location plays a fundamental role. The visual-location process is often carried out by human experts. The disadvantages associated with manual location are the large amount of time required and reduced efficiency as operator fatigue occurs. In addition, as electronic products now tend to be miniaturized, portable and with dense component layout, manual location is becoming unreliable. This has prompted the development of automatic visual-location systems to speed up the location process, increase production efficiency and improve manufacture yield rate. In this paper, we propose an automatic visual-location method for solder joints to address the problem of feature extraction in digital images, using the concept of potential functions (PF). In order to make the location method more suitable for the inductor industry, the virtual external electric field concept is introduced. The proposed location system, which uses modified binary potential functions (MBPF), has been implemented and tested with three kinds of inductors. The experimental results show that the proposed scheme performs with a high degree of accuracy, even with testing samples that are significantly different in appearance.