The rising stock of test is evident in a recent advertisement from a leading IC manufacturer, which broke new ground by publicly citing test as an invaluable part of its manufacturing process. The company revealed that it tests its advanced devices 1015 times!
This virtually incomprehensible amount of testing encompasses many testing steps throughout the process: at the beginning (design validation), at first silicon (silicon debug), during manufacturing (process testing of wafer characteristics and wafer-level testing of devices prior to packaging), and finally following packaging, when the packages themselves are subjected to testing to ensure functionality and performance.
The evolution of complex devices such as systems-on-chip (SOCs) signals another daunting challenge—namely, how SOC testers will deal with the emerging complexity of the disparate technologies of these chips. SOCs contain many pieces such as DRAM, microprocessors, flash memory, and A/D and D/A converters. Test engineers and engineering managers must figure out how to reduce the per-function cost of testing these devices.
As an industry, we must find a way to break new ground and develop clever new test products. We also need to think very aggressively about what’s needed to enable development of the lowest-cost solution for complex chips.
This means taking a close look at the human-resources angle. As device complexity continues to escalate, so too does the need for test engineers with a depth of training and expertise in a variety of areas from which they can draw as they develop new test methodologies.
Engineers who hire on at test companies typically have received their degrees in a broad area such as electrical or mechanical engineering or computer science and then developed their knowledge of test on the job. Consequently, they usually do not enter the industry possessing the cross-discipline understanding necessary to help ensure their success, particularly when it comes to testing increasingly complex chips such as SOCs.
All of this points to the need for a formal test-engineering discipline at the university level. Those who go to work in the test arena are increasingly entering the workforce unprepared to solve real-world problems.
Interdisciplinary test-engineering programs would enable universities to educate students about the key role of test in the semiconductor food chain. The programs also would produce a new generation of engineers specifically trained to understand and address a broad array of test challenges. Such programs would significantly benefit the semiconductor industry by providing test engineers trained to meet the demands of today’s challenging test environment.
Creating a New Discipline
Awareness of the role test plays in helping to ensure the commercial success of end products must start at the university level, which means educating students early on about the pervasiveness and value of test. Today, there’s no such discipline as test engineering per se. There’s no school where you can go to obtain a degree in test engineering. You obtain your E.E. or computer science degree, and then gradually you learn the test discipline.
If universities offered a more focused, interdisciplinary approach, this would help to get students more interested in test and in how to figure out all the ways the various tests interact with each other. By learning to identify where there’s an overlap of information, students would become more aware of how each part of the chip-making process, from design through test, is interconnected and begin looking at ways to streamline and optimize test procedures.
Adding a new generation of experts trained to critically examine and improve test efficiency with an eye toward lowering the cost of test is the true impetus for introducing a test-engineering discipline into universities. Essentially, it requires a shift in thinking for the academic community as well as a concerted effort on the part of the electronics test industry to help formalize and elevate such a discipline.
A cursory, thoroughly unscientific search of the Internet reveals that most schools with electrical and computer-engineering departments offer a course, sometimes two, on the topic of test. Here is a very small sampling, in no particular order, of some of the courses:
- University of Cincinnati: VLSI Testing and Validation
- Auburn University: VLSI Testing
- Rutgers University: VLSI Testing, taught at the Rutgers Center for VLSI Design and Testing
- University of Michigan: Digital System Testing
- University of Washington: VLSI Testing
- Purdue University: VLSI Testing and Verification
- University of Maryland, Baltimore County: VLSI Design Verification and Testing
- University of California, Santa Barbara: VLSI Testing Techniques
- University of Kentucky: Introduction of VLSI Design and Testing
- State University of New York, Stony Brook: Digital System Testing/Advanced Digital System Testing
- University of Massachusetts: VLSI Testing and Diagnosis
Obviously, this does not represent a comprehensive list of university course offerings. But it underscores an interesting point made by Dr. Wojciech Maly of Carnegie Mellon’s Electrical and Computer Engineering Department (see “An Academic’s Perspective”).
Dr. Maly maintains that because testing in and of itself is not necessarily a discipline to which hordes of students are likely to be attracted, the best way to indoctrinate them into this arena is via the world of VLSI design. He teaches a course on this topic and makes a point of incorporating testing technology information and techniques into the course work so that students understand the vital nature of the relationship between design and test.
This is a good start, but it’s just that—a start. Historically test has been knocked because it doesn’t technically produce anything, even though it ensures that manufacturers produce good devices, which saves time and money and prevents bad devices from being implemented in final applications. This has implications far beyond consumer-centric applications like gaming systems and PDAs, extending into industrial and automotive applications, medical systems, and commercial aircraft, to name a few. The higher the chip content and the more critical the application, the more important test becomes.
Formal test-engineering programs not only would provide the necessary curriculum, but also could, perhaps, comprise subdisciplines in which specialization in mission-critical test applications is encouraged. Students need to be taught about the big-picture benefits of test, including:
Higher-Yield Fabrication
You could make the argument that it’s very easy, in theory, to design complex chips, but the error rate will grow exponentially because you have so many things interacting. In turn, your yields are going to be seriously impacted. This offers some exciting opportunities for course work centered on test as an important component of yield assurance.
Quality Assurance
Quality assurance is a huge issue as chips grow ever larger and more complex. The danger of producing flawed chips grows rapidly without the implementation of programs centered on assuring quality through test. This approach has been taken in the world of vehicle manufacturing, and it’s a global challenge for IC test as well. Students need to look not just at test theory, but also to work on future approaches today.
Test for Design
Linking design and final manufacturing is crucial to understanding how testability affects device design. It moves students beyond test theory into a practical understanding of why testability needs to be incorporated into designs and why the supply chain is becoming increasingly interdependent.
How To Go About It?
So, what needs to happen to bring these kinds of programs to life within university engineering departments? What are some solutions that could speed implementation?
The cost issues associated with creating a new discipline are formidable: securing and maintaining equipment and related software, procuring the services of qualified faculty, obtaining university recognition of a new undergraduate curriculum and degree program (not to mention graduate- and post-graduate-level coursework), and attracting students to this new discipline through marketing programs.
Certainly, test companies must be more aggressive and forward-looking in working with universities. It’s within their own best interests to ensure that a plan is in place for the foreseeable future to populate the profession with strong test engineers.
This plan extends far beyond the traditional corporate university programs, which have centered on recruiting, donating test equipment, and creating scholarships or endowments that encourage students to focus their studies in a particular area. These kinds of activities, while valuable, will not suffice if test engineering is to be brought to life as a viable discipline.
One way to accelerate implementation would be through cooperative efforts between universities and test companies. Strategic consortia aimed at identifying and meeting the needs of the universities would create relationships in which the corporate world can find practical ways to fulfill the requirements of academia.
Another solution to help solve the issue of finding the right faculty members would be for companies to work out creative arrangements with qualified engineers and engineering managers, enabling these individuals to teach courses on an interim basis. This would provide universities with immediate access to credible resources as well as facilitate promotion of the new curriculum.
For many years, Advantest has provided graduate students from Stanford University’s Department of Electrical Engineering off-hour access to some of the industry’s most advanced test systems. This has provided hands-on, real-time exposure to the company’s state-of-the art technology, a unique opportunity that many in tomorrow’s workforce often do not have.
Conclusion
The semiconductor test industry stands at the threshold of a significant upswing in terms of both its perceived value and the amount of R&D and resources devoted to new and enhanced test techniques and programs. The question we must ask ourselves is this: Whether test professionals come to this business intentionally or by accident, how much more effective could they be if they were intentionally directed here?
For this to happen, it will require universities to implement programs that get people interested in test and help them realize the extreme value and importance of the role of test. And, concurrently, it will require test and complementary industries to take a hard look at how they’re investing in the future of the industry and decide what they can do to help ensure that test engineering finally comes into its own as a true, viable discipline from which will emerge the skilled test engineers of tomorrow.
About the Author
Nicholas Konidaris is president, CEO, and chairman of Advantest America. Previously, Mr. Konidaris was vice president of Boreas and Hypres, both early start-up companies. His career includes 14 years with Teradyne where he was a product and marketing manager. He holds an M.B.A. from MIT’s Sloan School of Management, serves on the board of directors for Ultratech Stepper, and is a member of TechNet. Advantest America, 3201 Scott Blvd., Santa Clara, CA 95054, 408-988-7700, e-mail: [email protected]
Return to EE Home Page
Published by EE-Evaluation Engineering
All contents © 2003 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
February 2003