The FATE of open architectures

Open-architecture ATE got a boost at the International Test Conference (see p. 11) when Advantest announced that it will ship its T2000 ATE systems, which conform to the Semiconductor Test Consortium's OpenStar specification. At the "Future of ATE" (FATE) workshop following the ITC (full disclosure: I served on the workshop's program committee), Intel engineers emphasized the importance to them of standardizing the "precompetitive" features of ATE, such as backplanes and software.

OpenStar proponents contend that standardization will enable test vendors to compete where it counts: in the instruments that plug into open ATE systems. Concomitant with this point is the implication that competition among backplanes and software doesn't count. Advantest America president and CEO Nick Konidaris told me that architecture was decided 20 years ago; the legacy is multiple, technologically similar but incompatible implementations where innovation is no longer taking place.

But Neil Kelly, VP and chief technology officer at LTX, rejects Konidaris's premise. Comments I've heard from test engineers suggest that factors like instrument-to-instrument synchronization among multiple clock domains can be more important than the performance of individual instruments. And even if Konidaris's premise is true, Kelly told me, freezing backplane and software standards would make the "no innovation" prediction a self-fulfilling prophecy. Backplane and software innovations, he said, are what make ATE different from rack-and-stack instrument collections.

Kelly also took issue with the "precompetitive" concept. "Precompetitive" suggests a group effort to achieve a level, standardized playing field from which true innovation and competition can commence. Some FATE participants cited as a model the Sematech consortium established in 1987 to solve semiconductor-manufacturing problems. But Sematech set out to achieve technological goals that its individual US members hadn't already accomplished on their own, such as the 1990 manufacture of 0.5-micron wafers using all American-made equipment. The OpenStar effort essentially asks companies to go through a "decompetitive" phase, where they roll back to a lowest-common-denominator architecture onto which they can build a standardized but not technologically superior version of what they are offering today.

OpenStar is a worthy experiment. I wish its participants every success as OpenStar competes with the slightly less open architectures of modular systems such as NPTest's Sapphire and Teradyne's Integra Flex. But compete it will have to do with respect to performance as well as price.

Contact Rick Nelson at rnelson@tmworld.com.