Caltech Engineers Test Self-Repairing Processor Prototype
An integrated circuit that heals itself when damaged is being tested by researchers in California
A team of engineers from the California Institute of Technology (Caltech) has demonstrated a prototype of an integrated circuit that can automatically diagnose and fix many issues that would have previously resulted in total failure.
The power amplifiers for millimetre-wave frequencies created at the Division of Engineering and Applied Science stayed operational after being subjected to severe stress and even being shot by lasers. The findings were published in the March issue of IEEE Transactions on Microwave Theory and Techniques.
It’s alive!
The new chips are not just more reliable, they also consume up to 50 percent less power. In the future, this technology could be implemented in almost any electronic system including computers, home electronics and mobile devices.
Today, even the smallest physical flaw or sudden spike in voltage can render a chip completely useless. Physical “wear and tear” also applies to chips, as they age, and repeated use changes the internal properties of the system.
To make integrated electronic circuits more reliable, the Caltech project funded by the US Defence Advanced Research Projects Agency (DARPA) and the Air Force Research Laboratory worked on self-healing.
The smart component can diagnose almost any problem using a network of on-chip sensors that monitor temperature, voltage, current and power. The information is fed to an Application-Specific Integrated-Circuit (ASIC) that can analyse the readings. If the chip’s performance is impaired, it can develop a work-around in less than a second, by adjusting the actuators.
“It was incredible the first time the system kicked in and healed itself. It felt like we were witnessing the next step in the evolution of integrated circuits,” professor of electrical engineering Ali Hajimiri told Phys.org. “We had literally just blasted half the amplifier and vaporised many of its components, such as transistors, and it was able to recover to nearly its ideal performance.”
The ASIC doesn’t need to store algorithms for every possible fault: it has been programmed to adapt to circumstances and keep hundreds of thousands of transistors in their optimum state, according to the sensor readings.
This first appeared on TechWeekEurope UK. Read the whole story here.