The short (and possibly long) term future for a semiconductor-based energy harvesting market depends far more on lowering power than finding ways to harvest more energy. That was the unanimous view of ISSCC 2013’s opening night panel on the topic.
And the four-man expert panel had one other important takeaway. You have to get your system definition right early on and manage the tradeoffs. With numerous energy sources to choose from and still more options for harvesting them, matching the right mix of technologies and optimizing the key energy-consuming components within them is critical.
Sywert Brongersma, senior principal scientist at the Holst Centre/imec, illustrated the challenge that remains by revealing that the Belgian research institute is no longer investigating on-body harvesting to look at other markets and sources.
He explained that not only is energy generation low – 4uM/cm2 for vibration and 20uM/cm2 for thermal – but that the game is not about providing an alternative to but rather beating batteries. “They [batteries] will never go away.”
Jerald Yoo, assistant professor at Abu Dhabi’s Masdar Institute of Science & Technology, offered the ISSCC audience another view on the same point. “We are in the era of invisible silicon,” he said. “The end-user can barely see what’s inside as the technology becomes more deeply embedded.”
In other words, the end-application is the thing. For the customer, how you get there – battery-powered or otherwise, simply doesn’t matter.
Energy harvesting is becoming more competitive in this respect. Christian Enz, director of the Institute of Microengineering at Switzerland’s EPFL, illustrated how advanced duty cycling technology was being used by schemes such as the EU-backed Wiserban Project. It recently developed a heavily-cycled eye-pressure monitor with power consumption of just 675pW/day.
But all agreed that there is a way to go. “There is still a gap between continuous device consumption and scavenging,” Enz noted.
Tenets of harvesting design
And here is where the need for advanced system-level design analysis comes into play.
Brongersma broke down five harvesting markets which have greater potential.
- Smart buildings
- Smart packages
- Intelligent tyres
- Predictive maintenance
- Body area networks.
Each has its own priorities – some require continuous operation, others do not, for example – and each could also benefit from different energy sources as well as architectures. For example, packaging may work best when driven by RF harvesting, tyres by vibration and medical applications by photovoltaics. Although even these are not always clear-cut choice: a building monitor could use either RF or PV, for example.
However, once that choice is made, the real challenge lies in the processes of optimization and, where possible, cycling.
Reminding the audience that technology trends are towards 30% per year improvements per conversion step for devices whereas PV efficiency gains by about 1% per year, Dennis Sylvester, director of the Michigan Integrated Circuit Laboratory, offered a key set of five tenets of energy harvesting design.
- Select the appropriate technology.
- Identify the use scenario and the energy bottleneck.
- Focus on the always-on circuit blocks.
- Apply aggressive voltage scaling everywhere, even to analog.
- Remember there is more potential in reducing power than increasing the energy harvested.
All four ISSCC speakers said that harvesting-based products applications remain on the right track, while acknowledging the market’s reputation as one of “neverending promise”. Progress, though, will depend on promoting the application over the technology and beating the batteries that have been with us for so long with the right methodology.