Case Studies: Enhanced Cylindrical Inspection
Boulder Imaging (Boulder, CO) is a systems integrator involved with machine vision. The software used with its machine vision systems is developed in-house and designed for specific applications. Boulder Imaging’s systems can be used to inspect any surface that requires a uniform finish, such as fiberglass, glass, paper or other Web applications. Furthermore, cylinder inspection, which is more challenging than Web applications, is a specialty of Boulder.
The system they have developed is called the Vision Inspector. One application of the system is to inspect the surfaces of printer drum cylinders, key components for printer cartridges used in laser printers. The machine vision system runs continuously, without any breaks.
All of Boulder’s machine vision equipment is used for industrial inspection in manufacturing plants that build products requiring surfaces with virtually no defects. There are two pieces to the Vision Inspector. First, there is the actual live inspection system that detects defects, gathers all information about defects and communicates with the robotics to sort them out. The second part is the process control station, which gives the operator a direct view of all defective parts that have been found, and classifies them. The system is, in essence, identifying defects and at the same time presenting this information in a statistical manner.
Says Carlos Jorquera, CEO and founder of Boulder, “The quality of the surface of a printer cartridge drum is directly tied to the quality of the printer page. Therefore, it’s very, very important to have a very uniform surface, otherwise field defects, field nonuniformities or any artifacts might show up on the printer page. Our Vision Inspector system inspects every single one of those drums made by our customer, a printer manufacturer.”
To get an image of something cylindrical in shape, Boulder uses Piranha 2 line scan cameras from Dalsa (Waterloo, Ontario) to put together a large image of the surface. Along with its own proprietary software, Boulder also uses Dalsa’s X64 frame grabbers to capture the image.
Jorquera explains, “We chose the Dalsa Piranha camera because it gets the resolution we needed to be able to see defects down to 20 micrometers in size. That requires us to use a camera with at least 6,000 pixels.” The camera’s light sensitivity allows Boulder to detect minute and difficult-to-see defects. The use of the camera also results in approximately 83 parts inspected per minute.
The high resolution of the Dalsa camera allows Boulder to image the whole length of the part. “Prior to using this approach, our customer was using up to three smart cameras from another source,” says Jorquera. “The cost savings from that change alone was about 40%.”
Jorquera sees machine vision components from Dalsa continuing to be important in the next generation of the Vision Inspector system. As Boulder continues to innovate, Jorquera wants the next phase to be able to incorporate adaptive learning, meaning that the machine vision system would be able to learn about defects from the data it generates.