在晶圓封測產業中，產品經過一連串的加工並逐漸改變其產品特性從進料到成品；並要確保其交貨到客戶後的產品品質。產品製造完成後，會針對每批產品做完整的抽樣並做不良狀況測試，以了解良率及確定是否可達成客戶或是公司的期望規格，並且可以提供評估產品品質好壞的重要指標，以了解製程品質的穩定性，同時為了提昇測試良率，往往會重工有不良問題之產品。 在過去研究中，這些研究描寫的在各個環境下多階段檢驗的問題使如何使有彈性、多樣性在檢驗的各個程序中。在分析的過程中，首先測試工程師必須先分析良率是否符合目標值，如果抽樣良率低於目標值，可考慮針對整批執行全檢動作。由於不良的原因可能來自於製程異常、測試設備不穩定，或是因測試作業不當而造成；除了必須分析造成良率異常的原因外，還要提升檢驗的效益。 本研究所要討論的是如何去決定檢驗策略使成本最小且能使品質達穩定，所要考慮的成本有檢驗、維修報廢等成本。 本研究收集案例公司於 97 年某產品的良率之相關資料以供研究分析，論文首先探討案例中所用的抽樣計畫為何，並尋找最佳之檢測策略並使達到最小成本以符合最佳的測試流程，為此論文之主要貢獻。 In IC packaging industry, a product usually under a series of operations that progressively alter the nature of the incoming material until it reaches the consumer in the form of a finish good. We using some inspect policy to ensure the quality of a finish good. In general, these papers illuminate certain aspects of the multistage inspection problem the most important of which are perhaps the flexibility, variety and complexity of options available in screening procedures. In order to find the inspection policy, this paper describes how to decide inspection policy in order to minimize the cost. Since our problem is to minimize the cost, the cost minimization at each QC gate is our objective function. The total cost includes the purchasing cost, inspection cost, rework cost, and scraping cost. This thesis attempts to propose a workable solution for inspection strategies process to establish how to allocate Sample size and maximum number of defective items that can be accepted in a given sample so that the cost of inspection for overall stage is minimized. Through a real example from macro SD probing process, it was presented to demonstrate the methodology.
