These Energy-Saving, Batteryless Chips Could Soon Power The Internet Of Things

In the beginning, it was speed and performance. We loved our electronic gadgets and PCs to be fast and furious, no matter if the energy consumption was over the top. The industry responded accordingly: the circuit design universe focused on speed and performance, so following Moore’s Law, doubling transistor density every 24 months, the goal was always to make things faster and more frequent, and laptops becoming more robust and being able to run bigger programs.

Now, things are starting to change: as we move into a mobile world, first with laptops, and then with phones, and now smartphones and tablets, people are all of a sudden aware of power issues. Desperately seeking a plug to save your device from a sudden death has become a common habit. And it’s only going to get worse: soon everything will be computing something and wirelessly communicating.

The Internet of Things holds a lot of promises, but if this means having to charge dozens of gadgets, from your smartwatch to thermostats, it could easily turn into a nightmare. Not to mention the energy bill at the end of the month. Luckily, it doesn’t necessarily have to be this way. Charlottesville-based company Psikick has been working on a technique called subthreshold processing. That could solve this kind of problems and provide us a battery-less future.

“Subthreshold processing,” the company’s CEO and co-founder Brendan Richardson tells me, “has been theorized and known about since the ’70s. Some of the early digital watches employed subthreshold circuits to conserve power and run on battery power quite efficiently. But it was never widely explored because until very recently power was not the focus of circuit design”.

Image credit: University of Virginia – Dan Addison

How does it work? The underlying concept is that, while transistors in a circuit with a supply voltage below a certain threshold are traditionally considered to be always off, there’s actually still a small current flowing, a leakage that could be exploited to perform useful operations. Enough to power EKG monitoring, wireless sensing applications, vibrations monitoring and other similar activities that need only a few tens of megahertz to run.

After years of testing, the Psikick team was able to design system on chips (SoCs) with the same or greater functionality than any comparable SoC found commercially or in research, but with a power consumption of 100 to 1,000 times lower. That’s great news per se, but where the revolutionary potential of this approach really lies is in the fact that this extreme energy efficiency enables them to be continuously and entirely be powered from harvested energy sources such as vibration, thermal gradients, solar, radio frequency (RF) or piezo actuation. No need for batteries anymore, as those chips will be able to power your monitoring application almost indefinitely. And that’s indeed a revolution.

PsiKick is not the only company operating in this space: there are a few competitors, like Austin, TX based Ambiq Micro, or IMEC eletronic research center, in Belgium.

One thing must be made clear: you won’t be able to replace your cell phone chipsets with subthreshold processors and throw away your smartphone’s charger anytime soon. Or do something like intensive image processing using only the energy collected by Psikick’s chips. “This won’t be possible,” the manager explains, “because you don’t get enough throughput. You’re not going to run these chips at gigahertz speeds. When you step down that subthreshold slope, you get a quadratic savings in energy, But the trade-off is that everything slows down. You’re going to run at a few tens of megahertz at most. That’s plenty to do EKG monitoring or vibration monitoring, but you’re not going to run, say, a high definition video on this chip.”

Even with these limitations, Psikick’s chips are likely to appear pretty soon almost everywhere when it comes to sensing applications.

“We think,” Richardson says, “that the technology is applicable to just about any application in the Internet of Things that you could imagine. Industrial process control, infrastructure monitoring, precision agriculture, biosensing, medical, consumer wearables, on and on. Smart homes, smart cities, smart roadways, whatever you want to think of.”

No wonder the industry is getting excited, with many customers across different verticals lining up to try the new chips. “Right now” – the Ceo says – “we are in the process of defining a platform chip that can be used in multiple applications in a particular vertical. Instead of building an ASIC for each application, we’re now defining what an application standard product would look like. For industrial process control, we can have one chip that can do lots of different things. In consumer wearables, we could have one chip that does a lot of things that are all common to consumer wearables, for instance”

Based on Psikick’s road map, the first chips will be in the marketplace by mid next year.