多品質智慧型參數設計－以環保點膠製程為例

Multi-response Artificial Intelligence Parametric Design －Example of Environmentally Conscious Adhesive Dispensing Process

蘇雅姍(Ya-Shan Su)

指導教授 ： 黃乾怡

國立臺北科技大學/管理學院/工業工程與管理系碩士班/碩士(2013年)

摘要

現行電子產業多以表面黏著技術及波峰焊製程進行印刷電路板組裝(Printed Circuit Board Assembly ; PCBA)。對於雙面電路板之組裝製程而言，通常會藉由波焊載具屏蔽第一面PCB上之表面黏著元件，以避免其焊點接觸錫波而再次熔融，造成元件偏移或沖刷掉落。然而，使用載具導致成本提高因而運用點膠製程代替波焊載具，且點膠之烘烤與迴焊同步進行，以提升生產效率。由於膠水屬於環氧樹脂之化學製劑，其硬化過程所需之烘烤溫度與時間有所不同，高溫度、長時間之烘烤會造成造成溫室效應問題。
本研究以音響產品PCBA為例，針對低成本之CEM-1板材及0805陶瓷電容，評估低烘烤溫度之膠材，並開發環保點膠製程，以達到節能減碳之環保目的。研究中運用田口方法規劃實驗，考量膠材種類、膠水熱固溫度、迴焊爐輸送帶速度及點膠壓力等控制因子，評估組裝可靠度與製造成本等多重品質特性，其績效指標分別為膠水推力強度與生產每片PCB板所需成本。應用變異數分析、主成份分析等統計方法，並建構基因類神經模型，分別決定最適製程參數組合。進而進行確認實驗，結果顯示，基因類神經模型所建議之最適參數組合優於其他兩種方法，得最佳製程參數組合為：膠材IR-130HF、熱固溫度140℃、迴焊輸送帶速度1.1公尺/分鐘、點膠壓力0.09Mpa。最後，將最佳製程參數組合導入點膠製程之量產環境，於觀測後續波焊製程中發生元件掉落之現象，結果顯示，最佳製程參數條件下生產之PCB元件掉落率1.5%低於初始參數2.5%。決策者使用新開發之低烘烤溫度膠材，並導入本研究所提出最佳製程參數組合，可有效提升組裝可靠度，同時降低產品之製造成本，兼顧環保效益。

關鍵字

表面黏著技術 ； 田口方法 ； 主成份分析 ； 類神經網路 ； 基因演算法

並列摘要

In the electronics manufacturing industry, the surface mount technology (SMT) and wave soldering process are used for the assembly of printed circuit board (PCBA). In the double sided printed circuit board (PCB), the soldering carrier is commonly used to cover the components on the first side. This is to avoid the component from being contact with the molten solder and falling off. However, using the carrier will increase the production costs. The adhesive dispensing process eliminates the necessity of carrier. The curing of adhesive and reflow soldering are performed simultaneously to improve the production efficiency. However, the curing of adhesive will further contribute to the greenhouse effect.
This research uses a sound card as example and investigates the adhesive dispensing process to conserve energy and to reduce the emission of carbon-dioxide. The composite epoxy (CEM-1) PCB material and the 0805 passive components are under consideration. The lead-free solder 58Bi/42Sn and a low baking temperature adhesive are used in this research. Taguchi method is used to investigate critical factors such as adhesive, curing temperature, conveyor speed and the dispensing pressure that may influence the assembly reliability and the manufacturing cost. The performance indicators are adhesive shear strength and the corresponding expense to produce each PCB board. The analysis of variance (ANOVA), principal component analysis (PCA), and the genetic algorithm based back propagation network model are exercised to handle the multiple quality characteristics problem and to determine the optimal process parameters. The results of confirmation test showed that the parameters determined by genetic algorithm based back propagation network model is better than the others. The optimal parameter are adhesive IR-130HF, curing temperature 140℃, conveyor speed 1.1m/min and the dispensing pressure 0.09Mpa. Finally, the optimal process is implemented in the production line to verify its effectiveness, showing that the PCB components drop rate decreased to 1.5%.

並列關鍵字

surface mount technology(SMT) ； Taguchi methods ； principal component analysis ； artificial neural network ； genetic algorithm

參考文獻

[10]姚威良，迴焊溫度曲線最佳化，碩士論文，樹德科技大學經營管理研究所，高雄，2010。

[15]程一鵬，使用模式樹建構SMT錫膏印刷製程品質管制模式之研究，碩士論文，國立成功大學工業與資訊管理碩士在職專班，臺南，2008。

[3]王芳芳，結合田口與多準則決策方法求解穩健供應鏈資訊共享策略，碩士論文，國立成功大學製造工程研究所，臺南，2007。

[16]程紀嘉，具多重反應值製程之要因分析，碩士論文，國立成功大學統計學研究所，臺南，2006。

[8]李文宏，IC封裝材料對模具正向及剪向黏著力量之研究，碩士論文，國立成功大學工程科學系，臺南，2005。