Martian software

In the October issue of ACM Queue is an interview Mike Deliman. Deliman
is the guy at Wind River who was responsible for the version of the
VxWorks operating system used on MER. He is now working at JPL.

He talks about the radiation-hardened microprocessors used on
spacecraft. In a sense they reverse modern engineering. Modern chip
engineering seeks to squeeze more and more stuff into each chip. The
smaller the chip components, the more vulnerable the components are to
radiation. So a rad-hardened chip emphasizes larger vias, resulting in
bigger, slower components. Add to this the extra qualification work and
the small production runs, and the PowerPC in MER is a good order of
magnitude slower than current domestic chips. The alternative is to rad-
harden the entire spacecraft, but this increases the spacecraft mass to
unacceptable levels.

The OS on MER uses about 2 Mbyte of storage. In total, MER uses about 30
Mbyte of storage. My digital camera has over an order of magnitude more
than this!

Deliman talks a bit about the memory management problems that caused
grief shortly after MER landed on Mars. He explained that JPL been given
everything they need to rebuild the OS from scratch - unusual for a
proprietary software product. Even if a vendor grants access to the
source code, the access is usually limited to seeing printed listings
and rarely if ever includes the build procedures, regression tests and
what not needed to build from scratch.

There is some speculation about "jamming" the radio transmissions with a
spacecraft. While admitting this is a real Cold War issue, it just isn't
feasible to hack a Martian spacecraft - any attempt to develop a Deep
Space Network would be detected and, presumably, shut down.

In short, it is worth the time to lookup the article.
--
Kevin Willoughby lid

The loss of the American system of checks and balances
is more of a security danger than any terrorist risk.
-- Bruce Schneier

In article ,
Kevin Willoughby wrote:
...the PowerPC in MER is a good order of
magnitude slower than current domestic chips. The alternative is to rad-
harden the entire spacecraft, but this increases the spacecraft mass to
unacceptable levels.

The other alternative is to just use "domestic" chips -- at least,
carefully-chosen low-power ones -- in the spacecraft. The radiation
hazard in space is greatly exaggerated, at least for relatively short
missions (like the MERs); there's a lot of superstition about the utter
necessity of rad-hard parts or massive shielding.

The MOST astronomy satellite is in an orbit where it gets more radiation
than the MERs did in cruise phase, never mind on the surface of Mars
(where the planet and its atmosphere between them cut the dose to about
1/3 of that in open space). 18 months and going strong... with not a
rad-hard component anywhere. Almost every electronic component in MOST
can be bought from Digi-Key; in fact, a lot of them were.

There is some speculation about "jamming" the radio transmissions with a
spacecraft. While admitting this is a real Cold War issue...

This exact concern was why the Apollo spacecraft's original specs demanded
that it be capable of flying the entire mission (including navigation to
the Moon and back) with zero help from the ground.

That level of paranoia didn't last long, however, and full autonomous
navigation capability was eventually sacrificed to free up scarce computer
memory. (They could still get home on their own, but couldn't do the
lunar landing without help.)
--
"Think outside the box -- the box isn't our friend." | Henry Spencer
-- George Herbert |

(Henry Spencer) wrote:
The other alternative is to just use "domestic" chips -- at least,
carefully-chosen low-power ones -- in the spacecraft. The radiation
hazard in space is greatly exaggerated, at least for relatively short
missions (like the MERs); there's a lot of superstition about the utter
necessity of rad-hard parts or massive shielding.

Right. The documented existence of radiation and documented problems
caused by it should be discarded as being superstition.

The MOST astronomy satellite is in an orbit where it gets more radiation
than the MERs did in cruise phase, never mind on the surface of Mars
(where the planet and its atmosphere between them cut the dose to about
1/3 of that in open space). 18 months and going strong... with not a
rad-hard component anywhere. Almost every electronic component in MOST
can be bought from Digi-Key; in fact, a lot of them were.

OTOH.. GP-B had had one confirmed radiation event and another
suspected in only four months of operation.

D.
--
Touch-twice life. Eat. Drink. Laugh.

-Resolved: To be more temperate in my postings.
Oct 5th, 2004 JDL

In article ,
Derek Lyons wrote:
...The radiation
hazard in space is greatly exaggerated, at least for relatively short
missions (like the MERs); there's a lot of superstition about the utter
necessity of rad-hard parts or massive shielding.

Right. The documented existence of radiation and documented problems
caused by it should be discarded as being superstition.

Where did I say that? Yes, there's radiation there. There's radiation
here too. Yes, it sometimes causes problems. The superstition is that
the *only* way to deal with them is to spend hundreds or thousands of
dollars per chip (I'm not exaggerating) on putting rad-hard parts
everywhere.

(Nor is the cash the only cost. The immense lead times on the parts are a
tremendous obstacle to an efficient development program. Those parts also
do bad things to your mass and power budgets, because often they're
grossly obsolete designs, so you need more of them to do the same job, and
they want higher voltages and higher supply currents.)

For *some* spacecraft, there is no choice, because the orbit they're in
involves high doses, or because the reliability and lifetime requirements
are hard to meet any other way. But for a lot of missions, the dose just
isn't that large, and the requirements just aren't that stiff, and there
are better ways.

...18 months and going strong... with not a
rad-hard component anywhere. Almost every electronic component in MOST
can be bought from Digi-Key; in fact, a lot of them were.

OTOH.. GP-B had had one confirmed radiation event and another
suspected in only four months of operation.

MOST has radiation-caused hiccups too. Probably more than that, but I'm
not sure because they aren't a big deal and nobody gets excited about
them. We expected them. The operations team deals with them. They don't
interfere significantly with the mission.

There are also some signs of component aging due to accumulated dose, in
one or two subsystems. But then, we're already past the end of MOST's
primary mission. And things are still working.

Trying to legislate these problems away with rad-hard parts is fearfully
expensive, both in direct cash outlay and in the less direct impacts on
the spacecraft and the development process. And it doesn't entirely work
anyway.
--
"Think outside the box -- the box isn't our friend." | Henry Spencer
-- George Herbert |

In article ,
says...
Is it available on-line?

To my surprise, yes:
http://makeashorterlink.com/?U1633661A
or
http://www.acmqueue.com/modules.php?
name=Content&pa=printer_friendly&pid=227&page=1

(If I had known, I would have initially posted the link rather than
summarizing.)
--
Kevin Willoughby lid

The loss of the American system of checks and balances
is more of a security danger than any terrorist risk.
-- Bruce Schneier

In article ,
says...
Kevin Willoughby wrote:
The OS on MER uses about 2 Mbyte of storage. In total, MER uses about 30
Mbyte of storage. My digital camera has over an order of magnitude more
than this!

You might remember, what did digital cameras have for storage say 4 years ago?
Also, with downlink rates, would more storage have done any good for the
mission?

More storage allows a different way to use limited bandwidth. The
Marscraft takes a lot of images and transmits thumbnails back to Earth.
The guys on the ground then decide which images justify transmitting
back home in full resolution, which can be subject to aggressive (lossy)
compression, and which are just not worth transiting at all. Image
transmission could also be prioritized, giving preference to those
needed for 'next day' mission planning, and allow slack time to be used
for images of mild interest.

This would require storage for, say, a couple of days worth of images.
IIRC, MER has megapixel imagers, so a gigabyte flash card feels like the
right size.
--
Kevin Willoughby lid

The loss of the American system of checks and balances
is more of a security danger than any terrorist risk.
-- Bruce Schneier

In article ,
Derek Lyons wrote:
...The radiation
hazard in space is greatly exaggerated, at least for relatively short
missions (like the MERs); there's a lot of superstition about the utter
necessity of rad-hard parts or massive shielding.

Right. The documented existence of radiation and documented problems
caused by it should be discarded as being superstition.

Mr. Spencer has helped design and operate electronic machinery which is
now in space. He knows whereof he writes.

In article , no-
says...
Mr. Spencer has helped design and operate electronic machinery which is
now in space. He knows whereof he writes.

True, but he isn't the only contributor to this thread who has deal with
radiation-related issues.
--
Kevin Willoughby lid

The loss of the American system of checks and balances
is more of a security danger than any terrorist risk.
-- Bruce Schneier