|
|
 |
|  | THE CONSUMMATE TEAM PLAYER | | | LSI Logic's Ronnie Vasishta understands the importance of the gap between design and production. | | | By Erik Sherman | |
|
The days of the independent genius as product designer are over. Given modern time-to-market pressures, user expectations, and the complexities of semiconductor technologies, it's become an industry axiom that leading-edge products require leading-edge cooperation among experts in a variety of disciplines. So why don't more chip companies realize that they need internal design chains?
After all, the gulf between design and production can be vast, with feature and control-flow concerns on one side, quantum physics and manufacturing on the other. Because of the demands of both disciplines, there are only a handful of people in the world with a substantial grasp of each. One of those is Ronnie Vasishta, vice president of ASIC Technology Marketing for LSI Logic Corp., and the semiconductor industry's version of an all-around team player.
"I think the number of people globally [who understand both] is not that many," says Dr. Handel Jones, chairman and CEO of International Business Strategies and an EDC columnist. "You're talking maybe 20 or fewer. He's very familiar with the design environment, the capabilities of third-party tools, and is also very knowledgeable in terms of the process space overall."
Vasishta's interest in the path between design and manufacture is part curiosity, part business necessity. The latter is easy to see, in his view. "You either have a design or a process technology," says Vasishta in a voice still echoing his native England. "But the device coming out and working at the end of the day is very closely tied to that underlying process technology and its features." To make a chip effective, and cost efficient, the two realms must join.
Mind the Gap
The gap between production and manufacturing is an important issue for semiconductor companies, according to Semico Research Corp. senior analyst Joanne Ito, who's met Vasishta at briefings. Fabless semiconductor companies have an even more difficult time, because the gap is complicated by the difficulty of managing communications between companies—the semiconductor vendor and the foundry. "Getting a design into a fab and into manufacturing is a real challenge," says Ito. And it becomes only more important as companies move to smaller geometries and new materials, which complicate design issues with quantum effects.
So when LSI wanted to move to a 0.18-micron process in 1996, Vasishta set up a cross-functional team with people from all areas of design and manufacturing. "We met on a very frequent basis—at least once a week, and more often as needed," he remembers. Subsets of the team, focused on such specific areas as process design interaction, met daily. For the first time, Vasishta began to see people from manufacturing and design actually begin to understand how to communicate effectively with each other. That interaction is one reason he likes life at a full-service semiconductor manufacturer: "If you're just a foundry, EDA company, or IP company, you'll never have the complete picture. We have that daily interaction now where we put the specialists together where we can solve the problem."
 These days, Vasishta's expertise lets him manage a range of critical concerns at LSI. He works with everything from corporate technology strategic direction and the definition and development of LSI Logic's technology products to direct marketing support for customers, sales, vertical market groups, and design centers. Vasishta's journey began following his graduation from Trent University in Notting-ham, U.K., in 1985 with a degree in electrical engineering. Although his education prior to college could just as easily have led to a degree in the arts, his heart was in gadgetry: "One time I tried to wire up my bedroom so I could control everything from my bed: curtains, lamps. It worked to a degree, but I wouldn't say it was production worthy."
While still an undergraduate, he worked for nearby STC Semiconductors designing test systems. On graduation, he took a full-time position, working both in testing and diffusion. He had wanted to move to a new fab plant that STC was going to open next to the original one. "I was intrigued with going to a new fab with new equipment," Vasishta says. But the company suffered heavy layoffs, so he took a job with Motorola in East Kilbride, Scotland working in diffusion. In 1987, LSI bought the plant that STC had originally planned to open, and Vasishta had a chance to return, moving to his now current employer.
He was the fourth employee there, walking into a building shell and becoming responsible for specifying and purchasing the equipment and materials, installing the process technologies, and ramping into production.
"We were a pretty young team to go and do that," Vasishta remembers. "It was a lot of long hours. I remember going in on a Friday morning and leaving on a Monday evening. My bed was bubble wrap under the desk while we were waiting for some [diffusion] experiments to complete."
At that time, most of the actual silicon work was performed by chemists. Industrial engineers directed the manufacturing processes. Electronic engineers designed the chip circuitry. And no one understood what anyone else did.
Design Dependencies
Vasishta began to see just how dependent design was on the specifics of production. "As you buy fabrication equipment, none of it works the way you think it will work," he says. "And in each process, you have hundreds of steps that need to be linked in a certain way. Essentially that gave me the grounding for understanding processing technology as I moved into design and new product development. I ended up with a lot of contacts in applied materials, and a lot of contacts with people on the test side.
 |
|
| "You often have to convince the engineering groups that what they're doing is not what
they ought to do." Ronnie Vasishta, LSI Logic |
|
That cross-barrier understanding and insight became a critical asset as LSI closed the U.K. fab but sold the equipment and process technology in 1991. Vasishta took the role of managing the technology transfer to the semiconductor vendor in Germany. He continued managing other third-party technology transfers for LSI until 1995, when he made a significant career shift and moved into marketing. "Being involved in technology transfer, I realized that a large part of the business was not in the wafer fab, but in communicating technology to third parties and convincing people that we had the right match for their need," he says. "For two years, I essentially worked with European customers enabling the technology and winning some pretty lucrative designs there [for LSI]."
"He does have a real knack for communicating information in very succinct and understandable forms," says Ito.
When LSI moved to a 0.18-micron process at its U.S. facility, Vasishta transferred to the U.S. headquarters in Milpitas, Calif., and began setting up his cross-functional team with people from both design and manufacturing. The company needed someone "who would be able to collate the requirements for the technology and transfer them to the engineering groups so they would have direction in their development," he explains. His new role would span process R&D, circuit design, the EDA design tool chain, and package technology. Vasishta had lessons to teach, and also new ones to learn.
"The thing that I found when I came here to the U.S., was that we have some tremendous engineering skills at LSI Logic," he says. "But you often have to convince the engineering groups that what they're doing is not what they ought to do." One example was with the shift to 0.18-micron technology. Traditionally, the company would develop only one type of transistor for each technology mode. But at the new geometry, it became clear that one transistor would not serve the requirements of all markets. A very fast transistor might serve the speed requirements of a router, but current leakage from the resulting design trade-offs would drain the batteries in handheld devices.
Another trade-off was between library performance and functional density when Vasishta was defining the ASIC's library requirements. "The obvious way that some companies have met these diverse requirements was to develop different process technologies for different applications," Vasishta says. But that is a dangerous road, because it creates the need to support multiple bodies of intellectual property for the different process platforms. At LSI, he instead asked the R&D department to develop separate modules for performance, density, and mixed-signal designs.
Relationship Mangement
Now Vasishta works on relationships with EDA vendors third-party IP integration, and core design methodology. "If you consider a typically vertically integrated company, my responsibility is now deciding what it is that we can build, what we can buy, and how we can integrate those things into the company."
Nevertheless, Vasishta doesn't see moving to new geometries as the real challenge. "The process technology will be there when we need it," he says. "Design complexity is what we need to work on more efficiently." The split between the two sides is still "a huge barrier." But what remains constant is Vasishta's work in constructing bridges to span the chasm.
|
|
ABOUT THE AUTHOR |
|
 |
|
Erik Sherman is a writer and photographer covering business and technology. He writes for such publications as Electronic Business, Newsweek, US News and Continental, the airline's in-flight magazine. Contact him at esherman@designchain.com. |
|
Back to the top
|
|
