Talking to Engineering Hall of Famer Lewis Counts on the eve of his retirement after 38 years at Analog Devices, I found “Mr. Analog” (as he is known by many at ADI) wasn’t waxing nostalgic. After all, the key to Counts’ successes at ADI was always an “out with the old and in with the new" attitude—a belief that engineering means there is always a better way.
Counts was chosen for our Engineering Hall of Fame because of his role in revolutionizing mixed-signal design via the complementary bipolar process, increasing bandwidth and data rates while lowering power consumption. He was also a pioneer in bringing higher levels of complexity to chip design via techniques such as laser wafer trimming.
Counts recalls that in his early collaborations with Barry Gilbert on rms-to-dc converters, a central tenet was that “the way they had always been done was, in my view, quite possibly based on bad reasons and ought to change.”
Counts led the then-radical push to use lasers to “trim and tweak” film resistors and to create ICs that were “complete and useful.” He began a quest to find and solve systems problems, eschewing the single-minded pursuit of “commodity” ICs at the lowest cost. He calls this the East Coast or “Philbrickian” approach, contrasted with “the Silicon Valley philosophy” that focused solely on part cost, or as he says, “the tyranny of yield.”
This continuous-improvement design philosophy got Counts involved in considering how the manufacturing process could be modified to boost the functionality of the end product, eventually leading to ADI’s use of complementary bipolar and silicon-on-insulator technology and more recently to the company’s move to SiGe processes for advanced RF applications. The underlying goal, says Counts, is to drive the process so as to get “more bandwidth per milliamp per buck.” In the mid-1980s, this push to high performance with low distortion opened the pathway to greater circuit integration.
Thus Counts became a pioneer on the path toward on-chip systems: setting the course for today’s complete circuits designed to address a particular problem and offer a solution. Counts credits Paul Brokaw as an important mentor who set the tone for ADI’s willingness to take risks by going in new directions.
“The end objective is radical improvements in technology that ultimately benefit the designer and the end user—getting design done quicker even as complexity goes up exponentially,” says Counts. “The focus today in the era of the cell phone is on networking and hardware integration, but requires continued improvements on the process side and analog side, plus the move to further integration.”
Counts says that while the roadmap for CMOS and digital technology is well publicized, process improvements for analog technology must also continue to scale, to pack in ever more performance. CMOS is “not great off chip” he says, and offers one-third the speed per volt of bipolar technology. The scaling on the digital side puts pressure on the analog side to improve physical interface and dynamic range, and to solve those challenges through integration, effectively dropping the power and cost per channel.
Increasingly, Counts says, the job of analog suppliers is to “modularize” content. To that end, the IC designer must interface with the end market and do more systems engineering. As an example of state-of-the-art integration, he cites a multichannel mixed-signal design integrating 128 channels on one chip, a solution designed for higher quality medical imaging that also helps limit patient exposure time to the scanning devices.
Counts is looking forward to a new role following retirement: he will focus on being ADI’s emissary to universities, with a particular focus on interdisciplinary engineering challenges. Counts points to microsensors, biomedical, and MEMS as areas of particular interest. He says he wants to influence what is being taught at the universities, because not surprisingly, he thinks its time for academia to consider new ways of doing things. Considering his track record in the world of analog design, campuses should get ready for Count’s version of Engineering 2.0.
Reprints
