Drives & Controls April 2022

33 www.drivesncontrols.com April 2022 MACHINE VISION n Vision-based bearing checker is on the ball P recision ball bearings need to meet special requirements to ensure their long-term, fault-free performance without lubricants in applications such as gearboxes. Ceramic ball bearings offer the best technical properties for this purpose and are used in many high-end applications. Producing the bearing balls requires the highest precision and process reliability. Even the slightest deviation from the optimum shape or the desired surface quality can result in bearing damage and possible failures. To minimise this risk, researchers at the Centre for Device Development (CeDeD) at the Fraunhofer Institute for Silicate Research ISC in Germany have developed an automated, robot-based system for inspecting ceramic bearing balls made of silicon nitride. The Ceramic Ball Check System (CBCS) has been operating since 2019. Its performance depends critically on a machine vision system. The developers split the checking process into two parts. The bearings’ roundness is first checked against appropriate standards, and this is followed by quality control of the ball surfaces. A robot feeds the balls into a test module, as well as ejecting faulty products and moving defect-free balls on to the next stage of the process. The target operating speed for the system is 2,000 balls per hour. Inherent properties of the ceramic balls posed challenges for the vision system. “The surfaces to be measured are highly reflective and require a special illumination system that minimises the reflections to solve the various measurement tasks,” explains the CeDeD’s head, Dr Andreas Diegeler. “In addition, the system had to be flexible, because various standardised ball sizes with diameters from 3–10mm had to be tested. These were very demanding requirements for a fully automated quality inspection system.” The system – thought to be the only one of its kind in the world – measures the bearing balls in three dimensions to an accuracy of 0.3μm. To achieve this level of accuracy, its uses precision vision systems with a resolution of 1μm. The system needs to detect any deviations from standardised roundness, as well as surface defects such as scratches, breaks, dents or colour inconsistencies. Such defects can be caused, for example, by overheating during the sintering process used to produce the bearing balls . A major challenge was ensuring that there no particles were present that could affect the measurements. “Even the tiniest grain of dust is identified as a defect,” Diegeler explains. “This means that the bearing ball production must be optimised in this respect as well". The test routine evaluates deviations from specified values automatically. If the tolerances for the ball shape or surface quality are exceeded, the defective balls are rejected. This allows conclusions to be drawn about any process errors during production. The vision system is integrated via OPC-UA supporting digitalised production, and allowing adjustments to be made for individual tasks. Diegeler and his team already had considerable experience with machine vision systems before embarking on the project. “In the past, we have implemented around 50 systems in seven different application areas,” he reports. “Machine vision has always been a core element of our measuring systems.” Despite this in-house expertise, Diegeler does not see himself or his team as imaging experts. They therefore carried out extensive research to find a suitable specialist vision partner. “Our starting point was the special illumination we needed because of the bearing balls’ reflections,” Diegeler recalls. They found a lighting manufacturer that was a partner of Stemmer Imaging, so were able to obtain all the vision system components from a single source. Diegeler sees promising opportunities for the future, particularly the possibility of incorporating artificial intelligence into the inspection technology. “With the CBCS automated ball-checking system, CeDeD has opened up a new application for high- precision and non-destructive optical testing technology,” he concludes. n German researchers have developed a vision-based system that checks the precision of ceramic ball bearings in three dimensions to an accuracy of 0.3μm. The “unique” system also checks for surface defects at a rate of up to 2,000 balls per hour. The vision system inspects the ceramic bearing balls to an accuracy of 0.3μm A robot feeds the ceramic balls into the inspection system and ejects any failures