Amiga.org
Coffee House => Coffee House Boards => CH / Science and Technology => Topic started by: Floid on February 22, 2004, 01:21:20 PM
-
Well, Slashdot today links to the EETimes, finally detailing exactly what went wrong. (http://www.eetimes.com/sys/news/OEG20040220S0046) Not Wind River's fault; more of a NASA/JPL SNAFU.
Note that the "R6000" mentioned is not a MIPS chip, but the PowerPC-based "Rad 6000" board we already knew about.
...and apparently those "256Mb" of Flash were in fact megabytes, which makes sense once you read what they're using them for. (Apparently repeat writes aren't a concern within the mission profile. ;))
-
Facinating!
-
"My daddy always used to say that if you wanted to put in a nail, you didn't do anythin' fancy, you just took a hammer and hit that son-of-a-bitch until it goes in there." - Soldier.
Which is my way of saying that I still think the NASA rovers were over-engineered.
-
KennyR wrote:
Which is my way of saying that I still think the NASA rovers were over-engineered.
No, if they hadnt had the features in question we either would have had a rover get to mars without the additional features that were added or a system that couldnt have recovered from the errors as it has. Why imply something is overengineered when it has allowed itself to be fixed from 150 million miles away.
-Tig
-
@Tigger
Making something self-repairing is not the same as making something so simple it wouldn't (and can't) have had errors in the first place. Russian engineering was always clunky and simplistic, but always very sturdy and practical at the same time, from T-34s to rockets. In some cases adding self-repair is even a detriment, not an advantage.
I'm very skeptical at the introduction of high level operating systems to probes at all - these are basically for the human interface. Remove the high level stuff and you can make the chips simpler and sturdier - not to mention cheaper.
-
Hmm... Although I understand what KennyR is saying, and often do agree along those lines, I think I'm with Tigger on this one. If anything, it wasn't engineered enough! The software department should have had something better in place to handle a simple memory allocation problem! The system should have also had some sort of "safe mode" built into it's boot process, so that it could do the bare minimum to get itself up and running and communicate with Earth after experiencing a failed boot attempt.
- Mike
-
Tigger wrote:
Why imply something is overengineered when it has allowed itself to be fixed from 150 million miles away.
-Tig
very VERY good question.
There's one thing: KISS (Keep It Simple Stupid)
and there's another thing: make hw and sw coping with EVERY circumstance. So you have to begin with calculating all circumstances. Kinda hard/expansive in space.
-
Isn't this why the term "mission-critical" was first coined?
-
KennyR wrote:
Making something self-repairing is not the same as making something so simple it wouldn't (and can't) have had errors in the first place. Russian engineering was always clunky and simplistic, but always very sturdy and practical at the same time, from T-34s to rockets. In some cases adding self-repair is even a detriment, not an advantage.
I'm very skeptical at the introduction of high level operating systems to probes at all - these are basically for the human interface. Remove the high level stuff and you can make the chips simpler and sturdier - not to mention cheaper.
Interesting PoV, which I'm sure has been discussed to death in NASAs engineering labs. I think your approach has been rejected simply because the mission profile is too complex to be handled by what you call something 'clunky and simplistic'. It's simply an optimisation problem. Given a mission profile where small failures are a certainty, is a design made up from *lots* of simple, sturdy and stupid components better than a design made up from *a few* yet smart and self-repairing ones? Remember, lots of components weigh more than a few, and weight is an expensive commodity to carry around. And you are always facing the problem that in if in case of clunky and simplistic things *do* go wrong, you have just spent hundreds of millions of dollars to put some metal and advanced plastics on a big round rock.
My point: the problem is too complex to be dealt with by 'simpler is better' mantras.
-
Hum, but simpler also means cheaper...
I think that NASAs agenda was towards having many low-cost missions, rather than a few high budget ones.
The use of `off the peg` and recycled parts must be a better solution to a cash strapped nasa.
But i suppose that this particular mission is very public and news worthy so i imaging that they did provide a lot of `redundant` features to the design...
Either way they have brought us marvellous discoveries...
-
blobrana wrote:
Hum, but simpler also means cheaper...
I think that NASAs agenda was towards having many low-cost missions, rather than a few high budget ones.
This is a low-cost mission.
The use of `off the peg` and recycled parts must be a better solution to a cash strapped nasa.
But i suppose that this particular mission is very public and news worthy so i imaging that they did provide a lot of `redundant` features to the design...
They used the same space qualed processor card they have used lately, used the FAA qualed version of VxWorks, which they have used in the past and got to reuse lots of code, the "simpler" design would have had to be custom and would have cost more to make. Also when this situation happened we would have lost Spirit if it was a simple system. The US is the only country to have successful landers on Mars, we've had 5, its silly to imply we are doing it wrong, when the other countries efforts have failed. If we lose Spirit or Opportunity tomorrow, they will still have each been a huge success, and in all likely hood we'll be hearing from them for at least the next 45 days or so.
-Tig
-
@Tig,
Didn't the Soviets/Russians succesfully land probes on Mars?
- Mike
-
Glaucus wrote:
@Tig,
Didn't the Soviets/Russians succesfully land probes on Mars?
- Mike
I'm trying to find something funny to say about the "Red planet" but I'm drawing blanks!
-
T_Bone wrote:
I'm trying to find something funny to say about the "Red planet" but I'm drawing blanks
hm, Mars as the god of war, the agressor. Mars as in Martians attacks the world (War of the worlds). Red, as being the color of blood, the color of the enemy.
-
Mars has been particularly unlucky for the russian space program, and unlike the success's of the venerahttp://nssdc.gsfc.nasa.gov/planetary/venera.html (http://nssdc.gsfc.nasa.gov/planetary/venera.html) spacecraft which landed on a more hostile planet...
And how they have tried...
http://www.geocities.com/goarana667/Marslist.htm (http://www.geocities.com/goarana667/Marslist.htm)
-
Glaucus wrote:
@Tig,
Didn't the Soviets/Russians succesfully land probes on Mars?
- Mike
Two close trys, lots of other misses. Mars 3 may have transmitted for a little bit less then 20 seconds upon landing in 1971, it then was DOA, there still is a debate about whether this was an actual message or whether it was radio traffic that happens when you drop a lander too hard. That could have been resolved if the orbiter had not lost fuel, not made its predicted orbit and crashed much earlier then expected, though there are some spectacular pictures from that craft. Mars 6, sent about 4 minutes of data on its descent before crashing into the surface. Due to damage caused by its spaceflight, most of the data was useless from the lander. And thats the closest anyone but the Americans has come, thats why we were cheering for the Beagle2, the ESA would have accomplished a task the Russians had not.
-Tig
-
Strange how the Ruskies have failed on Mars while have succeeded on Venus. What makes Mars so tricky?!?
- Mike
-
Glaucus wrote:
Strange how the Ruskies have failed on Mars while have succeeded on Venus. What makes Mars so tricky?!?
- Mike
Mars is really far away.
-
The Russians never put a man on the moon either. Actually its very surprising what they did manage to do, given that most of their GNP went on their military and nuclear arsenals.
-
Cymric wrote:
KennyR wrote:
Making something self-repairing is not the same as making something so simple it wouldn't (and can't) have had errors in the first place. Russian engineering was always clunky and simplistic, but always very sturdy and practical at the same time, from T-34s to rockets. In some cases adding self-repair is even a detriment, not an advantage.
I'm very skeptical at the introduction of high level operating systems to probes at all - these are basically for the human interface. Remove the high level stuff and you can make the chips simpler and sturdier - not to mention cheaper.
Interesting PoV, which I'm sure has been discussed to death in NASAs engineering labs. I think your approach has been rejected simply because the mission profile is too complex to be handled by what you call something 'clunky and simplistic'. It's simply an optimisation problem. Given a mission profile where small failures are a certainty, is a design made up from *lots* of simple, sturdy and stupid components better than a design made up from *a few* yet smart and self-repairing ones? Remember, lots of components weigh more than a few, and weight is an expensive commodity to carry around. And you are always facing the problem that in if in case of clunky and simplistic things *do* go wrong, you have just spent hundreds of millions of dollars to put some metal and advanced plastics on a big round rock.
My point: the problem is too complex to be dealt with by 'simpler is better' mantras.
Funny thing is that you say app. the same as I say, but with more words and smoother language usage. :-)
-
But did you notice how errors crept into his typing...?
if he had used fewer words then the chance would be lowered, (but the meaning is quite clear with the errors, due to the bulk context) :-)
-
Glaucus wrote:
Strange how the Ruskies have failed on Mars while have succeeded on Venus. What makes Mars so tricky?!?
With its orbit, its usually at least twice as far away from us then Venus, plus its twice as far from the Sun, which isnt good for solar power, and it has a thin atmosphere (compared to us or Venus soup).
-Tig
-
Making something self-repairing is not the same as making something so simple it wouldn't (and can't) have had errors in the first place.
Like a rock. :-)
I don't think that simplicity is the way to improve reliability. Redundancy is a much better way.
Just look at todays web/email/database servers! Loads of redundacy there! You'll find that a server is a more complex machine than your average workstation. And it's that way for a reason.
These devices certainly don't need to be made simpler, just more reliable (or perhaps they can be tested a little more?)
Whichever way they go, they sure are quite successful in doing what is not an easy task.
-
KennyR wrote:
The Russians never put a man on the moon either. Actually its very surprising what they did manage to do, given that most of their GNP went on their military and nuclear arsenals.
Yeah, but that have been to the moon and brought rocks back from the moon. Its not a technical feat they are incapable of doing, they have however killed all 8 of their Mars landers and basically all 16 Mars efforts have been pretty dismal from the USSR. Next year we launch the Mars Reconnaissance Orbiter, which will be able to find all those crashed vehicles since it will be able to pick up items as small as a dinner plate on the surface of mars all the way from Orbit, and then in 2009 we launch the Mars Science Laboratory which theoretically will run for years on its nuclear power cells
-Tig
-
Hm, so there's gonna be nukeplants put on a big pile of explosives? Wouldn't want to be around it when the launch fails :nervous:.
-
Found this link in some a.orger's sig :lol:.
(http://www.rebelion.org/tragicomix/mars_spiritcolor.jpg)
-
odin wrote:
Hm, so there's gonna be nukeplants put on a big pile of explosives? Wouldn't want to be around it when the launch fails :nervous:.
:lol:
-
odin wrote:
Hm, so there's gonna be nukeplants put on a big pile of explosives? Wouldn't want to be around it when the launch fails :nervous:.
The US has done 7 I believe in the past, a challenger like explosion of the craft in question would not harm the system. The Russians may well have done more then us, my link to that info is not operating at the moment.
-Tig