Is Machine Vision Right for You?

As a design engineer at an OEM creates a machine for production, the first basic question is: "Do I want to inspect this part, or this product at all?"

Yes, inspection leads to higher quality, but there are instances where the cost of inspection and the product's quality requirements versus its value may not warrant inspection. Think about toothpicks: If there is a toothpick or two in the box of 500 that is not quite right, most people are not exactly worried about it.

But for many products, inspection is a real necessity. To this end, many OEMs build machine vision inspection into the systems that they sell to end users. Machine vision can add value to the whole system by increasing the productivity and accuracy of the manufacturing process and reducing costs for the system's user.

So how does a design engineer know when machine vision is right for their system? Although the first, basic machine vision systems were introduced in the 1970s, the industry is just revving up in terms of mainstream adoption. This may lead design engineers to wonder if it is right for their application, as well as trying to justify the cost versus the benefits of building machine vision inspection into a design.

Industries with highly complex products, such as semiconductors and electronics, have traditionally driven the market for machine vision due to their complexity and miniaturization. But today, all types of industries from automotive to pharmaceutical to paper processing are relying on machine vision to inspect products and increase the quality of the output. Industry experts predict that within the next 20 to 50 years, machine vision will become a universal factor across all industries, and that almost every product produced will be inspected by a machine vision system.

There are four main reasons for using machine vision:

Anatomy of a Machine Vision System

Once an engineer determines the need for machine vision, the next challenge is to assemble the building blocks of the system. Among the essential components: lighting, part positioning, cameras, optics and control logic, as well as image acquisition hardware, processing software, and engineering services.

Chances are most vendors are not expert in all of these areas. So look for a supplier who not only knows its core competencies but has developed long-standing partnerships with other suppliers to provide other components needed for the system.

A good example of a company that has successfully incorporated machine vision inspection into its products is Asymtek (Carlsbad, CA). The firm has won many awards for the design of automated fluid dispensing systems used in the assembly and packaging of semiconductors, printed circuit boards, flat panel displays, electronic components, and medical/biotech products.

Two important applications for Asymtek's machine are underfill, which pulls the heat away from a semiconductor chip placed on a board, and encapsulation, a process that covers a chip and keeps the air off of it leading to less corrosion and greater structural integrity.

In these applications, the system must guide the fluid dispensing head to a precise target. Imagine guiding this process "blind"—that is, without seeing the surface that the fluid needs to be placed on, an almost impossible task. Machine vision makes this possible by letting the dispensing head "see" where the head needs to go, and then moving it to the exact position to dispense the fluid. The motion control setup, which positions the board on the conveyor, goes hand-in-hand with the machine vision system to make inspection as accurate and repeatable as possible.

Asymtek boosted the power and usability of its dispensing equipment by adopting a machine vision system as a standard feature on its Century, Millennium, and new Axiom 1010 dispensing platforms. The vision engine, produced by Coreco Imaging, provides an enhanced pattern recognition system. Among the key components: High-speed CCD cameras, Coreco's PCVision image processors, and Coreco's Sapera software to automatically correct work piece misalignment. For lighting, the system features an LED array with computer brightness control.

Together with Asymtek's software, the vision engine can take a series of measurements to find die edges and calculate the corner of the die. The vision engine provides several enhancements to Asymtek's pattern recognition systems as demonstrated by an increase in speed of up to 30 percent for detection of standard reference marks. Customers experience greater throughput and yield as a result of the faster image processing and more robust image analysis. These results earned Asymtek Intel's prestigious Supplier Continuous Quality Improvement (SCQI) award, its highest honor for its suppliers.