Since 2014, MO.S.A.I.C’s team of engineers have designed, built, and maintained vision systems for sectors including automotive, food & beverage, paper, logistics, and pharmaceutical production.
A key inspection system in their automotive vertical conducts surface quality inspections on car and truck brake disks. Quality assurance is integral to the production of brake disks, which are made through a casting process characterized by using sand as the mold material. From start to finish, the production process for brake disks encompasses sand-casting, sand removal and disk separation, cleaning/blasting, quality inspection, and disk machining phases.
Prior to the upgrade, human operators would evaluate the quality and condition of the disks, checking for sand and sintering traces, and removing bad or non-optimal units from the production line to avoid damaging the machining tool.
Marco Pistilli, IT engineer, machine-vision system and software developer with MOSAIC, notes, “with human inspectors, it is natural that product samples will be assessed differently by different operators. The likelihood of these differences increases after humans have worked a few hours and are tired.”
MOSAIC sought an enhanced quality-control solution—using machine vision technology—to identify bad units more reliably and effectively, “because with a vision system,” Pistilli continues, “in addition to the enhanced quality control, the inspections are repeatable 24 hours a day and the machines perform faster than any operator could.”
The sand-casting process complete, a finished brake disk undergoes a scan to check that the surface is free of particulate traces. Image sensors are more responsive to lower wavelengths, which accounts for the unique purple tone of the image. Working with purple light best brought out the defects, allowing the MOSAIC team to shorten the exposure and thus speed up the vision inspection.
Their team decided upon Matrox® Imaging’s software and components, largely “because of the excellent technical support, software features, performance, and great price-to-quality ratio their products afford,” Pistilli affirms.
For the highest quality standards, the custom vision system was tuned to highlight subtle defects that only experienced quality inspection workers would notice. MOSAIC selected flowchart-based Matrox Design Assistant® X vision software because of its effectiveness, precision, and rapid response.
The software runs on an HP workstation equipped with an Intel® Core™ i9 processor and three Matrox Concord PoE frame grabbers, which acquire and process the images coming from nine Teledyne DALSA cameras—seven Genie Nanos and two line-scan models. Each camera has a different shooting position, capturing images while the disks pass through the inspection tunnel on the moving production line. A custom-built lighting system uses MidOpt filters on camera lenses and spotlights. The vision system interacts with a programmable logic controller (PLC) via fast-scan modules and an encoder; hardware triggers come directly from the PLC.
The new Matrox Imaging-based vision system performs quality-control inspection faster, with greater accuracy and repeatability, than the prior manual inspection process. The result? Fewer non-optimal brake disks passing through production.
The MOSAIC team had three objectives for the new vision system. First, they wanted to inspect every angle of the disk surface without stopping production or needing to invert or rotate disks as they passed. The short cycle time made physical displacement of the disk an encumbrance to the process, so integrating nine different cameras to catch every disk angle helped address this concern.
They also wished to conduct quality assurance for more than 200 different disk models using, ideally, the same vision setup. “By developing an effective background removal algorithm in Matrox Design Assistant X to process all the disk models with the same logic and processing algorithm, there is no need to redefine regions of interest,” Pistilli reports.
Finally, precise image acquisition was central to system development. Matrox Design Assistant X’s convenient Operator View lets the operator manage recipes, create sets of parameters for each camera on the line, and customize the image color-map to best match their needs. A filmstrip recalls previous pictures and relative datasets for further analysis. A dataset is generated for each analysis for further investigation from quality-assurance department.
The Operator View within Matrox Design Assistant X streamlines management of recipes and parameters. The filmstrip format makes it easy to retrieve previously acquired images and datasets.
Thanks in in part to the successful implementation of this automotive vision system, the MOSAIC team are exploring building the system on Matrox Imaging hardware and software. “Matrox Design Assistant X software allows us to quickly deploy feasibility studies and application demos within the production software. The completeness of its tools helps us to work on both 2D and 3D systems,” Pistilli concludes. “Truly, Matrox Design Assistant X offers us all we need to deploy an effective machine-vision system.”
