Right but if we can just simply "radiation-sheild" off the shelf devices, why didn't we do that instead of custom building a robot made of specifically radiation hardened parts that are also much more expensive?
Why was the most important consideration by far for the processor in the previous NASA project was designed from the ground up to be radiation-hardned, but yet we can launch this no problem?
I just don't get it. I also don't appreciate the downvotes. I'm not trolling or looking to start an argument. I don't have dog in this race, I'm genuinely confused.
The answer is that modern consumer memory packages have come a long way from core memory. Standard ECC tech is extremely resilient to bit-rot and can recover in cases where older memory technologies would have failed catastrophically.
And in the case of real memory corruption, i.e. where cosmic radiation overwhelms the ECC controllers ability to correct the bit rot, well .. "reboot and retry" has been a standard for years, and will continue to work as a strategy as long as it is considered by the software architects of the system.
I think it has a bit to do with how mission critical something is. With the Mars rover, if it fails that is a lot more money and time down the drain than if a small phone satellite does.
As I commented on a Curiosity article, I hope they eventually tend more towards using lots of commodity hardware running instructions in parallel and working off probability rather than any one processor having to be perfectly deterministic.
I'm a huge fan of the possibility of heavily shielding off the shelf parts. It seems logical to me to try to incorporate it into unmanned craft.
However, I'm guessing there are probably people who have simulated this and it probably hasn't worked out to be very economical for past missions. I think part of the problem is that the level of shielding required would really limit your payload to orbit. It's probably going to take a huge amount of material seeing as even the Space Shuttle couldn't act as a perfect buffer between computers and radiation.
From a NASA article[1]:
"Designers also found out that laptops would crash when the shuttle passes through the "South Atlantic Anomaly," which is an area where the magnetic field draws in to Earth, again offering less radiation filtering for spacecraft flying through it."
Completely speculating here, but I would imagine that part of this experiment is to get operational experience with using standard grade electronics in space.
From the OP's article, the satellite has:
"...a watchdog circuit that monitors the systems and reboots the phone if it stops sending radio signals."
Anything with a human or with a high price tag attached to outages can't work with that caveat. If DirectTV's satellites did that, people wouldn't stay signed up for long. Imagine a Mars rover having to reboot every half hour... not very convenient.
But perhaps NASA might want to know if they can cheaply build "disposable" orbital experiments... i.e. ones that can complete their missions in a few hours. The sort of reliability where you just have to deal with the odd 30 seconds of rebooting every so often could be alright, especially if the payload can be stowed on a cargo ship and simply shoved out an air lock like the SuitSat[2].
Again spitballing, maybe they will find that things work out more or less OK if they send up five moderately shielded Raspberry Pis and they "vote" a la the Space Shuttle's flight computers[3]. If so, maybe that opens up space experimentation to universities with smaller budgets to squeeze onto resupply/manned launches that happen to have a spare square foot of space.
