There are different strategies to react on a detected weld defect. A very expensive one is to check the generated weld seam after welding and to put the part back into the clamping device for a further weld, if the joint does not fulfill the requirements. If a joint is vulnerable to defects, this procedure results in high costs. Therefore, there are different approaches to inspect the weld seam during the welding process in an in-process or post-process monitoring. The inspection systems reach from single sensors to camera-based systems. Coaxial systems are often preferred because of their high flexibility and small interfering contour. Most of these systems can decide if there is a defect or not, but can’t obtain the shape and type of the defect.
In the framework of the WELDone project (funded by German Federal Ministry of Education and Research), the Institut fuer Strahlwerkzeuge developed a camera system that uses two images on one camera chip, resulting from process emissions with different polarization orientations. With the aid of a fast FPGA (field programmable gate array) the orientation of the hot workpiece surface is obtained on-line. With this information attributes of several different weld defects can be recognized. In principle, it is even possible to reconstruct the topology of the surface and not only detect a defect, but also characterize it.
Since different defects require different treatments of the weld, the weld repair system requires a method to determine if there are defects and which type and size they have. The structures observed by the newly developed monitoring system and the measures taken after decision will be discussed exemplarily for two different defects in overlap welding. For the case of a hole, generated by melt ejection, the preferred method to close the hole and keep the sheets connected is to weld again with adjusted power. An undercut needs a different treatment. Here a defocusing yields best results.