應用迴歸分析與田口方法提升測試良率：以K公司為例

Apply Regression Analysis and Taguchi Method to Improve Test Yield: – a Case Study of K Company

梁興岳(Hsing-Yueh Liang)

指導教授 ： 邱銘傳

國立清華大學/工學院/工業工程與工程管理學系碩士在職專班/碩士(2019年)

摘要

積體電路技術發展已進入成熟階段，以此為基礎發展出微機電系統技術(Micro-electromechanical Systems，MEMS )。不同於傳統的 IC 產業鏈，微機電系統技術因結構較複雜，包含化學、光學、機械功能，其設計、製造、封裝、測試，依產品應用不同而有所差異，因測試設備需模擬各種MEMS感測器使用環境，測試機台設備因較複雜，精度也要求較高，造成測試設備成本要比一般IC測試設備成本要高。
陀螺儀測試分類機因是測量角速度，是屬於慣性原理。對於測試慣性感測器機器設備需提供轉動，震動，翻轉等物理條件捕捉感測器的輸出信號，以檢測出產品好壞，測試之穩定性與持續精度以確保測試品質是一項重要指標。
本研究開發了一種利用迴歸分析和田口方法解決上述問題的綜合方法。迴歸分析的目的是找到關鍵因子，並使用田口方法實驗設計以找出更好的最佳參數。因此，分析了每個對照實驗的S / N比和最佳水平，以達到最佳組合公式。實際產品驗證10批，驗證結果顯示震動雜訊測試項之(在量測其水平震動雜訊) 測試良率由原本的0.8-1.2%之間，下降到0.2%以下，已達客戶標準0.69%之內。

關鍵字

微機電系統技術 ； 陀螺儀 ； 測試分類機 ； 田口方法 ； 迴歸分析

並列摘要

The development of integrated circuit technology is in mature stage, based on which Micro-electromechanical Systems (MEMS) technology. Different from the traditional IC supply chain, MEMS technology which includes chemical, optical, and mechanical functions is complicated in structure. Especially the design, manufacturing, packaging and testing vary in product applications. Since test equipment needs to simulate various MEMS sensor usage environments, test equipment is more complicated and requires higher precision, which results in higher test equipment costs than regular IC test equipment.
The gyroscope test sorter is a type of test equipment that measures the angular inertial velocity. Hence this machine needs to provide physical signals such as rotation, vibration, and flip to capture the output signal of the sensor and determine the yield of product. How to improve the testing stability and continuously accuracy of yield to assure test quality is an important mission.
This study developed an integrated method that utilized regression analysis and Tahuchi method to address the above issue. The purpose of regression analysis is to find the key factors, and the Taguchi method is used to design the experiment to figure better process parameters. Therefore, the S/N ratio and the best level of each control facto were analyzed to achieve the best combination formula. According to the testing results of 10 lots, the Noise (in measuring its horizontal vibration noise) of test yield declined from 0.8-1.2% to less than 0.2%, which has reached within 0.69% of the customer standard.

並列關鍵字

MEMS ； Gyroscope ； Taguchi Method ； Regression Analysis

參考文獻

英文文獻

1. Castro, S., Dean, R., Roth,G., Flowers,G.T., And Grantham,B.(2007) Influence of acoustic noise on the dynamic performance of MEMS gyroscopes. In International Mechanical Engineering Congress and Exposition, American Society of Mechanical Engineers. ASME IMECE 2007-42108

2. Chen, S. H., Ko, C. S.(2018) The Effects of Different Parameters on the Temperature Field of Plasma Assisted Machining Titanium Alloy. Journal of Chinese Society of Mechanical Engineers, Vol.39, No.5, pp.295~301

3. Chen, W. S., Yu, F. J., Wu, S. H.(2011) A Robust Design for Plastic Injection Molding Applying Taguchi Method and PCA. Journal of Chinese Society of Mechanical Engineers, Vol.7, No.2, pp.1~8

4. Chen, Y. T., Tsai, S. H., Shei, H.J.(2016) ANOVA for composite fan of prediction and analysis. Journal of Chinese Society of Mechanical Engineers, Vol.65, pp.1~17