[AR] Re: LEO radiation shielding
- From: Henry Spencer <hspencer@xxxxxxxxxxxxx>
- To: Arocket List <arocket@xxxxxxxxxxxxx>
- Date: Thu, 5 Dec 2019 16:25:25 -0500 (EST)
On Thu, 5 Dec 2019, Ian M. Garcia wrote:
Does anybody here have experience with Cubesat (or any satellite)
electronics in LEO?
A bit... :-) We have 23 satellites in orbit -- would be 25 by now if the
Vega launcher wasn't on hiatus after a failure -- every one that made it
off the launcher intact has worked, and almost all are still working.
These are mostly slightly bigger than cubesats, although there are a
couple of 3Us among them. The MOST astronomy microsat died last spring,
in its 16th year in a fairly high LEO. CanX-2 (a 3U) is still working in
its 12th year, AISSat-1 in its 10th...
I'm starting to look at this and comparing rad hard vs COTS with and
without shielding. Rad hard options are crazy expensive and low
performance compare to COTS, but I am worried about radiation for a
satellite that could be out there for a while.
For not-excessively-high LEO and not-ridiculous mission durations, the
need for rad-hard is a myth, period. There are still some people (e.g. at
NASA) who claim it's vital, but many, many satellite-years of experience
now disprove this. We don't use rad-hard parts, nor do we make any
particular attempt at shielding (the spacecraft structure supplies some).
We do take some basic precautions (bulk memories have error correction,
power parts are derated, irreversible operations require ready/arm/fire
sequences, and the spacecraft are designed so that software failures can't
kill or damage the hardware), and we test crucial parts in a proton beam
to look for bad behavior on single-event upsets (software crashes are
okay, wild electrical misbehavior or smoke is not). Unfortunately, our
proton results, and specific part choices in general, are considered
proprietary.
The expense isn't even the worst part of rad-hard parts -- ask about
delivery schedule. Many-month delays are absolutely deadly to any sort of
efficient development process, as is being terrified of ruining even one
part because it takes so long to replace; being able to order your parts
from Digi-Key and have them in your hands tomorrow is a huge advantage.
Rad-hard parts also make for bigger boards and more boards and more cables
and higher power consumption because they don't put nearly as much
functionality on each chip. The added design complexity is particularly
bad -- *most* electronics problems nowadays are design errors, not random
failures, and the more designing you have to do, the better the chance
that you'll goof somewhere.
...reading that lead is actually not used that much because it's too
soft, so instead they use tantalum, tungsten, or just thicker aluminum.
Not only is lead awkwardly soft (although some lead alloys are stronger),
but it's not nearly as dense as tungsten. Aluminum is not great as
shielding, but it's cheap, and just making the structure thicker usually
costs very little (if you can spare the mass).
But then I continue reading and it turns out that materials high in
hydrogen are better, so they are starting to recommend polyethylene
(HDPE) instead of metals? And then the latest is that HDPE combined with
a metal liner is actually even better because the metal captures
secondary radiation. What the hell? This is all very confusing.
Hydrogenous materials like HDPE do produce less secondary radiation, and
adding a metal liner can help intercept what does get produced, but the
tradeoffs are complicated (HDPE's low density is a disadvantage for this).
Because of the secondary-radiation problem, it can actually be better to
shield the hell out of one side and leave the other side bare, than to
put thinner shielding on both sides!
People who need a little bit of localized shielding still tend to use
tantalum or tungsten.
It *is* a confusing subject, with a lot of arcane literature and a marked
lack of easy-to-use design guides for practical engineering. Fortunately
it's not a big issue for LEO.
Henry
Other related posts: