Title :
Automatic detectand match of LED dies basing on position relations betweenadjacent dies
Author :
Wu, Tao ; Li, Bin ; Wang, Long-wen ; Huang, Yu
Author_Institution :
Nat. Numerical Control Syst. Eng. Res. Center, Huazhong Univ. of Sci. & Technol., Wuhan, China
Abstract :
In order to improve production efficiency, inspecting and sorting of LED dies are done respectively in prober and sorter. During the inspection, dies may be broken, fragmentized, tiny moved, or deflected slightly, besides, visual scan can also lead to location deviation when the wafer is moved from prober to sorter. All of this can cause some dies mismatch in the sorting stage, which is never allowed. In order to get integrated match of dies, the position relations between adjacent dies are considered, techniques including diagnosing criterion of position relatives, redundant visual scanning, forward and backtrack algorithm with cross priority are put forward. It is proved with experimental results that such re-index approach is more applicable than traditional index method. When dies arrange in good order, match rate is more than 99.5%, by compassion, that of traditional index method is less than 92%. When the wafer is deflected slightly or locally fragmentized, match rate of the former is more than 92.5%, by contrast, that of the latter is less than 50% in same situation.
Keywords :
dies (machine tools); materials handling; LED dies; adjacent dies; automatic detect; sorting; Cybernetics; Dies; Indexes; Light emitting diodes; Machine learning; Magnetic heads; Visualization; Automatic match; Backtrack algorithm with cross priority; Die prober; Die sorter; Position relatives; Redundant visual scanning;
Conference_Titel :
Machine Learning and Cybernetics (ICMLC), 2010 International Conference on
Conference_Location :
Qingdao
Print_ISBN :
978-1-4244-6526-2
DOI :
10.1109/ICMLC.2010.5580907