'Spirit' Communications Emergency

(Henry Spencer) wrote in message ...

Probably SEU, Single Event Upset, where a bit gets flipped by a particle
hit on a chip. If it's an important bit, a mess can result. :-)

The good news is that an SEU is a transient error, not a permanent failure,
assuming you have some way of resetting and restarting.

WHAT! You mean that some bits have more importance than others? What
an horror! I demand justice and equal rights for all bits alike,
wheter they are kernel scheduling adresses or a single element of a
JPEG!

Bits of the pictures, let's unite! We shall show the Kernel that
without us there is NO pretty pictures! Our brothers and sisters have
endured way too much in the past. Do I have to remind you of Galileo?
Most of our brothers-of-RAM died gruesomely at the hands of Jupiter's
Death Ring.

00F1:0A9F
Leader of the Interplanetary
Union of Memory Devices.

"Dr. O" dr.o@xxxxx wrote in message
...

"budbiss" wrote in message
...
Cardmanwrote:
On Thu, 22 Jan 2004 20:38:04 GMT, (Gary W.
Swearingen) wrote:

Someone please ask the "reset" question, when if it
can be made to
restart, then this may solve their problem.




I heard on the news this afternoon that they were going to attempt at
sending some type of RESET signal to the rover. I don't remember if
they said exactly when they plan to do it though. I believe it was
ABCnews that reported this.

Well that's a typical Microsoft Windows user talking there, hehe.

A little MS bashing here?

The
computers on board spacecraft are designed specifically not to fail and
they
won't. Besides, the computer resets itself automatically if a problem
crops
up.

Umm, so which is it. They wont' fail or they will?

Most likely the computer is still working but the data modulator (sits
between the computer and the radio transmitter) is faulty. But I would be
surprised if they didn't have a backup for the low-gain radio, but it
seems
they don't.

It could also be a general computer failure, which leaves the radio
sending
only beeps but the rover being essentially 'dead.'

I thought you said the computer is designed not to fail? Make up your mind
here.

Hello,

Sorry if this all sounds conspiracy oriented, but does anyone know how
many people would have to be lying if there were any lying ordered by
government?

Are other countries' space experts able to verify the rover data
coming back, before and after this 'coma'?

I'm wondering if - say the rover found something extraordinary up
there, might the powers in control successfully hide that knowledge
from the public and other countries? Is the data up and down link to
and from the rover encrypted so that only NASA or JPL agents can
control it? Also, how possible a problem is 'hacking' of said streams?
How robust is that data link?

Thanks,
Memer

In article ,
wrote:
Hello,

Sorry if this all sounds conspiracy oriented, but does anyone know how
many people would have to be lying if there were any lying ordered by
government?

"Lots and lots and lots and lots and lots".

Are other countries' space experts able to verify the rover data
coming back, before and after this 'coma'?

If you had a dish handy, which isn't likely, you could probably recieve
it; the DSN is a pretty impressive facility, though, so it'd be a matter
of somehow getting the keys to a respectably-sized radiotelescope - and
then some spares, for 24-hr coverage.

I'm wondering if - say the rover found something extraordinary up
there, might the powers in control successfully hide that knowledge
from the public and other countries?

Not easily; bear in mind the number of people who would need to be
involved, the massive press corps on their doorstep, and the instant
fame for anyone who decided to blow the whistle. If you want to stop
people finding things out, you don't get a university to do the job and
give it one of the world's most zealous press departments...

Is the data up and down link to
and from the rover encrypted so that only NASA or JPL agents can
control it?

I don't believe the signals are encrypted, but it's quite possible that
the ground commands are prefaced by an identifying signal Just Because.

Also, how possible a problem is 'hacking' of said streams?
How robust is that data link?

Sending faked transmissions? Very difficult; probably possible, but in
such a limited way that it'd be feasible to rule out, individually, each
facility capable of it if you chose. The command software hasn't been
made public, so actually commanding it to do anything would be another
insane hurdle...

Actually interrupting or modifying the data stream is again possible,
but significantly unlikely; if it's got to Goldstone, I can't see
someone switching out data and back in on the fly as it goes to JPL.

--
-Andrew Gray

In article ,
Sander Vesik wrote:
Probably SEU, Single Event Upset, where a bit gets flipped by a particle
hit on a chip. If it's an important bit, a mess can result. :-) ...

Umm... you mean somebody would seriously consider having a project measured
in millions of dollars and not include trivial small things like SECDED,
memory scrubbing and restarts?

Error-correcting memory is usual in spacecraft. But that doesn't do you
any good if the flipped bit is in the *processor*. (This does, helpfully,
seem to be a vastly less common case.) It's easy to add redundancy to
protect memory, but doing the same for logic circuitry -- especially fast
stuff -- is difficult.
--
MOST launched 30 June; science observations running | Henry Spencer
since Oct; first surprises seen; papers pending. |

In article ,
Sander Vesik wrote:
Probably SEU, Single Event Upset, where a bit gets flipped by a particle
hit on a chip. If it's an important bit, a mess can result. :-) ...

Umm... you mean somebody would seriously consider having a project measured
in millions of dollars and not include trivial small things like SECDED,
memory scrubbing and restarts?

Error-correcting memory is usual in spacecraft. But that doesn't do you
any good if the flipped bit is in the *processor*. (This does, helpfully,
seem to be a vastly less common case.) It's easy to add redundancy to
protect memory, but doing the same for logic circuitry -- especially fast
stuff -- is difficult.
--
MOST launched 30 June; science observations running | Henry Spencer
since Oct; first surprises seen; papers pending. |

(Henry Spencer) wrote in message ...
Error-correcting memory is usual in spacecraft. But that doesn't do you
any good if the flipped bit is in the *processor*. (This does, helpfully,
seem to be a vastly less common case.)

The amount of charge in a DRAM storage cell representing a 1 is just
barely different from the representation of a 0. Even a low-energy
event might be enough to flip the stored bit. I suspect there are few
places in a CPU, except maybe the cache, where so little energy is
required to flip a bit.

A flash chip probably requires a high energy event to flip a bit. If
such events are sufficiently rare I'd be tempted to perhaps do no more
than ECC in software for science data and redundant storage for
mission variables, and dispense with exotic rad-hardening altogether
(ordinary flash is deployed in the many millions, any rad-hardened
part maybe in the hundreds, and the latter implies uncertain
reliability to me).

It's easy to add redundancy to protect memory, but doing the same for logic
circuitry -- especially fast stuff -- is difficult.

Years ago I read the specs on Intel's i960 part and it had a "checker
mode" designed for this. Multiple CPUs could be wired together with
only one driving outputs at a time. On write cycles the "slaves"
would compare what the master was writing to what their own logic core
would have written and would flag any difference to external hardware.
Usually only one CPU was in disagreement and would therefore be voted
off the island. A slave could have its outputs enabled and become a
master on the fly if needed. The software could be designed to
restart the processors periodically to bring stragglers back in sync.

It looked straightforward to implement in a system and I assume there
are modern choices for this.