Consider time to market early in test process

-May 14, 2013

Early R&D design test activities are traditionally more focused on bench lab characterization than on ATE (automatic test equipment). There is a natural reason for that. R&D test is often about fitting measurements rather than sorting chips. The first goal is to make sure that something we have never confronted in “real life” actually fits the simulations. As a result, using conventional bench test equipment only (either on package or on wafer, using a prober) for R&D test lab is the obvious course.

But let’s approach the subject by taking more end-market considerations into account. Today’s research centers must increasingly focus on direct industrial partnerships. The time-to-market parameter is perhaps the most critical issue in any upcoming project we have. Taking into account mass-production issues in the first steps of the proof-of-concept can provide a strategic edge. (Not to mention that showing an ATE test environment to your potential partners obviously inspires much more confidence in you, compared to presenting a demo board with loads of cables surrounded by a pile of six different test tools extending to the ceiling!)

R&D test should not only be about proof of concept, but also about proving that as disruptive as our new architectures can be, they still are compliant with an industrial perspective. Therefore, taking industrial test into account in the very first steps of the design flow should be strategic in every R&D design center that wants to target final market customers, and not only IDMs or foundries.

On the other hand, purchasing the same test equipment as large “test houses” without the same volume is very expensive. How can R&D manage to afford such investments? It’s simple: use its strategic position in the industrial landscape, and sell its test knowledge. Insights from R&D test teams can be very valuable for test-equipment providers who must anticipate the upcoming standards roadmap and cover the associated testability issues. If you are one step ahead of what exists in the test market, the “work-around” solutions you have developed definitely have value.

What about manpower? We can’t afford to dedicate one test engineer for each designer. On the other hand, performing debug processes without access to in-house test expertise can be very time consuming.

For the bench environment, each designer has the necessary “lab skills” to perform the measurement. Just helping the designer with board design, packaging and test-environment programing should be enough. And let’s face it: modern test equipment is more user friendly and requires less expertise than it used to. But for the ATE environment, you need one full-time, dedicated test engineer since the equipment is too specific for one to master both design and ATE skills (although having mastered one in each field gives a strategic edge).

So what’s the magic ratio? I’d say one test engineer for every 10 designers. What do you think?

About the Author:
Jerome Prouvee's Profile

Related Articles

Loading comments...

Write a Comment

To comment please Log In