Justin:

On Friday, January 17, 2014 4:11:11 PM UTC-5, [email protected] 
wrote:
>
> Hi everyone,
>
> I am part of a student-run society, at university, that is responsible for 
> designing a CubeSat satellite. We are attempting to use the BeagleBone 
> Black as the on-board computer, however we have concerns with bit flips and 
> single-event upsets that can be caused by radiation. Of course we will be 
> using some shielding, but I would be interested to know if similar projects 
> were undertaken and also what you guys think about the BBB's tolerance to 
> radiation.
>

I once ran a Hitachi SH-2 in Indiana University's proton accelerator (which 
isn't nearly as powerful as actual space) for a few hours, and could 
trigger SEUs anywhere I wanted to in the SoC about as fast as I could turn 
on the beam.  I'm guessing that the smaller process geometry, lower 
operating voltage, and greater complexity of OMAP will make it perform even 
worse.  (For the record, I have no knowledge of OMAP in CubeSats because I 
haven't been paying attention lately.)

That's not to say that you can't use the BBB in LEO, but if you do then 
you'll need to be prepared to mitigate the damage or your overall mission 
will fail in short order.  In our case, we had a VERY sensitive current 
meter in the circuit that would power-cycle the whole board at the earliest 
sign of a latchup.  We also planned to do periodic power cycles to clear 
micro-latchups that wouldn't sink enough current to trip the current 
sensor.  We had several layers of double-sided watchdog timers, and we 
checked every one of our drivers (we weren't running Linux) to make sure we 
didn't have any polling on status bits that wouldn't eventually time out in 
some other way if the hardware got stuck.  Finally, the current threshold 
that would trigger a power cycle was software-controlled, so we could raise 
or lower it depending on what we knew we were doing at the time.

Tragically, the board never flew due to a long list of bureaucratic snafus, 
delays, and the other usual suspects.  But it still taught me quite a bit 
about how to harden software for the real world.  :-)

-- 
For more options, visit http://beagleboard.org/discuss
--- 
You received this message because you are subscribed to the Google Groups 
"BeagleBoard" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to