A maturing AOI industry moves forward

Not many years ago, vendors of automated-optical-inspection (AOI) equipment expended much of their marketing effort convincing electronics manufacturers to adopt the technique as part of a comprehensive inspection-and-test strategy. Today, the benefits of AOI itself are rarely questioned, but some of the issues that customers have always faced remain: cost of ownership, whether to use 2-D or 3-D inspection, and how to store the captured image data.

Economic pressures challenge the manufacturers of AOI systems to improve the performance of their products and lower the all-important cost per use. One way they can do this is by building flexible systems that can be adapted to multiple projects. As a product matures, its process yield generally rises, so it needs less inspection to achieve the same level of quality. As a result, AOI equipment that was once fully utilized can have spare capacity to apply elsewhere, allowing the customer to allocate the cost of the system to several projects.

Jeff Bishop, product marketing engineer for Agilent Technologies, put it this way: “Return on investment has become more complicated. It has to include some more intangible considerations. For example, if I buy a machine for this particular product or process, can I easily transfer it somewhere else? A more flexible machine may prove less expensive even if the initial outlay appears higher.”

Part of the flexibility issue lies with where to place AOI in the process. It can generate equally important data at various stages: immediately after deposition of solder paste, after component placement and before reflow, and after reflow. But buying equipment for each of those venues is obviously an expensive proposition. According to Bishop, “Generally speaking, companies perform paste inspection and then pre-reflow or paste inspection followed by post-reflow. Not many manufacturers inspect at all three stages. For one thing, managers are reluctant to authorize pre-reflow, objecting to a step that only verifies that the pick-and-place machine is out of whack. Adjust the machine and the extra step becomes unnecessary. In the final analysis, companies often decide where to perform AOI based on results from the x-ray inspection system downstream.”

Systems offering higher throughput also significantly impact inspection cost. “Even if those systems are more expensive,” said Bishop, “customers can easily accept a $10,000 or $15,000 price premium. Cellphones, automotive, and similar applications feature relatively small boards, say, 5x8. If inspection produces a cycle time of perhaps 15 seconds per board, the faster throughput and handling that cuts the effective time to 7 or 8 seconds represents a huge savings.”

Ultimately, Bishop doesn't expect AOI to become much less expensive than it is today. “Prices have already declined enormously, from the neighborhood of $200,000 in 1999 to perhaps $70,000 to $100,000 today,” he said. “There isn't really a lot of room to lower the price any further unless a steep drop in camera prices or other factors lower manufacturing costs. Vendors will pretty much stick to improving existing hardware designs and upgrading the software.

“I see a lot of companies abandoning this end of the business. Initially, starting an AOI company didn't require a lot of capital. Rather than try to make a splash in the overall market, a company merely had to identify a niche, corner it, and push outward from there. Price pressures and increased capability for the same price will likely drive some of the more marginal players away.”

Besides considering the cost of an AOI system, a customer also needs to determine whether 2-D or 3-D inspection works better for a given application. As with x-ray, opinions differ as to which option provides the optimum balance of throughput, cost, and fault resolution. If managers are considering 3-D at all, they may consider using the approach only during ramp-up to full production or while tuning the process to bring it under control, relying on 2-D for more routine situations.

Two-dimensional imaging tends to be faster and less expensive. It can measure the area of solder paste on the pads to determine fairly accurately whether the amount of solder is sufficient, and it can see bridges that would act as shorts on the finished board. On a loaded board, either before or after reflow, 2-D can determine if the components are positioned correctly.

In contrast, 3-D techniques such as triangulation or laser inspection can also determine the height of solder on the pad, offering a more accurate measure of solder volume. The jury is still out, however, on whether you should consider insufficient solder volume a defect. Bishop explained, “The information to date is preliminary as to the correlation between AOI results and the defects detected at x-ray. Some companies have found that if solder falls below a certain percentage of nominal volume, the joint will show up later as a defect. But not everyone shares that experience.

“Solder-paste inspection has become much more common in the past few years,” he continued. “As recently as the mid '90s, companies relied primarily on 2-D and volumetric sampling. Today's 3-D equipment is more effective, easier to program, and less expensive than it used to be, so it is becoming much more common.

“The ultimate driver is quality. At the demand of customers, manufacturers—especially contract manufacturers, but it holds across the board—have to ensure the quality of the products they deliver. This attitude is very common in the US and more recently in Western Europe. Automated techniques have caught on more slowly in Asia, where cost consciousness and the relatively lower cost of labor has skewed the economics a bit.”

As the capability of AOI systems has exploded, so has the amount of data that they produce, and storing and tracking the data have proven increasingly difficult. Consequently, data handling and dispersal become important considerations when selecting a system.

“Storing images and raw measurement data inherently takes up more space than storing a “mere” numerical representation. Bishop commented, “In the past, people kept many gigabytes of images, only a small portion of which were used to create the Pareto charts and other analytical tools. Expect to see inspection companies provide better statistical process control (SPC) and process-feedback software. Right now, people are trying to limit storage to only the relevant information about the product—pictures of components, pictures of boards, pictures of bar codes—primarily for liability purposes. For cellphones and other consumer goods, the storage period can be quite short. Medical, military, and other high-reliability products require keeping the data much longer. Still, there is no reason to store every data point for every system down the line. Decide what you need to store and conform to that limit.”

Bishop explained that some companies distribute the data to make storage easier. “Companies have become smarter about storing data and have therefore reduced its volume,” he said. “They have begun to divide the data rather than keeping everything at all locations to minimize the amount that has to be stored in any one place. Managers hold onto some of it, while some is retained by line workers. Some data is even given to customers, who then become responsible for it.

“If we are going to use the data to improve the process, we have to analyze it in a more timely manner. Empower operators to identify and predict problems in real time and generate reports automatically. Perhaps the source of the problem is not the process but bad components or bare boards. What kind of paste are you using? How fast are the boards going through the paste step? If you keep all the data, there is so much of it out there that analyzing it could take months. Considering the short life of a product generation these days, you may not even be making it anymore.

“It also helps if you use the data at all levels. That is, apply your analysis of the data generated at x-ray to improve AOI. Conversely, if x-ray inspection has missed something, can you find it at the earlier step?

“The point is to make the best use of all the tools at your disposal. X-ray and AOI provide complementary results. X-ray will remain a production tool because of the explosion of solder joints in recent years and the number of them hidden under BGAs and similar packages. On the other hand, x-ray can't read component labels or detect certain polarity defects. You will always need them in combination.”