Credence takes final test to big leagues

Electronic News, August 11, 1997

Santa Clara, Calif.--Credence Systems started out 20 years ago writing test programs for Sentry testers, made by Fairchild Test Systems. The company--originally known as Semiconductor Test Solutions, or STS--moved up to making its own automatic test equipment a few years later. Last year, the company posted sales of $239 million. Business is off that pace this year, but Credence is clearly moving into the big leagues of ATE, taking on Advantest, Teradyne and LTX.

The tester manufacturer has a long and acquisitive history. STS bought ASIX Systems in 1989 and then became known as Credence Systems. Credence went on to buy the Semiconductor Test Systems division of Tektronix in 1991, entered the memory test business with the purchase of EPRO in 1995 and just a few months ago bought Test Systems Strategies, Inc. (TSSI) from Summit Design for $5 million. TSSI started out similarly to Credence, writing test programs for Tek testers, and has over the years evolved into a business providing software that helps bridge the gap between design and test.

Credence became a public company in 1993, at the beginning of the biggest boom period in semiconductor equipment history. It won its niche in the ATE business by focusing on lower-cost test technology, implementing custom CMOS circuitry (its stock trading symbol is "CMOS") while most other ATE suppliers were still using more-expensive ECL devices. Several other test companies have since come around to the idea of using CMOS.

Wilmer (Bill) Bottoms is the chairman and CEO of Credence. The one-time president of the Semiconductor Equipment group at Varian Associates went into the venture capital business in 1985 and joined the board of Credence. After Credence's founder, David Mees, was killed in a 1991 plane crash, Dr. Bottoms became chairman of the company in 1992, and he took over the CEO's post in July 1996. He met recently with Electronic News' editors in Silicon Valley for a breakfast interview.

EN: We hear a lot in the equipment business that back end or test doesn't get as much attention as it should. There's a history of customers not wanting to spend a lot on test. They don't see it as a value-added proposition. How do you do business coming from an underdog-type position, as opposed to the other big equipment makers?

Dr. Bottoms: "The degree to which testing is performed on integrated circuits is not a decision made just by the manufacturer of those integrated circuits, it's a decision made by the customers of those manufacturers. And in the early days of the semiconductor equipment business, when ATE was new, the testing was actually done at more than one location. There was a test done by the manufacturer before they sent it out, and then the people that received it; if those quality levels were not reliable, they did the test again when they received it to ensure that it would meet their requirements. Over time, in order to make the total system more efficient, the people who are the customers of the IC producers have stopped doing the test themselves, but they have demanded higher and higher quality tests by their vendors.

"So the degree of test which you get is something that's determined, not so much by the IC manufacturer, but what their customer wants to pay for it. For example, if you get a little IC that goes into a greeting card or an IC that goes into a throwaway calculator, it may not have been tested at all. They may just symbol it into the calculator and if it doesn't work, they throw the calculator away, because the yield is so high that the cost of testing doesn't add any real value, or enough value, in the eyes of the person buying those ICs. But if you're buying something that's in any way critical or something that's complex, the probability that it doesn't require testing gets smaller. And I think we all know that if you look at the number of transistors shipped, a higher and higher percentage of those transistors are shipped in complex systems, and the customers, more and more, are requiring complete testing of those systems.

"And more recently they have been requiring testing at speed. It turns out you can test an IC that's designed to operate at 200 megahertz at 50 megahertz. You can test an IC designed to operate at 50 megahertz at 10 megahertz, and you know that it is functionally operational, but you don't know what its speed limitations are. There are applications where the speed limitation is not critical and people have said that's okay. But as the bus speed has improved and the central processor speed has improved in PCs and similarly in other consumer products because one moves, everything kind of moves along with it, there's been a requirement for test at speed, because these things have to work together with each other, and if they're untested at speed, they may not work.

"And so what's happening now is people who are actually buying the ICs and building the systems are requiring a higher level of test, not a lower level of test as they go to more complex systems. I think it is true that people plan much farther in advance for their front-end investments than they do for their back-end investments. And I know that people spend a lot more time and energy, typically, on the front end than on the back end, but I believe that the customers will continue to demand higher and higher levels of quality, which can only be assured if the device is tested."