One Of A Kind Space Hardware?
NASA JSC Solicitation: Single Board Computer For Space
“NASA/JSC has a requirement for a Single Board Computer for Space with the ability to clear errors in 1mS to meet the time to first failure estimate of 3 thousand years. This fidelity in error correction is necessary to meet the expected level of redundancy for a human rated spacecraft. A market survey indicates that Maxwell has the only commercially available singe board computer that will meet the Government’s requirements.”
Keith’s note: Wow. Only one company on Earth makes this device. 3,000 years without an error. Amazing.
I wonder how many others would qualify if the 3000 year (!) figure were reduced to 1000 or 500. And why is 3000 years the benchmark anyway?
Because Maxwell being a young company hasn’t had time to continually test a board for more than 3000 years. Or something.
It is interesting that the specific environment in which this device is supposed to be able to perform with a “time to first failure” estimate of 3000 years is not mentioned in the specification.
Is that 3000 year value for a desktop environment or while being bombarded with protons and heavy ions at one of the nuclear research facilities NASA uses to test onboard electronics?
I surely don’t understand why they need a 1ms ‘error clear’ or a 3,000 MTBF. How long is this thing going to run?? EO-1’s processor has been up there for 11 years in Polar LEO without rebooting, that’s not good enough?
Sorry, but I do spacecraft systems for a living and I can’t even imagine why a 1ms recovery is required, much less a 3,000 year MTBF. EO-1’s Mongoose V (a MIPS R3000 derivative) has been up in Polar LEO for 11 years without rebooting, and the RAD-750 available from BAE is flying on lots of stuff, both single string and redundant (LRO for instance)(well, LCROSS too, but we smashed that one!). That announcement baffles me…
Deep Impact (now EPOXI) is flying dual redundant RAD750s and has not experienced a reset in ~6 years.
The cynic in me says that someone with connections had this board and the specs were written to ensure a single bidder.
It says
“This fidelity in error correction is necessary to meet the expected level of redundancy for a human rated spacecraft. “
I would assume Orion’s boards are set so what else? The Space Exploration Vehicle (SEV) in its in-space configuration?
I think you haveto look at the solicitation more carefully.It says”human rated”. You have to wonder if they even know what requirements are needed or are just taking the most conservative design possible. No wonder costs are so high.
Obviously, some folks are planning to build a multi-generational starship. If you’re cruising slow and steady to Alpha Centauri at .1 g, then one error every 3,000 years might just be one error too many. 🙂
That does seem really high. I thought maybe it’s normal but rummaging around I found an MTBF figure of 44 years for a BAE RAD750 board.