Low power design

Researchers at MIT and Texas Instruments have developed a low power technique for making integrated circuits that are up to ten times more energy-efficient than those on the market today.

The team recently demonstrated their methodology by developing a low power version of TI's MSP430, a widely used microcontroller.

The key to the design was to find ways of making the circuits on the chip work at a voltage level much lower than usual, said MIT Prof Anantha Chandrakasan. While most current chips operate at around 1V, the new design works at 0.3V.

Reducing the operating voltage, however, is not as simple as it might sound, because existing ICs have been optimised for many years to operate at the higher standard voltage level. 'Memory and logic circuits have to be redesigned to operate at very low power supply voltages,' Chandrakasan said.

One key to the new design, he said, was to build a DC-to-DC converter - which reduces the voltage to the lower level - right onto the same chip, which is more efficient than having the converter as a separate component. The redesigned memory and logic, along with the DC-to-DC converter, are all integrated to realise a complete system-on-a-chip solution.

One of the biggest problems the team had to overcome was the variability that occurs in typical chip manufacturing. At lower voltage levels, variations and imperfections in the silicon chip become more problematic. 'Designing the chip to minimise its vulnerability to such variations is a big part of our strategy,' Chandrakasan said.

So far the new chip is a proof of concept. Commercial applications could become available in five years in a number of exciting areas, Chandrakasan said.

In some applications, such as implantable medical devices, the goal is to make the power requirements so low that they could be powered by 'ambient energy,' Chandrakasan said - using the body's own heat or movement to provide all the needed power. In addition, the technology could be suitable for body area networks or wirelessly-enabled body sensor networks.

The research was funded in part by a grant from the US Defense Advanced Research Projects Agency.