Advanced Imaging


Advanced Imaging Magazine

Updated: January 12th, 2011 10:01 AM CDT

There's a Hole in the Bucket

Proper illumination solves a 'simple' inspection that wasn't quite so simple
Container showing
© Arch Consulting
Container showing "created" defects. The bright spots are defects. It is important to have defects that are as close as possible to the actual production defects to ensure inspection integrity.
product defect
© Arch Consulting
This defect is much closer to the type that production-line inspection must find. It is no surprise that it is also much more difficult to inspect for these defects.
case of plastic bottles
© Arch Consulting
The process for all solutions must be lighting, then optics to solve the application. In the case of plastic bottles and containers, the defect to be inspected for can be as small as a pin hole.

By Glenn Archer, Arch Consulting Enterprises

Illumination and its importance in vision applications cannot be understated. The process for all application solutions must be lighting then optics to solve the application. Many times, the customer can't supply defects, so the integrator has to create similar defective parts to properly test illumination. Once this is accomplished, the integrator can determine the complexity of the camera and or vision system needed to gather the data in a consistent repeatable manner to determine good and bad product.

A Container Application

A plastics container manufacturer approached us about finding a solution to a quality issue. Specifically, we needed to detect holes and voids in containers. The voids could be so small that they qualify as a pin hole. The manufacturing process would, on occasion, leave voids or holes and this can have particularly bad results for the end customers' desire for freshness inside the container. The hurdles to be overcome in this application were to minimize the number of cameras, "fit" the solution into a very tight space, and inspect containers traveling at high speed—approximately 11 containers per second.

The client was in the process of reworking the production line and it was an opportunity to retrofit a vision system into the manufacturing process. A key aspect of the application was determining what size hole the camera must detect. The customer had difficulty determining exact dimensions since the primary objective was to produce containers with no holes or voids. Many discussions were held to review vision-system capabilities and the differences in accuracy vs. precision vs. pixel size vs. optical distortion. We settled for detecting anything greater than 0.005 square inches.

A big issue in determining the criteria for success for any vision-system installation is how the customer will challenge the system to validate the effectiveness of the operation. This typically means that the customer needs to "create" reference good parts and more importantly "create" reference failed parts. When an inspection system is challenged with known failed parts and every failed part is rejected then the customer can have confidence in machine vision.

Creating Defects

For this particular application it is necessary to create defective containers since the number of bad products produced was extremely small—along the order of one bad part per 5,000. Since the customer was sampling the production output, there were very few defective containers to use for testing. We produced defective containers with holes in them, but as is the case in most applications, the produced defects are very different than the defects produced by manufacturing.

Specifically, the holes normally found in a plastic container are usually found in the center area of a container that was produced with extra thin walls. This can be a small area with thin wall dimensions, and the hole can be much smaller still. We would later determine that holes could be 0.010 inches square, and that is was next to impossible to produce a thin-wall-related hole for testing. When we "created" a hole by sticking a thumb tack through the side wall or bottom of the container, we would then test this defect part in the system and could find every failed part.

However we would not know if this would solve the application because true thin wall conditions may be present and not have holes. These may show up as extra bright regions on the container but not be truly defective. Our "created" defects usually looked like small bright areas on a dark background. It is pretty simple to find bright spots on a dark background. It is much tougher to find bright spots on an already brighter than average background.

1 2 next

Subscribe to our RSS Feeds