One of the injection molding machines operated by Walter Söhner GmbH and Co., a manufacturer of precision plastic parts, produces about 5,000 coil forms an hour. In a standard three-shift cycle the output totals more than 120,000 parts per day. The machine already has produced 62 million coil forms in two different versions. The most recent 50% of these parts has benefited from a new quality inspection system. "Thanks to reliable monitoring by the machine vision system, we have not delivered a single faulty part to the customer in a year. In other words, we have achieved an error rate of zero parts per million (ppm) in real terms," Jochen Neugart, head of production for Walter Söhner, says.
The company had already successfully deployed industrial machine vision systems for quality assurance purposes in other areas of production. But the new application has broken entirely new ground. The company's goal was to reliably detect sporadically occurring defects in plastic coil forms. When supplying the automotive industry with components like these, it is vital to eliminate defects that could disrupt subsequent production processes. Walter Söhner required an absolutely dependable quality monitoring solution using industrial image processing and analysis that provided fully automatic sorting based on clear pass/fail criteria.
The company obtained quotations from two system providers. According to Neugart, his company also was keen to identify a supplier that could provide consistency with regards to know-how and collaboration. Another key factor was the ability of the supplier to implement the solution within a short period. The planning, bidding and decision-making process was completed within two months.
The decision was to purchase the solution offered by Octum electronics GmbH, a Cognex PSI (Partner System Integrator). This decision was partly based on the quality of the quotation and partly on the prior experience of successful solutions from the company.
The coil forms are placed at intervals of 30 millimeters on a special fixture, in two rows of eight, and then inspected to check that they are fully formed. With a belt speed of 9 meters per minute, the maximum analysis time per fixture is 1.6 seconds. The high-precision shape inspection of each coil form must be completed within 100 milliseconds.
During the inspection operation, the entire contour of each component is inspected and compared to a specified set of values by two cameras simultaneously capturing images from different positions and perspectives. The two cameras are angled at 40 degrees to the horizontal, and are offset at 90 degrees to each other. With two rows of parts in the fixture, two cameras and one stroboscopic lighting unit are deployed for each row.
These demanding requirements are satisfied using a PC-based vision system, supplied with advanced and reliable image processing and analysis tools. The key components of the high-performance vision system supplied by Cognex Corp. (Canton, MA) is the MVS 8120 frame grabber and the PatInspect software package.
PatInspect is a further development of Cognex's PatMax software technology. The software's tools are invariant-they are insensitive to changes in the object's position, orientation or scale, and unaffected by fluctuating light and contrast conditions.
The software adopts the principle of geometry-oriented object description to differentiate parts according to a series of individually and sensitively determined error criteria. Minor adjustments can be made quickly during production by modifying a few parameters. Inspection levels can be adjusted to define the quality required.
Despite the complex requirements, the project was implemented quickly. The complete testing station, comprising the handling system, part presentation equipment and quality control system, is located alongside the injection-molding machine.
After a short start-up phase the testing station was at full production capacity, and the error rate of zero ppm was achieved within one month.
Neugart described another benefit that is of even more significance from their customer's perspective, "We demonstrate a thorough understanding of quality and demonstrate our substantial technological expertise. We are able to show the customer that more extensive automation and more efficient production go hand-in-hand with much higher quality."
With the software, inspection levels can be adjusted to define the quality required.
Check out the October 2020 edition of Quality: Understanding laser trackers, Industry 4.0, All-in-one QMS solutions for practical data management, how Edge AI improves the visual inspection process, and much more!