Why Revisit a Solved Problem?

To manage test economics effectively, seek the lowest cost solution for each board that enters the test department.

As a newcomer to the circuit-board test industry, Iï¿½ve found itï¿½s exciting to be at the center of new device and packaging technologies that stuff more and more functionality into less and less space. Test managers have a major challenge: developing test strategies that cope with new technology while their managers demand higher test yields delivered with ever-shrinking budgets.

The fallout of the recent downturn has made the job even more challenging because there now are fewer test engineers to do it. Everyoneï¿½s to-do list just keeps growing.

Test engineering centers around managing the tension of meeting complex technology demands while minimizing unnecessary cost. This never-ending optimization process requires both technical and economic savvy on the part of the engineer.

But from this newcomerï¿½s perspective, I see more enthusiastic concentration on technology than on economics. Thatï¿½s natural. After all, overcoming technical challenges as elegantly as possible is what makes engineers become engineers in the first place.

Thereï¿½s no shortage of technical issues facing the test industry: lead-free soldering processes, increasingly dense boards and tiny components, exotic low-voltage semiconductors. But thereï¿½s a costï¿½in all senses of the wordï¿½to focusing too much on the technology and not enough on the economics. In fact, we estimate that in North America alone, OEMs and contract manufacturers spend $200 million more annually on board test than they need to in fixtures, programs, support, and upgrades (Figure 1).

Figure 1. Estimated Annual ICT Operating Costs in North America
Source: CheckSum

Over the past year, Iï¿½ve met with more than 100 test engineers and managers whoï¿½ve patiently explained the subtleties of circuit-board test to me. Two themes stand out:

Test managers at both OEMs and contract manufacturers have standardized, or plan to standardize, on traditional does-everything in-circuit test (ICT) systems for all their board types and consider in-circuit to be a solved problem, freeing them to turn to other pressing issues.
These same test managers know these traditional testers are technical overkill for a sizable majority of their board types, which could be tested less expensively on simpler equipment.

The reasons behind this apparent inconsistency are simple. Changes in the fault spectrum over the past five years have effectively placed boards in two categories: those that require traditional backdrive digital vector testing and those that either donï¿½t need it or whose digital test complexities lie beyond the state of the backdrive art.

But in most cases, test managers have standardized on the traditional in-circuit tester equipped with sophisticated backdrive capability and other advanced features. That way, their test department is ready for any challenge the design group may send its way; they are fully covered by tester insurance and can get back to dealing with the other technical challenges that confront the test department.

Until now, tester insurance has made sense. With continuously evolving technology, new equipment purchases have been easier to justify when a new technological capability was gained.

But insurance comes at a price. When used to test boards that donï¿½t need it, high-capability testers waste resources, not just the tester, but also in the higher cost overhead of more complex test programs, more expensive fixtures, and higher maintenance costs. Unfortunately, too often test engineers have become desensitized to these costs because of internal accounting procedures that make these ongoing costs look free.

But they only look free. One contract manufacturer of high-end boards told me that even for its mix of sophisticated products, 80% received no added benefit from the digital capabilities of traditional ICT. Yet, the contract manufacturer isnï¿½t tracking the added cost of this tester insurance.

Regardless of accounting procedures, cost-reduction pressures only will intensify for OEMs and contract manufacturers. Typical gross margins for a contract manufacturer are stuck in the single digits. Senior managers are awakening to the reality that money frequently is left on the table at board test. The era of test engineers being able to buy high-priced just-in-case-we-need-the-capability-sometime-in-the-future testers is drawing to a close as the economic forces in the organization look in every corner to ferret out cost savings.

As senior managers drive to constrain test budgets, they still will require the test department to deal with increasingly complex technology and meet ever-tighter production schedules. This reality creates a stark choice for test managers: forego new equipment purchases such as optical or X-ray inspection or find a new way to implement existing test requirements at a lower cost.

Reducing total test cost requires re-opening the door to that seemingly solved ICT problem.

The industry-wide habit of putting all boards from the simple to the complex on the same high-capability tester thatï¿½s the designated standard invariably drives up overall costs. And the theoretical savings that a standard is supposed to deliver rarely offset the extra fixturing, programming, and support costs that traditional testers entail.

A high-capability tester is right for a high-complexity board. But putting a simpler board on a high-capability tester is akin to planting a rosebush with a backhoe. The job gets done, but thereï¿½s a lot more overhead cost compared to putting that simpler board on a tester that has lower application and support costs.

Thatï¿½s where tester portfolio management comes in. Just as an investor will distribute funds into a variety of financial vehicles based on clear criteria, the tester portfolio manager distributes board types among testers of differing capabilities based on criteria such as board complexity and production volume. This reduces the total test cost by avoiding unneeded tester overhead.

ï¿½But our goal is to reduce the variety of testers,ï¿½ argues the test manager. ï¿½Besides, time-to-volume pressures donï¿½t allow sufficient time for complicated test-strategy planning. And my staff of test engineers is way too small to train on a bunch of different testers.ï¿½

These all are valid points, but the tester portfolio can be as simple as having a traditional tester (the one you already own) and a simpler, much lower cost in-circuit system. Tester portfolio management can be as easy as deciding to put the next board type in the production queue onto the least expensive tester that can do the job; tracking the costs of operating, fixturing, programming, and support; and calculating the savings.

ICT now is almost 40 years old. In a mature market, itï¿½s essential to focus more heavily on the economics than on the technology. Cost comparison has become the rule because it is increasingly difficult for any in-circuit tester to differentiate on technology. Taking this economic point of view, many test managers have seen standardization as the most sensible route to ICT cost reduction.

But legacy issues make standardization almost impossible to achieve, and it ends up being discussed more frequently than accomplished. This is especially true of large contract manufacturers who have grown through multiple acquisitions. Their tester portfolios resemble an ice-cream parlor with five variations of vanillaï¿½absorbing all the costs of variety but gaining few of the benefits.

Rather than wishing for a single standard, a more practical approach is to seek the lowest cost test solution for each new board that enters the test department, even if it means investing in yet another but much lower cost test system. Given the wide variety of boards built today, I believe high-end testers and very low-cost platforms will coexist on an increasing number of test floors, delivering much lower total testing cost than a single high-end standard.

Embracing a tester portfolio instead of a single standard may seem counterintuitive, and it may look like solving the problem all over again. Thatï¿½s what a major automotive electronics manufacturer thought until its costs and technological requirements were analyzed. Test cost was reduced by more than 60% without any detrimental effect on yields, and savings now are some $1.5 million per year.

To save the rest of the $200 million spent annually on fixtures, programs, support, and upgrades, test engineers must act on answers to the following questions:

Have I thought through the true cost of trying to put every board type, regardless of complexity, onto a single does-everything tester?
Does the fault spectrum of the next board to be tested really require the cost and effort to implement backdrive digital test?
Which has the faster payback: investing in performance upgrades to my conventional in-circuit tester or adding a low-cost in-circuit tester to my portfolio?

Test engineers and managers do not need to become economic gurus to manage test economics effectively. There can be a big payoff. Savings that arise from re-opening the supposedly solved problem of ICT can leave room in a shrinking budget for new tools that will help test engineers deal with all that new technology.

About the Author
John VanNewkirk is president and CEO of CheckSum. Previously, he worked in China for Bain & Company and ran a steel processing business there for VSC Holdings. Mr. VanNewkirk has a B.S. from Brown University and an M.B.A. from Harvard Business School. CheckSum, Box 3279, Arlington, WA 98223, 360-435-5510, e-mail: [email protected]

November 2004