Optical inspection spots defective ball bearings

29 Aug 2007

Illumination by ring light could be the best way to see potentially dangerous flaws in ball bearings.

spotting defects

A researcher in Australia has used an optical microscope with a ring light illumination source to examine the surface of ball bearings, and believes that the technique can easily see defects, cracks and waviness (Measurement Science and Technology, 18, N73).

Tuck Ng of Monash University told optics.org that the technique had not been tried before for this application. "Ball bearings have been tested using video imaging, but the problem always lies in the illumination," he said. "The surface of the bearing is both reflecting and non-flat, which makes it hard to implement detection algorithms."

Ng's ring-light method introduces an added pseudo-dimension, a so-called 2.5 dimension, that provides a simulated representation of the bearing's three dimensional surface.

Varying the distance between the ring light and the bearing produces a distinct ring-image reflection that is scanned across the surface of the bearing and examined optically by the microscope. Any local defect reveals itself as an anomaly in the image of the ring, observed visually by an operator or potentially by an image sensor.

The set-up used a 74mm halogen ring light illuminating a 5 mm stainless steel bearing, and a 5x magnification microscope, but the technique is invariant of the magnification used in the microscope. "The regions inspected on the ball bearing are only dependent on the distance between the ring light and the vertex of the bearing," Ng said.

shining light

Defective ball bearings can cause catastrophic machine failure, and Ng's process could allow the vital inspection stage to be carried out faster and easier than current methods.

"You could apply this technique in a production line with minimal cost," he said. "In order to improve throughput you could install multiple stations and it should still be within the budget of most operations. All you basically need is a ring light, a video camera and a computer running an algorithm programmed to look for flaws."