Issue link: https://iconnect007.uberflip.com/i/831357
78 SMT Magazine • June 2017 tomatically identify the presence of lead on a PCB. Conclusions We have presented a newly-developed tech- nology that enhances the capabilities of 2D X- ray inspection for the electronic industry. It provides additional material type and thickness information that has not been available to date in 2DX inspection. The key to the technology is a physical structure known as the Multi Absorption Plate or MAP. This is coupled with sophisticated ma- chine learning algorithms and training meth- odologies. Several cases were discussed to dem- onstrate applications in the electronics, securi- ty and food safety industries. Below are key en- hancements gained through MAP technology: • Enables users to visualize samples in a new way by looking at a materials image as well as a standard grayscale absorption contrast image • Detection of defects and impurities that are invisible in the regular 2DX grayscale image • Quantitative thickness and material information returned to the user • Can be fitted to most flat panel detectors • Near real-time operation • Ability to adapt algorithms for automated inspection applications The authors are very interested in discus- sions, ideas and collaborations within the elec- tronics industry in order to focus the develop- ments and take full advantage of this new, ex- citing technology. Acknowledgements The authors would like to thank Tamzin Laf- ford for her great help during the final stages of the paper development. SMT 2D X-RAY INSPECTION WITH MATERIALS AND THICKNESS IDENTIFICATION Figure 9: Illustration of how the MAP technology can be used to measure precise Cu track thick- ness information. This can be used to generate thickness profiles (top right) or a full three- dimensional representation of the copper thickness across the PCB (bottom). Figure 10: Materials curves for lead-free and lead- based solder samples obtained using the MAP technology installed in a Dage Diamond 2DX in- spection system. The two materials curves are clearly separated, showing that a machine learn- ing algorithm could be developed to automatical- ly identify the presence of lead on a PCB.