Intelligent X-ray inspection for quality control of solder joints

Today automated X-ray inspection is a reliable technique for inline process monitoring and 100% quality control of solder joints. In order to further optimize X-ray inspection there are three key tasks: (1) improvement of X-ray sensors with respect to resolution and contrast; (2) development and integration of three-dimensional (3-D) tomographic reconstruction algorithms in order to separate superposition caused for example by double-sided boards; (3) improvement of image processing and machine learning algorithms in order to provide a detailed quality profile of solder joints necessary for inline process control. In this paper, primarily (1) and (3) are addressed, An inspection system is introduced, which is based on a hierarchical inspection strategy in order to meet speed as well as reliability requirements. Within this concept two-dimensional (2-D)-inspection based on single X-ray projections and 3-D-inspection based on multiple projections are combined. Most solder joints are rapidly evaluated based on single 2-D-projections while in complex situations one or more off-axis projections are required. For example, superpositions caused by double-sided boards are separated by 3-D-reconstruction based on multiple projections from different directions. The advantages and limitations of 2-D and 3-D techniques are discussed within several examples. For 2-D-inspection image preprocessing, classification and defect learning by neural networks are addressed. For 3-D-reconstruction of solder joints digital laminography, model based reconstruction and microtomography are investigated