Electronic component solderability assessment algorithm by deep external visual inspection

Electronic component solderability is an essential capability in electronics manufacturing. It determines an electronic component's ability to be reliably and repeatably soldered onto a circuit board in an automated production scheme. However, conventional solderability assessment is a destructive test that is performed on samples only. A novel, automatic, and non-destructive method for assessing the solderability of electronic leads based on deep visual inspection (DVI) is presented. The solderability is correlated to the surface reflectance which is degrading by corrosion and intermetallic reactions. The inspected components' solderability is assessed by acquiring micro-features in images through a combined supervised and unsupervised machine learning system. The former is designed to evaluate the homogeneity of the leads and the latter is designed for assessing soldering leads by classifying them into leads with good and poor solderability, based on their apparent-age. The models are trained using multiple images of a variety of soldering leads from different ages and therefore different solderability conditions. The experimental results show that components from the same package do not have homogenous solderability. Furthermore, while the leads mostly fit their apparent-age, some have solderability corresponding to apparent-older components. This method allows for continuous real-time 100% component screening, with very high classification accuracy. The presented method paves the way for a radical improvement in manufacturing quality by assuring that only components with good solderability are used during assembly.