Guest commentary: It’s time to abandon the gold standard in PCB test
John VanNewkirk, CheckSum -- Test & Measurement World, 7/12/2007 6:57:00 AM
The test engineering community generally sees in-circuit test (ICT) as a “solved problem,” having evaluated and selected in-circuit strategies and testers in the mid-to-late 1990s based on criteria and capabilities needed to diagnose the solder and parts defect classes prevalent at that time. These “big iron” testers that met the needs of the test community ten years ago still constitute an ICT “gold standard” in factories around the world.
Test professionals agree that component technology together with manufacturing process changes over the past five to ten years have shifted the fault spectrum—the range and magnitude of solder, parts, and process defects that ICT needs to identify. Paradoxically, though, even while every criterion that was important to selecting testers ten years ago has changed, many test managers continue to rely on the same ICT gold standard as if nothing has changed. Just one example: once crucial ICT capabilities such as digital vector test are essentially superfluous in light of today’s PPM level digital failure rates.
In fact everything has changed. Today’s electronic assemblies pack in more functionality per cubic centimeter than anyone thought possible just a few years ago—all driven by a massive evolution in parts and process technologies: system on chip, denser packages, and thinner traces. Parts quality is up. So are manufacturing yields. Consumer products of unprecedented reliability are today’s norm. All of this is good news. But it profoundly affects test strategies and test costs.
Companies that pretend that as far as ICT is concerned it’s still 1995 means that modern boards are tested on yesterday’s “standard” testers designed to find defects that now rarely, if ever, occur. The result is inefficient use of resources: needlessly high acquisition and maintenance costs together with complex test fixtures and arcane software processes that needlessly drive up implementation cost. Now that electronics manufacturers must supplement in-circuit test with ancillary test and inspection steps like AOI, AXI, and boundary scan to obtain the necessary coverage, it doesn’t make sense to sink the same amount of resources into ICT as it did a decade ago.
Choosing a tester that keeps pace with evolving technology
What should the test manager who accepts that it’s 2007 and not 1995 actually do?
The first step in improving any process is to evaluate the performance of the current system. How well is your current tester meeting the more complex needs of today’s test strategies?
Second, there is no need to evaluate anything if you are unwilling to not only accept change but to act on it. Test engineers can be notoriously risk averse, tending to feel safe with “big iron.” After all, there’s no upside to letting defects escape to the next stage and “big iron” has been the standard for over a decade. But holding to that now obsolete standard—pretending ICT is a “solved problem” requiring no further examination—while all other variables have changed is poor engineering practice. In a fast-changing world yesterday’s safe choice cannot be today’s best choice.
The test manager open to change asks difficult questions like these:
• Is tester X with vector stimulus rate of 10 MHz meaningfully superior to tester Y with 1-MHz rates when 1-GHz rates are really what’s required?
• Have I dismissed less expensive testers as “toys” because of preconceived notions that they’re architecturally too simple and therefore inadequate for my uniquely complex boards?
Other questions to ask include these:
• Is the test platform adaptable to a wide variety of applications?
• Are multiple test methods available and straightforward to implement?
• Can the test platform be a “big ICT and little functional with some ISP programming thrown in” tester one day and a “little ICT with boundary scan together with a hot bed check” the next day?
• Does the system follow industry standards?
• Can I add-in third-party solutions for specific needs like boundary-scan or am I locked in to using the tester vendor’s (usually more complex and expensive) proprietary method.
Although they’re still important, there’s a lot more to choosing a tester than the specs printed on the back page of the brochure. The old test selection criteria of speed and fault coverage are necessary but not sufficient. A world where advancing technology and shrinking product cycles demands as sharp a focus on test strategy as on tester specs means criteria like platform flexibility, adaptability to new technologies, and ease of use must take their equal place alongside the traditional tester selection metrics. And above all, cost: cost to purchase, cost to apply, and cost to support.
Companies that have gone to the trouble to reevaluate their test equipment options rather than assuming that their “big iron” ICT continues to be the only acceptable standard have generally been surprised at how their selection criteria have changed in the past decade. They’ve discovered that inexpensive test flexibility is more important than leading-edge measurement specs needed only for rare corner cases. Low-cost fixturing and straightforward programming invariably trump elegant digital test capability. ISP programming and boundary-scan need to be easy and inexpensive to implement on the tester.
The upshot is that the majority of “big iron” users who have evaluated alternatives have realized that their “big iron” is no longer the optimum ICT solution and have moved to more flexible, lower cost platforms, saving time and money.
In an environment of accelerating change in component and production technology the “ICT problem” is neither static nor solved. The international monetary system moved away from the gold standard in the 1970s because the world economy demanded greater economic flexibility. In light of today’s fault spectrum and production cycle realities, reevaluating in-circuit test strategy and equipment demands the same bold action. It’s time to abandon the “old” standard, and take a fresh look at what you really need in an in-circuit tester.
John VanNewkirk is president and CEO of CheckSum.
Related Content
Related Content
Sponsored Links
TMW Resource Center
Browse Resources by Type:
- What’s New in NI Multisim 10.1? For Education
National Instruments | White Paper
VIEW NOW
National Instruments | White Paper
VIEW NOW
National Instruments | White Paper
VIEW NOW
View More
