Silicon Module Quality Control Process for the CMS High Granularity Calorimeter using ML Tools

In the CMS experiment at the Large Hadron Collider (LHC), the upcoming High Luminosity phase necessitates the replacement of the current endcap calorimeter with the High Granularity Calorimeter (HGCAL), comprising nearly 27,000 silicon modules. A challenge during the assembly of these modules is ensuring the accurate alignment of silicon sensors with the printed circuit board (PCB) through step-hole connections, which are key to wire bonding. We build on our previous work that employed deep learning techniques, including the YOLO algorithm, for automated quality control of wire bonds. Here, we introduce an algorithm to further assess the spatial offset between silicon sensors and the hexaboards (HB) by detecting the boundaries of silicon cells and the centers of step-holes. The tool can be utilized online and it is fast, accurate, and intuitive. The algorithm enables precise identification of positional offsets and assists in optimizing sensor placement to improve the integrity of the module assembly process.