Reducing production time in pick-and-place scenarios with optimized 3D printed matrix trays

Abstract

The aim of this thesis is to optimize the automated pick-and-place process of surfacemount devices on a printed circuit board. Therefore, we develop a 3D-printed tray feeder, from which the robot takes the components to place them. The software is based on a relational database and a user interface to manage the data and guide through the process. Relevant for a placing order which aims to minimize the placing time are the minimization of time-consuming nozzle changes and the avoidance of collisions between the nozzle and already placed components due to different component heights. Also, the tray area is minimized with a first-fit decreasing height strip packing algorithm with strip width variation. We add several pre-processing steps like an aspect ratio modification and a new rotation modification. This algorithm calculates fast and the new rotation modification applies to both analyzed examples. The 3D-printed tray feeder works successfully, but there are time-consuming preparation steps due to propriety file formats of the machine software. The developed workarounds are still fast and easy especially for inexperienced users, but are a new potential source of errors. That is why currently a 3D-printed tray feeder is worthwhile especially for components with former complicated feeder situation. The use for all components would require a better connection with the machine software. This is a limitation of the tray area minimization too, as the software does only accept similar parts in a regular grid. Additionally, the positive effects of the pre-processing steps could increase with further effort in small calculation time