The Xilinx View On Neutron SEUs

The effect of cosmic radiation on ICs is nothing new to the industry or Xilinx:

  • in radiation-intense environments like the upper atmosphere and in space, SEU events are common and systems must be designed to compensate
  • on the ground, this effect is far less intense. Cosmic Rays striking the atmosphere generate a neutron flux that peaks at 60K feet and decrease to a small fraction at ground level. The energy generated by these particles upon collision with an IC has only recently been observable, however, at a very low level.

No instances of atmospheric neutron-induced SEU reliability issues have been reported to date:

  • concern has been raised because of theoretical calculations
  • the relatively small number of neutrons reaching devices at sea-level makes the likelihood of an SEU-induced error very small

Xilinx has taken this issue seriously; we've done our homework:

  • based on Xilinx work with space-qualified products, SEU effects are well characterised for our devices
  • accelerated testing normally performed for atmospheric neutron SEU predictions are difficult to correlate to real-world behaviour
  • Xilinx performs a series of ongoing experiments called 'Rosetta' where boards each containing 100s of devices are monitored for long periods of time at various locations of differing elevations and compared to accelerated testing.

These non-accelerated experiments give more realistic results than those extrapolated from high-density neutron beam testing.

Xilinx has been running tests on SEUs for years and publishes the data: see http://www.xilinx.com/ xlnx/xil_prodcat_product.jsp?title=aero_radiation

'NSEU Sensitivity of SRAM-based FPGAs' MAPLD 2003 Paper C5. Joe Fabula, Austin Lesea, Carl Carmichael, and Saar Drimer Abstract: fabula_a.html.

Presentation: c5_fabula_s.ppt Upset with Neutrons.

IC designs are continually improved to compensate.

Xilinx 90nm FPGAs show approximately 15% better results than previous technologies, despite predictions that 90nm would be more susceptible.

The Mars 2003 mission used Xilinx devices to control the descent and landing, including rockets firing. Virtex devices are used in the rover for all the motor control functions, wheels, instruments, etc.

To determine how important SEUs are on the ground in an application, we need to know the actual probability that the device we're using will experience such an upset.

Designing robust circuits to tolerate Neutron SEU effects is just one of the many challenges of advanced IC technology. The 'head in the sand' alternative is to stay with older, lower complexity technology which is not really viable. Then there are other reliability concerns associated with increased chip count, PCB size, number of interconnects, etc.

SEU are just not a big system level concern for most designs at this stage. Moving forward, the right approach is to quantify the phenomena as accurately as possible, take practical measures at the IC design level and use system design techniques as required.

The IC manufacturers are most qualified to test their devices and develop applications collateral to enable their customer to reach the required system objectives.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish