Even if you pay attention to the CPU industry, Atmel isn’t likely to be a company you’re familiar with. But its low-power processors could change the way we interact with devices and the burgeoning Internet of Things. Founded in 1984, the company focuses on embedded computing, microcontrollers, and automotive processors — precisely the kind of hardware that powers the equipment we interact with on a daily basis, without ever realizing it contains a microprocessor or three. Atmel is making waves at present for its new Smart SAM L21 family of processors, which draw so little power they can reportedly run for decades and be powered by energy harvested from body motion.
First, the basics: The L21 family is based on ARM’s Cortex-M0+ microprocessor series. The M0+ is an embedded chip and a fairly modest one — it’s an optimized version of the Cortex-M0, with one fewer pipeline stages to reduce power consumption and a few features of the more capable Cortex-M3 and M4 families.
What sets the Atmel SAM L21 family apart is that they’ve been designed to use ridiculously low amounts of power — just 35 microamps per MHz when active, and 200 nanoamps of electricity when in sleep mode. With power consumption that low, an Atmel L21 core that didn’t wake up very often could conceivably run for decades off a battery. Even more interestingly, Atmel claims the microcontroller can be powered simply by human energy capture.
“Atmel is committed to providing the industry’s lowest power technologies for the rapidly growing IoT market and beyond for battery-powered devices,” said Reza Kazerounian, senior vice president and general manager for the company’s microcontroller business unit. “Developers for IoT edge nodes are no longer just interested in expanding the life of a battery to one year, but are looking for technologies that will increase the life of a battery to a decade or longer. Doing just that, the new 32-bit MCU platform in the Atmel | SMART family integrating our proprietary picoPower technologies are the perfect MCUs for IoT edge nodes.”
Atmel isn’t revealing which process technology its L21 core uses, possibly because these types of processors tend to be built on older nodes and focus on minimum cost rather than top-notch performance. Instead of relying on a cutting-edge 14nm or 16nm process, the company has emphasized sophisticated power gating methods that aren’t much different from what we’ve seen companies like Intel and AMD adopt. Each area of the chip is designed to be power gated, and the core aggressively shuts off segments of the die that aren’t in use.
In larger chips, we’ve seen this approach adopted to avoid blowing power budgets and ensure that mobile battery life is maximized when the CPU is doing relatively simple tasks. The Cortex-M0+ isn’t powerful enough to run even a device like a smartwatch today. But the fact that Atmel adopted such sophisticated power gating methods shows how technologies adopted to preserve battery life at the high end of the market trickle down into much cheaper, simpler parts.
The ability to charge electronics via human power is an old dream, and partly limited by battery technology as much as by circuit design. Simply advancing microcontroller design won’t solve all those problems, but it does simplify one key technological challenge.
No comments:
Post a Comment