Situation:

A high-volume manufacturer had vested several years and large capital investment with an inspection equipment provider to improve the speed, consistency of results, and eliminate tedious manual inspections of its products immediately before packaging and shipping operations. These efforts, entailing multiple enhancements to one of the vendor’s recommended “standard” inspection systems, had demonstrated only limited success, missing product throughput goals, with an offline system capable of detecting only a portion of the types of defects known to occur in their products.

Target:

Taking a different approach, can an “inline” inspection system, capable of higher speeds, and able to detect a significantly higher proportion of the defect types be developed?

Plan:

When approached by the manufacturer to explore potential alternative solutions to the problem, we embraced a team approach to the problem; engaging with a conveyor vendor already utilized in the customer’s production line, camera, camera lens, and lighting vendors, a software developer, and a university. Acting as hardware integrator, we sought to identify alternative camera systems, research, and identify optimal lighting systems, and provide prototype hardware integrating both the camera/lens combinations with “best in class” lighting systems capable of delivering the required high-resolution digital images to the software developer. The software developer and university researchers were to design, build and integrate software tools capable of analyzing these images; identifying and classifying defects in units and report each unit’s location in the material flowing through the system via their unique data matrix identifier. All of this was to take place on a conveyor system already handling the products in the manufacturing line without slowing the production rate.

Result:

The prototype system was not only capable of identifying all of the critical types of defects, it also proved this capability on defects far smaller (<5 micron) than either the original system or human inspectors had identified. Additionally, the system did this “in-line” without impacting the overall manufacturing line’s production speeds. The prototype was then refined; with multiple systems built and installed at manufacturing sites worldwide.

Costs:

• Manpower: 5 Full-time Equivalent (FTE) hardware, software and quality engineers in 16-month project
• Production System(s) cost: ~$200,000 (replaced $1M+ systems)