Surface mounting technology is a integrated technology that electronic components were soldered and encapsulated in printed circuit board. Generally, the setting parameters of glue dispensing is entirely decided by the engineer's knowledge and experience, but cannot be easily taught by the experiences. The Taguchi method can effectively resolve the excess parameters and levels that makes too large orthogonal array and many number of experiments. And the Artificial Neural Network can model, predict and resolve the problem of experimental design with a large number of continuous parameters and levels. This was aimed at glue dispensing process experiments of surface mounting technology. We obtained data from the destructive test of vertical and horizontal thrusts. According to the cause-effect-diagram, this research chooses controllable factors of crimping height, the setting temperature of heat, the location of glue dispensing, and this research uses the methods of Taguchi Method, TOPSIS, Artificial Neural Network and complete permutation to analyze the problem of multiple quality characteristics, in order to obtain the optimal parameters combination. The results reveal that the best optimum factors level: the crimping height is 0.2 mm, the setting temperature of heat is 150°C, the glue location of dispensing is level 3.
表面黏著技術(Surface Mounting Technology, SMT)是一種在印刷電路板上的電子元件銲接之封裝印刷電路板的組合技術。一般對於點膠參數的設定完全取決於工程師的知識與經驗,不易進行經驗傳承。而田口方法可以有效解決參數與水準過多而造成直交表過大、實驗次數過多的問題,且類神經網路可建模預測,並解決大量連續性參數及水準最佳化設定之問題。本研究針對SMT點膠製程實驗,進行直推力與橫推力的破壞性試驗,以取得數據。透過要因分析選定影響因子為壓著高度、熱固溫度及點膠位置,使用L_9(3^4)直交表重複兩次且雜音因子為印刷電路板上兩個不同位置,而本研究將使用田口方法、理想解類似度順序偏好法、類神經網路建模與完全排列組合法來分析多重品質特性的問題,以找尋最佳的參數設定組合。結果顯示,本研究最佳因子水準組合為壓著高度0.2mm、熱固溫度150°C及點膠位置水準3。