Newsgroups: sci.space.policy
From: Henry Spencer <henry@zoo.toronto.edu>
Subject: Re: How expensive is space?
Date: Wed, 21 Feb 1996 16:28:57 GMT

In article <4g39qq$gji@utaipx02.uta.edu> TJM2326@utarlg.uta.edu (Timothy J McGaha) writes:
>Radiation. Down here we don't worry about it much, since we've got some
>protection from the atmosphere. But up in low Earth orbit and even more so
>in geosynchronous orbit, the radiation environment is much more intense...

It's not that bad.  There are two aspects of the problem:  electronic devices
failing for good due to accumulated dose, and transient errors due to single
energetic hits.

Accumulated dose is a non-issue in LEO if you are a little bit careful
about which parts you use; a wide range of standard commercial parts will
tolerate 5-10 years in LEO without difficulties.  Transient errors require
some attention in circuit design and software design, but can be dealt
with quite successfully -- they are relatively uncommon in LEO (and are in
fact extremely rare outside the South Atlantic Anomaly, although most LEOs
do pass through the SAA regularly).

Ordinary consumer electronic gear runs just fine in LEO, in fact, within
the shuttle.  More-or-less off-the-shelf commercial laptops are now used
quite extensively for non-critical shuttle operations.

Environments farther out require a bit more attention to this sort of
thing, but even so, with some shielding and some attention to design, it
is a tolerable problem outside the Van Allen belts themselves.  (Building
stuff to function for long periods within the belts is another story.)
Significant parts of Clementine's electronics used commercial chips.

We're not talking about exotica here.  Slightly different designs are
needed, and more attention to error handling is necessary, but the cost
difference is small if the engineers know what they're doing... and *if*
they aren't also striving to meet severe requirements for low power,
minimum mass, 99.99999% reliability, etc. while dealing with mounds of
paperwork.  The radiation environment is a nuisance but not a huge
problem. 
-- 
Space will not be opened by always                 |       Henry Spencer
leaving it to another generation.   --Bill Gaubatz |   henry@zoo.toronto.edu

From: "Jordin T. Kare" <jtk@llnl.gov>
Newsgroups: sci.space.tech
Subject: Re: Satellite design question
Date: 6 Feb 1997 23:42:35 GMT

In article <5cp2ug$3v@hacgate2.hac.com> Ben Muniz, bmuniz@ccgate.hac.com
writes:
>>	What is so space-specific, other than propulsion systems? Why
>>can't a spacecraft use an off-the-shelf processor, properly packaged, of
>>course?
>
>In one word: radiation.  If you could design proper shielding (to prevent 
>Single Event Upsets & other events) that wasn't too heavy, it wouldn't be an 
>issue, but for some reason s/c designers like to launch sensors & 
>transponders, not lead.  :-)

It's not very hard to shield against some types of radiation, especially
if you're shielding a modest volume; a centimeter of aluminum, for
instance, can drastically reduce the radiation dose due to low-energy
particles in trapped-radiation belts like the Van Allen belts or
Jupiter's radiation belts.
However, there's always some tail on the particle spectrum, so you tend
to need some level of inherent radiation resistance -- a level which can
be pretty easily met these days with appropriately-chosen commercial-grade
devices.

There's been much more concern in my experience about single-event
upsets (caused by high-energy cosmic rays, and thus essentially
un-shieldable*) than about total dose effects, at least for missions in
LEO or moderate-duration GEO-and-up missions away from solar maximum.
SEUs can cause anything from glitches to nondestructive latchup to
catastrophic failure; fortunately, even most commercial devices no
longer suffer catastrophic failure (although lots of early CMOS did).
You can cope with SEU's by sensing errors or latchup (excess current
drain) and power-cycling, but that adds complexity (for the error-
detection and power switching) and thus decreases reliability.  Also,
of course, power-cycling or rebooting in the middle of a critical
operation may be a Bad Thing -- or it may not matter at all for, say, a
sensor controller.

So you look at the probabilities, and the possible consequences, and
your budget :-) and maybe do some testing, and make hard choices about
how much risk to take.  Or, you go by the book and demand a rad-hard,
latchup-proof, space-qualified device and end up flying 10-year old
technology :-( (Vacuum tubes are completely proof against SEU's :-) :-))

*[[It's possible, of course, to reduce the radiation load to what it is
on the ground -- but it takes 14.7 pounds per square inch of shielding
:-) ]]

>BTW, the radiation tolerance of a chip depends on what it's made of (GaAs is 
>more resistant than Si). 

In many cases, the details of the manufacturing process determine the
radiation tolerance of the chip.  Certain manufacturers' design rules
and processing approaches (things like doping levels and substrate
layer thicknesses) inherently yield rad-resistant devices; others yield
very rad-sensitive devices.  You can use off-the-shelf parts in space
*provided* you pick the right manufacturer (and he doesn't change his
processing without telling you!  Guaranteeing that the part you get
next week is the same as the one you got last week and tested is a
problem even in routine electronics manufacturing.  It's a lot worse for
if you want *exactly* the same part 5 or 10** years later -- which is a
typical military/aerospace requirement.)

**Or 20 or 50, if the Space Shuttle keeps flying as long as NASA says it
will :-(

Jordin (Rad-hard, but not fault tolerant) Kare

*** Note:  Jordin Kare is no longer an employee of Lawrence Livermore
National Laboratory.  He is a Participating Guest.  While the email
address jtk@llnl.gov will be valid for the immediate future, snail-mail
should be sent to Jordin Kare, P.O. Box 3042, San Ramon, CA 94583**