Montreal's AV&R Vision & Robotics develops inspection system to analyze pressed glass
The process of fabricating reflectors and lenses for lighting components requires glass, melted at a high temperature, to give the final product its shape.
Understandably, the outcome must be free of flaws, but it’s during the glass preparation and pressing stages that problems can—and do—occur. Defects could include foreign components embedded in the glass (like rocks or gas bubbles), incorrect part dimensions, cracks or locally missing material at the outer diameter of the parts. Some of these defects, like missing material at the circumference, can create leaks between the lens and the reflector when they’re sealed together.
As an inspection solution to prevent such inconsistencies, Montreal-based AV&R Vision & Robotic created a network of four inspection processors linked via Ethernet to a centralized statistical analysis database, which performs the quality inspection of reflectors and lenses after the glass-puck pressing process. AV&R developed the PC-based system—which, according to the company, offers flexibility for future improvements based on the results of the inspection system—using the National Instruments (NI) software library and toolkits.
For image acquisition, AV&R used a high-resolution GC1280 GigE Camera designed by Burnaby, B.C.-based Prosilica Inc., which manufactures digital cameras for machine vision and industrial applications. The GC1280 runs 27 frames per second at 1280×1024 resolution over the GigE Vision-compliant gigabit Ethernet interface. AV&R selected the Ethernet protocol because of its robustness even when the powerful glass induction oven emits highly electronic noise.
To enable communication between the software and the cameras, AV&R implemented NI-IMAQdx driver software, which allows users to acquire, save and display images from thousands of different cameras, and created its image inspection routine using the NI vision library. The result is a system that carefully manages inspection results because the data is pushed in two directions: pass/fail results are sent using an NI PCI-6514 digital I/O interface to a programmable logic controller (PLC) for rejection of defective parts; at the same time, inspection data is sent via Ethernet to the statistical database computer for monitoring.
In most of its machine vision applications, AV&R is able to control all factors, such as acquisition devices and lighting, to ensure maximum performance. In this application, however, the product generates its own light because the glass is incandescent. To deal with this variable, AV&R’s engineers were able to gain enough temperature control for the glass to stay safely within the camera’s range; normalization of the image ensures enough product stability to avoid external light sources. But these high-heat conditions (as high as 160°F in summer) required that the acquisition equipment be shielded with an enclosure and cooled using a Venturi device (a flow device for generating a vacuum by means of air flow through a restricted orifice) working with compressed air. If the cooling system is not working properly, a security feedback device in the system generates an alarm to avoid any deterioration of the equipment.