View Single Post
  #4  
Old March 18th 13, 04:38 AM posted to sci.space.station
snidely
external usenet poster
 
Posts: 1,303
Default Curiosity and other rovers issues with flash memory

"Jochem Huhmann" wrote in message
"Brian Gaff" writes:

Considering that flash memory is not knew, and is I understand used
extensively in the new capsule for humans and probably on the iss but I
don't know that for sure, they do seem to be having a lot of issues with
it
corrupting. One assumes this is due to radiation, but it could also be
due
to some inherent instability caused by read/writes or temperatures
experienced. Its interesting to see it happening already to the latest
rover while its only just started to happen ont the older rover.
I do hope someone is looking into this problem as its not a problem you
really want if your crew is many days or months away from earth in some
critical system.


I'm fairly sure that this is just one of the problems you have to deal
with by redundancy and backups. You can't expect to have no memory
corruption (or outright failure) at all, even under the best
circumstances. **** happens, everywhere and always, even in space.

On Saturday, Brian Gaff queried:

That is true, but one would imagine that memory protection is rather a vital
thing to have issues with, considering the number of years its been on
spacecraft. back in the old days they actually used to fly tape recorders as
I recall.These tended to stick in the varying temperatures out there.
Brian


The type of Flash memory is also important. NOR-based Flash tends to
be more stable, but costs more. It may also be slower. NAND-based
Flash is dirt cheap, which is good because you have to provide a bunch
of extra cells ... even on Earth, it is pretty much necessary to use
ECC, and to use wear-leveling algorithms.

Perhaps it was the ECC circuitry that triggered the failover. Note the
redundancy: the error(s) occurred on the A-side computer and the
B-side computer took over running the ship.

And Curiosity's project manager says this:
"The hardware that we fly is radiation tolerant, but there's a limit to
how hardened it can be," Cook said. "You can still get high-energy
particles that can cause the memory to be corrupted. It certainly is a
possibility and that's what we're looking into."
http://arstechnica.com/science/2013/03/flash-memory-issue-forces-curiosity-rover-into-safe-mode/


A quick look for relevant information doesn't lead me to any technical
data on the Flash memory involved (2GB per computer, per Wikipedia),
but I did see a comment by a veteran satellite programmer that there is
only so much you can do in hardware to protect from these sorts of
events (did it happen after a signficant solar flare?), so you always
have software looking for errors. Which is what happened here.

/dps

--
But happiness cannot be pursued; it must ensue. One must have a reason
to 'be happy.'"
Viktor Frankl