Interface development for cost effective automated IC orientation checking systems

Abstract

IC packages, in semiconductor industries, are being marked through certain sequence by either laser marker or ink markers. A critical step in the final visual inspection is the orientation check of ICs that can be determined by detecting images of pin-1 notch or pin-1 dimple, which are molded/marked on the IC packages. This study accomplishes a software interface for a simple cost-effective approach to develop an automated low cost vision system that is capable of detecting the IC orientation on a JEDEC tray for a laser/ink marker system used in the IC industries. The algorithm has the competence and flexibility to distinguish correct orientation for various types of IC packages (QFP, PGA, BGA, flip chip etc.). The workable prototype possesses input variables, control system and output variables along with software implementation using the pattern recognition template based on RGB (red-green-blue) color method for every pixel of the images. Images of Pin-1 dimple or pin-1 notch of the IC packages are considered to perform image matching with the software development. The accuracy level has been achieved to a threshold value of above 80% of the total pixels scanned between the matched images. With this threshold value, matching efficiency between two similar images has been attained near about 100%, based on the proposed algorithmic results. Illumination variation, vibration of the machine on conveyors and handlers, computerpsilas timing gap with the Web camera and low resolution (320times240 or 277times245 pixels) of the taken images have made the image matching a complicated one. However, all the constraints are taken into account for setting the threshold value and obtaining the matching accuracy and have been achieved the appropriate matching algorithm to implement the cost effective automated IC orientation checking system.