Military Embedded Systems

Rad-hard requirements increase as space market slows


June 13, 2013

John McHale

Editorial Director

Military Embedded Systems

Radiation immunity and signal processing requirements are increasing for spacecraft and satellite electronic systems as designers look to add more capability for government and civilian applications. Meanwhile, the U.S. military and commercial markets for space electronics remain flat, thanks to the slow global economy and government budget cuts.

Despite what has been a stagnant economy, military and consumer demand for increased bandwidth from their intelligence and communication systems has not slowed a bit. Regardless of budget cuts and program cancellations, system integrators want space systems with high-performance signal processing payloads that can survive in a variety of high-radiation environments for longer periods of time than ever before.

Space designers go where space designers have gone before. They need to trust where they get their radiation-hardened (rad-hard) electronic components from because swapping out bad parts in a satellite or spacecraft that has already launched is not an option. Stricter radiation requirements make proven flight heritage even more critical, making the barrier to entry in the space market formidable if not impossible for suppliers whose widgets have yet to leave the Earth.

“You can’t cut costs in space, especially when you’re sending astronauts to the moon or Mars,” says Doug Patterson, VP, Military & Aerospace Business Sector at Aitech in Chatsworth, CA. “They don’t have enough fuel to go three quarters of the way to Mars and have something go wrong. If it does, they can’t get back as they need to loop around the back of the planet for a slingshot effect to get them started on their way home. Our space customers are pushing technology higher and higher, further and further, and for longer periods of time in space. As a result, component selection is becoming much more rigorous. Customers that would typically request 20 kilorad (krad) products now want a 100 krad pedigree. This also makes the development process even longer as radiation life testing needs to be done. Just running the test itself takes time and, on top of that, you have to make sure you have the right components, which is also a time-consuming process, although design cycles are coming down as the components produced with silicon on sapphire are more inherently rad-hard than other CMOS technologies.”

“I think the current DoD [Department of Defense] budget cuts and sequestration are making integrators more cautious, but I think the need for radiation-hardened components is greater than ever before,” says Jim Kemelring, CTO of Triad Semiconductor in Winston-Salem, NC. “The number of satellites being launched into space is just amazing. Rockets are going up and launching not one, but multiple satellites at a time and all the electronic components onboard each one need to be rad-hard. Missiles and avionics are also demanding more rad-hard parts. Even designers of medical X-ray technology are looking at using rad-hard components, thinking maybe their equipment would last longer with [that] type of rad-hard component.”

“As you go up in radiation tolerance, the challenge is maintaining the same or similar performance levels,” says Monty Pyle, VP of Sales & Marketing at VPT in Bothell, WA. “For radiation-hardened components it is about the components you select, addressing their potentially differing footprints and/or electrical characteristics. It is not just a simple matter of swapping out components. We also have stringent TOR – Technical Operating Report – requirements that are becoming more and more fixed and demanding every year. Aerospace TOR refers to reports developed by Aerospace Corp. and flowed down as requirements on space asset procurements that cover technical requirements on electronic parts, mechanical parts, materials, and processes involved in the manufacture of components used on space-based systems. The requirements include guidance on analysis, component deratings, prohibited part types, part element evaluation, and screening that in many instances exceeds the requirements of MIL-PRF-38534 Class K.”

“Qualifying and testing ASICs takes longer every process technology generation, it seems, as the complexity of the customer designs onboard has only grown,” says Peter Milliken, Director of Semi-Custom Products at Aeroflex Colorado Springs. “Integrators are dreaming lots of wonderful things in software that can be quite complicated to manifest in silicon – especially when there are more than 10-20 million gates to play with in a component. The System-on-Chip (SoC) design flow has become more and more demanding. Customers use FPGAs to guarantee operation before they fly, so they can program FPGAs until you get it right. A good ASIC design flow makes sure the design and quality are right the first time. Right now, 90 nanometer is the technology of the day and we are on track under a government program to have our UT90nHBD ASIC library, design flow, and manufacturing offering fully 90 nm qualified by the end of the calendar year. For more complex flip-chip performance applications, we are executing in accordance with our Aerospace Corp. reviewed qualification plan with a certification expected in the middle of calendar year 2014. The toolkit has already been released to the community and can be downloaded from our website.”

“As far as rad-hard designs, we see a separation in markets in commercial and military rad-hard,” says Al Ortega, Marketing Manager, Military and Space Products, Microsemi High Rel Group in Lawrence, MA. “In military programs there are more requirements for higher radiation tolerance. We are seeing a lot of requirements out of Aerospace Corp. for additional testing of Single Event Upset (SEU) prompt dose and neutron radiation. The military segment wants more reliability and the Aerospace Corp. is constantly involved in increasing requirements in terms of reliability – wanting Enhanced Low Dose Rate Sensitivity (ELDRS) and SEU immunity at a minimum.”

Increased requirements for SEU immunity and reconfigurability also drove the AFRL to fund Xilinx’s Virtex-5QV FPGA. Many FPGAs for extreme rad-hard applications have been one-time programmable, says John Bendekovic, Director of Aerospace and Defense Sales at Xilinx in San Jose, CA. “Once a satellite is launched, whatever it took up there would have to stay up there for its mission life without any way to reconfigure it from the ground. What Xilinx has done is develop a reconfigurable FPGA that is SEU immune and latchup immune. There is a demand for reconfigurable computing in the upper rad-hard environments as complex, processing-intensive payloads are enabled by reconfigurable logic. If integrators can get some factor of reconfigurability in the system, they can access it from the ground and reconfigure it based on mission requirements. ASICs are not able to provide that feature. This is not the first Xilinx brought to market with reconfigurability, but its SEU immunity is the watershed that sets it apart from other FPGAs for space.” The device has an SEU immunity of > 100 Mev-cm2/mg and a Total Ionizing Dose (TID) rate of > 1 Mrad(Si). It has as fast as 450 MHz DSP technology with flexible embedded processing and as many as 130,000 logic cells.

Funding restrictions also are forcing integrators to look more closely at what they need more than what they would like to have. “Some U.S. military programs are looking more closely at what requirements they really need,” Ortega continues. “If a program is not frontline and more experimental in nature, they may not need 100 or 300 krad pedigree products. If it is not mission-critical, they may take a step back regarding quality and radiation immunity.”

Sidebar 1: Rad-hard market trends

(Click graphic to zoom)



“DoD programs today don’t have the luxury of test driving multiple products anymore, so they are being much more selective in spending only on what is essential for their program,” Pyle says. “We are seeing a higher level of critiquing in terms of what they really need. For example, does their launch vehicle really require 100 krad total dose? Or, what are the radiation requirements for that satellite over time? Maybe a 100 krad high dose rate is extremely unlikely in low Earth orbit, but continual low dose is very likely; therefore, they should use a product rated for ELDRS.”

“One bad trend stemming from the shrinking geometries of space components and push for reduced Size, Weight, and Power (SWaP) is that as sizes shrink, we end up packing more and more transistors on a single device, which enables more performance in a smaller footprint; however, the device becomes more susceptible to contact radiation as one SEU or SEE can fry multiple transistors at once,” Patterson says.

Reduced SWaP and enhanced signal processing

The more rigid rad-hard specifications often are coupled with demands for enhanced signal processing and higher-density memory devices in sensor payloads used for Intelligence, Surveillance, and Reconnaissance (ISR) applications.

“The biggest trend we are seeing is for more signal processing capability onboard the satellite to drive scientific and military sensor payloads for hyperspectral imaging, radar, etc.,” says Ken O’Neill, Director of Marketing, Space Products, Microsemi SoC Group in San Jose, CA. “The science community wants more data and sensors with higher resolution, which requires more processing on the satellite payload. It is exactly the same challenge with Unmanned Aerial Vehicle (UAV) payloads, as we are dealing with a limited communication bandwidth down to the ground. In a UAV payload, FPGAs help fulfill the need for onboard processing that has historically been driven by new families of rad-hard ASICs and is now being driven by new families of FPGAs for onboard processing. We are in an advanced stage of development with an FPGA product that has significantly enhanced DSP capabilities.” For signal processing applications, Microsemi also produces the RTAX-DSP space-flight FPGAs, which add embedded radiation-tolerant multiply-accumulate blocks that integrate DSP functions into a single chip without any external components for code storage and without using multiple-chip implementations for radiation mitigation.

“The primary trend today among our military customers is an increased demand for higher-density nonvolatile memory products and a higher demand for higher-speed and higher-resolution A/D and D/A converters,” says Larry Longden, VP and General Manager Microelectronics at Maxwell Technologies in San Diego, CA. “They need more nonvolatile memory to run computers as the software becomes more complex in satellite systems. One of the big drivers is having the ability to store multiple images in flash to replace ‘be able to keep reprogramming’ with ‘provide multiple images for the’ new Xilinx FPGAs to support reconfigurable computing. Traditionally, our biggest product has been our EEPROM device. This year we will introduce new NOR flash and NAND flash products to provide higher-density nonvolatile memory. They have SEU-hardened flip-flops to protect against heavy ion radiation effects.”

SWaP demands also are pushing military space system designers toward a higher level of integration. “[Many] today still go about developing rad-hard systems like they did in the 1970s and 1980s,” Kemelring says. “They use off-the-shelf components such as op-amps, A/D or D/A converters, or build their own out of different components and maybe occasionally use an FPGA – or they use the old 4000/7400 devices to build up logic circuits. That is not how anybody makes anything anymore. If they could integrate all these components on one chip, they’d improve their SWaP by an order of magnitude.

Sidebar 2: CompactPCI in space

(Click graphic to zoom)



“At Triad Semiconductor, we make a mixed signal Via Configurable Array (VCA),” Kemelring continues. “It is a semi-custom ASIC in that it takes silicon-proven analog and digital resources that are radiation-hardened by design or process into VCAs that can be configured in one chip with a single mask layer. VCA technology integrates analog and digital resources onto preconfigured ASIC arrays. The digital and analog regions are shielded and isolated to protect sensitive analog circuits, then the entire array is overlaid with a global routing fabric.” The VCA has a TID of > 1 Mrad(Si) and an ionizing dose rate of > 5×108 rad(Si)/s, migrates ELDRS effects, and operates as high as 70 V, according to the Triad data sheet. “Semicustom ASICs could also be reused as the components have already been qualified for radiation environments,” he says. “The only thing that would change would be the single configurable via layer to modify the IP. The AFRL is really behind us on this and is trying to drum up funds to get us to go further. Custom rad-hard mixed signal VCAs also are available. The VCA still has yet to go through Aerospace Corp. for full radiation testing.”

Power management

“Our customers want faster systems with more baud rate, more bandwidth, more computing capabilities – all at lower power,” says Tony Jordan, Director of Standard Products at Aeroflex Colorado Springs. “That is our focus – to get more compute capability while keeping power consumption flat or reducing it. It is about increasing the logic per square millimeter. The request for increased computing capability is all about sensor processing – acquiring the data, processing the data, and either moving the data or making a decision. ASIC technology excels in these types of computing- and communication-intensive applications. Aeroflex is working on high-efficiency power, clock, voltage supervision/monitoring, and fault monitoring solutions.”

“The increase in processing power of today’s devices has increased demands on power systems,” VPT’s Pyle says. “Increased device speeds plus transistor density require operation at low voltage and high current. This is tough on a power supply and not compatible with efficient distribution. For systems with multiple low-voltage load requirements, users should have one isolated converter to provide the critical isolation barrier and feed the power to multiple, high-efficiency, nonisolated Point Of Load (POL) converters. This brings improvements such as better efficiency and reduced size and weight.

“VPT’s new SVR series grows from the company’s SV series. There was a need for a higher-radiation-level product – SVR is 100 krads TID while the SV is 30 krads – and new requirements with TOR through the Aerospace Corp.,” Pyle continues. “The SVR also is 85 MeV-cm2/mg standard compared to 44 MeV-cm2/mg standard for SEU resistance. The SVR product family includes DC/DC power converters, EMI filters, and a point of load converter. It has pin-for-pin compatibility with existing designs and is flexible due to its hermetic hybrid construction. Both SVR and SV are available in Class H and K, have ELDRS, and have a temperature range of -55 °C to +125° C.”

Sidebar 3: MIL-STD-1553 still getting wins in space

(Click graphic to zoom by 3.0x)



“Going forward, we need more components onboard with better efficiencies to manage all that power consumption,” Microsemi’s Ortega says. “We are doing a lot of work with higher efficiencies in DC-to-DC converters and experimenting with wideband semiconductors in those designs. We expect to come out with high-efficiency products in the next year or two. All new technology will be Gallium Nitride-based technology, and we will integrate those with our hybrid products as well. Microsemi has had capabilities in DC-to-DC converters through our Power Management Group (PMG) in La Mirada, CA, and is developing point of load converters.”

(For more on rad-hard power trends, see the article from Crane Aerospace.)


Figure 1: Rad-hard listing

(Click graphic to zoom by 1.9x)