Standard off-the-shelf vision systems cannot meet the requirements of many complicated visual inspection applications. One good example is an inspection system we created last year with two seemingly incompatible requirements: high magnification and large field of view. This application required a visual inspection system with a fairly large field of view to locate the area of interest and align the part, the ability to view extremely fine detail on exceptionally small components, and the speed to improve throughput.
High magnification implies a small field of view, but a small field of view typically makes it difficult to see which portion of a larger object a system is imaging. This problem arises in applications like semiconductor and electronics inspection, and biological imaging. We considered using a zoom lens, but such a lens requires valuable time to change focus. An alternativea microscope-style rotating turret that houses different focal-length objectivesalso requires time to change views and introduces alignment problems.
Microscope optics suggested the solution, nevertheless. The objective design of many microscopes offers some flexibility in the magnification produced. Most microscopes designed for human eyes, for example, allow the user to switch eyepieces. Systems that fit a camera to a microscope use a video coupler designed to match the image size to a specific camera-array size. For a single microscope, a number of video couplers may magnify the same image from an objective lens by different amounts. This is, basically, what the application required; the final image is the same size, but the field of view differs for differing magnifications.
The application required a larger range of magnifications than the video couplers provided, but the solution concept was similar; the design retained the same objective on the front end of the system, but we used two different lenses on the back end. Both back ends could work equally well with an infinity-corrected objective.
Two beamsplitters - one for introducing illumination and one for splitting the image into two paths - help solve the dual magnification dilemma without moving parts.
We can find magnification of a system using an infinity-corrected objective by dividing the focal length of the tube lens (a lens or set of lenses set between the objective lens and the camera system) by the focal length of the objective (see figure). Our team chose two tube lenses with focal lengths that provided the necessary magnifications and fields of view.
In an ideal scenario, a system operator would be able to view both magnifications simultaneously. To accomplish this, we incorporated a 50-50, non-dichroic plate beamsplitter after the objective lens. Adding this capability involved a tradeoff: The system provided two images at once, but it was possible that light rays coming from off-axis positions would no longer pass through the aperture of the tube lens. Our design minimized the distance between the objective and tube lens to avoid vignetting, which resulted in lower light levels at the edges of the image.
Another tradeoff with the new split-image design was increased system sensitivity to illumination levels. To conquer this problem, we used a second beamsplitter to introduce integrated in-line illumination in the system.
The final design features a dual field optical system with no moving components, which eliminates several alignment issues. The larger-view system shows a field of view of about one millimeter, while the second view system has a factor-of-five higher magnification and a correspondingly smaller field of view. Because the users view two fields simultaneously, they can measure in both fields at the same time.
We built the prototype almost entirely of off-the-shelf components, reducing cost and speeding development. Later, we customized mounting components to fit the needs of the machine into which the vision system was incorporated. oe
Greg Hollows is imaging product line manager at Edmund Industrial Optics, Barrington, NJ.