Space avionics systems require power balanced performance in rad-hard processors
By John McHale
The methods for making radiation-hardened processors and integrated circuits (ICs) have not changed much, but the chips themselves are getting smaller, faster, and generating more heat than ever before in avionics systems in space applications.
"When we focus on processors it is on the processor suite" and its elements, says Vic Scuderi, manager of satellite electronics for BAE Systems. Right now Department of Defense (DOD) users are focused on horse power, -- in other words MIPS or million instructions per second, he adds.
The RAD750 processor from BAE Systems in Manassas, Va., clocks in with as much as 260 MIPS, Scuderi says. These high speeds also create more thermal and power challenges, he adds.
"In space it is difficult to cool processors and supply power," says Keith Nootbaar, senior director for microelectronics and precision sensors at Honeywell Aerospace in Plymouth, Minn. Designers are "asking not about performance, but about the power and performance ratio."
Users can manage the MIPS rate through software depending on the power requirements of their applications, Scuderi says. "It allows them to dial down the MIPS."
Honeywell also enables designers to manage their clock rate, Nootbaar says. Each of their RHPPC 603e processors has three different clock settings, he adds. The 603e is based on PowerPC technology from Freescale Semiconductor and "from a total dose standpoint its immunity is greater than 1 megarad," Nootbaar notes.
Space designers demand that type of functionality Scuderi says. It is part of a low power strategy for electronics, he adds. Later this year a new version of the RAD750 will range from 400 to 500 MIPS and be hardened to one megarad, Scuderi continues.
Both BAE Systems and Honeywell do radiation-hardening by process, which provides much greater radiation protection than the hardening by design methods that commercial foundries use, Nootbaar says. Rad-hard by process results in zero latchups, he adds.
"We use silicon on insulator," which provides 10 times the single event upset (SEU) performance of bulk silicon, Nootbaar says. Also in lower earth orbit there is a need for heavy proton immunity, and ICs from commercial foundries are more susceptible to proton upsets, he adds.
Another reason the DOD community is moving away from commercial foundries is that many are off-shore and this is seen as security risk, Nootbaar says. "There is a concern that with offshore production there could be "intrusions into the circuitry," he explains.
The DOD created the Trusted Foundry program to deal potential risks – such as design tampering – that may occur when off-shore facilities are used.
Looking toward the future Scuderi says that feature sizes of ICs will continue to get smaller. Currently "we are at .15 nanometers," he adds.
The smaller sizes bring benefits but new "challenges because as feature sizes shrink, SEUs, heavy ions, and solar flares" will have a much bigger effect on the chips. Scuderi continues. He declined to go into specifics on how they will attack that challenge.
As for five or 10 years down the road, Honeywell's Nootbaar says he sees carbon nanotubes as the next technological game changer.
The molecular design would create ICs that are high-speed, low-power, and inherently rad-hard, Nootbaar says. "Radiation does not affect the molecular structure," he adds.
Nootbaar predicts that there will be a carbon nanotube memory device in about five years and processors in about 10.