Professor Wood Chiang and his students have created an analog-to-digital converter with record-breaking power efficiency, and their breakthrough is featured in the IEEE Journal of Solid-State Circuits.
The Ministry of Science in Taiwan and a consortium of technology companies sponsored the team - including Chiang and research students Eric Swindlehurst, Hunter Jensen, Alexander Petrie, and Yixin Song - and collaborators at National Chiao Tung University and the University of California, Los Angeles. The BYU team focused specifically on creating an improved ADC.
“ADCs are literally everywhere and in every electronic piece of equipment because they are the interface between the analog world that we live in and the digital world where information is transferred,” said Professor Aaron Hawkins, Chair of the Electrical and Computer Engineering Department. He said that a power-efficient chip matters in phone and mobile applications where battery life is obvious.
Chiang’s team created an ADC that consumes only 21 milli-Watts of power at 10 GHz for ultra wideband wireless communications. Other converters at comparable speeds consume hundreds of mW or even Watts of power, so not only does their converter have the highest power-efficiency in the world, it holds the record by a substantial margin once resolution is also factored in.
“It’s like the equivalent of creating the world’s most fuel-efficient vehicle,” Chiang said. “Our team is able to create one that’s better than everything else.”
The ADC works on the principle of the binary search where each step in the algorithm reduces the search range by half, according to Swindlehurst. He explained that the full circuit interleaves eight channels, each running at 1.25 GHz, to achieve the desired 10 GHz sampling rate.
Chiang said Swindlehurst is the principal author of this work, which led to the successful defense of his Ph.D. degree.
The project overall was a high-risk undertaking that required meticulous planning and execution, according to Chiang. The team spent four years working on the project - three years to design the chip and one year to test it. A single mistake in the design would have taken an additional year to correct.
“We were on a lot of time constraints to meet tapeout deadlines,” Swindlehurst said. Tapeouts were only a couple times a year with a three-month turnaround time, so he said the team had a limited amount of time to test, debug, implement new features, and layout the circuit before the next tapeout.
“It’s like building a little city,” Chiang said. “There are so many details that went into this project.” He said the chip had thousands of connections and if even one was reversed, it could destroy the entire thing.
But, Swindlehurst said, despite the hard work and long hours, it was something he enjoyed. It didn’t feel like work at all.
They took careful steps and ran simulations before the official tapeout - the one that would take a year to correct if anything went wrong. But even simulations fail to capture everything that could happen. Chiang said a lot of human factors could interfere.
Fortunately, the students had perfected the design. Chiang remembers the day they were finally able to prove the chip worked. “We were ecstatic,” he said. “We were jumping up and down in the lab, shaking hands, high-fiving each other.”
Swindlehurst said this was the most rewarding part of the whole experience. “When those risks paid off, it felt really good,” he said.
They first published their findings in a conference a year ago, and recently their findings were published in the most prestigious journal in the circuit design field: the IEEE Journal of Solid-State Circuits. The journal is extremely selective, and the most recent BYU paper it published was authored by Professor David Comer in 1996.
A lot of publicity likely waits on the horizon. “Now BYU is officially known in the circuit design community at large,” Chiang said. He said he will probably be traveling and speaking to spread the ideas from this project, and his students will see doors and job opportunities opening up.
Chiang plans to use ideas from the project for his other research with circuits to seek the next power-efficiency world record.
“The design techniques we used to minimize power and maximize speed are likely to be used in next generation communication circuits,” Swindlehurst said.