The distinction between vision systems used for inspection and those used for real-time control is blurring fast. All but the most basic vision systems are now equipped with digital interfaces that allow easy connection to the machine control system.
The control system uses vision system data to make decisions, either automatically or with operator intervention. These decisions often result in adjustments to the process that correct variations before they can cause rejects.
A number of trends are converging to increase vision system performance requirements in both inspection and real-time control applications.
Producing products in ever-smaller lots means manufacturers must be able to quickly and automatically tweak vision systems to effectively measure different product types. More stringent quality requirements continue to force manufacturers to replace manual inspection with automated vision systems. And high-speed manufacturing and real-time control often tax the processing speed of vision systems.
Figure 1: Sub-Micron Alignments
A vision system determines edge detection and part position for the alignment step in assembly of optoelectronic devices. These applications typically require sub-micron positioning accuracy. (Source: National Instruments)
There is also a definite trend toward moving vision systems further upstream in the manufacturing process. The intent is to identify problems as soon as they occur instead of at final inspection. This requires inspection of discrete parts as opposed to final assemblies, so vision systems for these applications must be inexpensive and easy to install.
Machine builders use advanced vision systems for many different reasons, but the overall objective is consistent. "At a high-level view, the goal of the vision system is to capture and process information," says Robert Pearcey, director of electrical engineering for Cookson Electronics Equipment (http://www.cooksonelectronics.com), Franklin, Mass. "The information is then converted into data that has value for the process." Cookson manufactures electronic assembly equipment for the SMT and semiconductor packaging market.
Let's Get Real Time
When vision systems are used as inputs to real-time control systems, tremendous gains in speed and precision can be realized. Palomar Technologies (http://www.palomartechnologies.com), Vista, Calif., upgraded a microscope targeting system from manual to fully automatic operation using machine vision. Palomar manufactures precision assembly systems including die bonders and wire bonders for customers in the wireless, photonics, defense, and hybrid semiconductor markets.
The company uses Cognex vision systems to enable rapid part location using flexible vision alignment algorithms. Palomar's customers choose from multiple automatic alignment options depending on the orientation and lithography patterns of their components. These can include area alignment, single point alignment, and pad-to-bump alignment.
One of the most important design decisions for machine builders using vision for real-time control is speed versus flexibility. "Opening up the process window to enable multiple part orientations or recognition of same-function parts from different vendors enhances program flexibility and reduces rejected parts, but may add costly time delays," says Eddie Wills, product manger for component assembly systems at Palomar. "Our integrated control and vision systems provide our customers with the ability to optimize their process window decision."
Stress Engineering Services (http://www.stress.com) also used a machine vision system to replace a manual system that relied on microscope targeting. Its precision integrated robot (Figure 1) uses a National Instruments vision system to perform edge detection and part position determination during the alignment process used in the assembly of optoelectronic devices. These applications typically require sub-micron positioning accuracy.
Automated vision systems can also be used to great effect when replacing systems that rely on hard tooling and fixturing. "Both manual and fixturing methods are costly in a high-mix or high-volume production environment," says Pearcey. "In some cases, it is not possible to develop the machinery or the process without the use of vision imaging and processing."
Vision systems on Cookson equipment are used to close the loop on positional feedback for alignment, inspection of deposited materials, surface coverage, and height measurements (Figure 2). The inspection of solder paste deposits on printed circuit boards is especially critical to reliability of the stencil printing process. Solder brick deposit images are acquired by the vision system optics and camera and translated into data for processing by the vision algorithms.
Figure 2: Positional Feedback Precision
Vision systems close the loop on positional feedback for alignment, inspection of deposited materials, surface coverage, and height measurements of solder paste deposits on printed circuit boards. (Source: Cookson Electronics)
There are areas where machine builders and their customers must pay particular attention when implementing a vision system. "It is best to consider the vision application as the equipment is being designed," stresses Pearcey. "The component to be inspected needs to be held within the focal range of the vision system, and process speed may affect the vision capture timing."