微型銑刀崩刃自動光學檢測系統之研究

Abstract

PCB生產製程中，由於產能需求大、精密度要求高，再加上板材複雜化，因此高轉速、高精度、高進刀速的成型機已漸被採用，而要實現高速成型及高精度加工目的，PCB成型銑刀的品質便是關鍵因素之一。崩刃便是影響銑刀品質的一項嚴重瑕疵。如果利用有崩刃之微型銑刀切削電路板時，將會影響PCB使用時間與品質。目前銑刀製造廠大多數是利用人工目視檢測崩刃瑕疵，以確保銑刀的品質。然而當銑刀外徑越來越小時，人工目視檢測就漸漸顯的不敷使用。本論文的目的，是在發展出一套適用於檢測微型銑刀崩刃瑕疵的自動化光學檢測(AOI)系統。由於同一型號的銑刀外觀可能會因批次的不同而有所差異，這將會增加開發此系統的困難度。因為銑刀在刃身與魚尾的外觀上截然不同，所以我們分別開發兩套子系統，用以檢測銑刀不同部位的崩刃瑕疵。第一個子系統主要視檢測刃身部分的崩刃瑕疵，主要是利用影像重組與相關的檢測演算法，找出崩刃的位置。此一子系統的成功率高於96％。第二個子系統主要是偵測魚尾部分的崩刃瑕疵。此一系統是利用所開發之光源系統配合品質管制圖，進行魚尾崩刃瑕疵的偵測。其成功率高於99.2％。此一檢測系統的特性是微型銑刀可以隨意置放於夾具上，即可進行檢測，不需要精準的校正即可使用。實驗的結果也顯示了此一系統能夠正確、穩定、快速地偵測不同外徑的微型銑刀之崩刃瑕疵。

In this paper we proposed an auto-optical inspection (AOI) system which can inspect the microrouter collapse automatically. The microrouter is a tool used to cut a printed circuit boards (PCB). The trend towards higher density and smaller size PCB requires finer microrouter to cut with higher precision. A few kinds of defect could occur on the microrouter and cause unexpected damage to PCB. Among these defects, collapse is the most critical one to be detected. The microrouter manufacturing company used to rely on human inspectors to control the microrouter’s quality. As the diameter of the microrouter gets smaller, it is more difficult for human inspectors to examine the microrouter’s defect. The same type microrouters from different batches may have different geometry that increases the difficulty to develop the AOI system. Due to the geometry of microrouter body and fishtail are different, the proposed AOI system was divided into two subsystems to conquer this problem. The first subsystem focuses on the microrouter body, it utilizes the image reconstruction and associated inspection methods to identify the position of collapse. The successful detection rate is above 96%. The second subsystem focuses on the fishtail. Some extracted invariant features of the fishtail are heighted by the developed lighting device and utilized to build quality control charts to detect the collapsed blade. The successful detection rate of the fishtail subsystem is up to 99.2%. These two subsystems are designed based on the concepts of mounted-free and easy use. Experimentation results revealed that the proposed AOI system can successfully detect the collapse of different sizes and different types of microrouter accurately and robustly.