PDA

View Full Version : Space Shuttle Computers



jgravatt
2003-Sep-25, 04:01 PM
I heard somewhere awhile back that the computers used on the shuttle are extremely antiquated by our consumer standards. I forget exactly what the reason was however--extensive testing or something? Does anyone know the reason, and approximately what level of stuff they have on the shuttles now?

diddidit
2003-Sep-25, 04:19 PM
Well, they're late-1970's technology, but I believe they're being updated.

However, my limited contact with the space business brought me the term "space-qualified." Space-qualified means $$$$$ and absolute, proven reliability. I can't imagine there will be, say, a Pentium 4 on the thing (at least as a flight computer - maybe for experiments) - probably P1 at the most.

did

Andreas
2003-Sep-25, 04:35 PM
Space-qualified means $$$$$ and absolute, proven reliability. I can't imagine there will be, say, a Pentium 4 on the thing (at least as a flight computer - maybe for experiments) - probably P1 at the most.
Are there radiation hardened Pentiums already? I remember they went up to 486 in recent years. (Of course, there are many other CPUs apart from Intel's for space environments.)

Also, you don't have to go to space to see "obsolete" hardware in use. How many P4 3GHz computers with 1GB RAM, 200GB harddisk and GeForce FX graphics can you count in any modern jet cockpit? "Consumer standards" are totally different from industrial and aerospace standards.

jgravatt
2003-Sep-25, 05:10 PM
Oh yes, I do think I remember now that they have to be radiation hardened...

parejkoj
2003-Sep-25, 05:29 PM
Are there radiation hardened Pentiums already? I remember they went up to 486 in recent years. (Of course, there are many other CPUs apart from Intel's for space environments.)

Yup. They're about a million dollars a pop though, so not many actually are in use. They aren't exactly Pentiums, but something close in terms of processing power. They actually came out about a year ago, so maybe the price has gone down since I heard about them.

There are actually two main reasons that spaceflight computing power lags behind consumer technology: rad hardening and development freezes. The first one was already mentioned: it is quite a challenge to make a processor/memory/bus/supporting hardware not go haywire when smacked with energetic particles. The delay between comercial and rad-hardened processors (I'm not sure about memory) is about 4-5 years. Keep in mind that the MHz rating of a chip is actually useless when comparing processors, so if you see comparisons based on that, they can probably be ignored...

Development freezes are necessary because you have to actually build the thing that you design and at some point everyone has to agree on what is going into it. Since it usually takes a few years to build, and then a couple more to test, any spaceflight system (at least with the operating procedures JPL and NASA use now) has another 4 years or so delay from when the freeze happens to when the thing is launched.

diddidit
2003-Sep-25, 06:42 PM
For some reason, it's stuck in my overburdened head that the original shuttle computer had roughly the processing oomph of an Apple II. Which is fine - for the most part, it's a glorified timer.

did

parejkoj
2003-Sep-25, 07:19 PM
For some reason, it's stuck in my overburdened head that the original shuttle computer had roughly the processing oomph of an Apple II. Which is fine - for the most part, it's a glorified timer.

did

I'd say that sounds about right. Remember, you don't need massive processing power to do certain, very specific tasks. There's a reason I hadn't upgraded my computer in 5 years! (until last week: need a newer video card for games that I've started wanting to play... my requirements are getting too general :) ).

And it is more than a glorified timer: it has to coordinate readings for a whole mess of sensors and check them against "safe" values, while also monitoring fuel flow and... you see where this is going. It's just that it don't need as much power as one might think, since all (or most) of the tasks are well known before hand, and you can design it with that in mind.

Betenoire
2003-Sep-25, 07:57 PM
Maybe the problem is they just can't get a computer without a Windows virOS on it these days.

jscotti
2003-Sep-25, 08:26 PM
The original shuttle GPC computers were designed starting in 1972, so the technology had to be frozen and built and tested well before the first flight in 1981. As others have already said, you don't need much speed for what the computers are called on doing during flight. They were updated in the late 1980s with a program begun in 1984 which replaced the original computers with machines that were about twice as fast and have about 2.5 times the original memory. Reliability is the key along with redundancy, and the code used has been tested and flown over and over again. Any replacement today would need similar testing. The late 1980s replacements was able to run the identical software used before the upgrade.

The crews do use PC compatible computers on the flight deck for non-mission critical functions like e-mail and even controling experiments or monitoring the orbiters ground path while doing Earth observations at the windows.

Here's a couple links to webpages describing the shuttle computers:
http://spaceflight.nasa.gov/shuttle/reference/shutref/orbiter/avionics/dps/gpc.html
http://www.wikipedia.org/wiki/AP-101
http://www.hq.nasa.gov/office/pao/History/computers/Ch4-3.html

Jim.

Avatar28
2003-Sep-25, 09:57 PM
I thought the new glass cockpit now used rad hardened 486 computers. They were installed a few years ago I believe.

George
2003-Sep-25, 10:01 PM
I do remember the Hubble was upgraded with the 486.

Firefox
2003-Sep-26, 12:24 AM
I thought the new glass cockpit now used rad hardened 486 computers. They were installed a few years ago I believe.

Atlantis was the first one equipped, during her last OMDP, I believe. Columbia received her new cockpit in 2001. I'm not sure whether Discovery or Endeavour have theirs yet, though I'm sure they will pretty soon.

On a side note, I've heard that NASA has reorganized its OMDP program to where ferry flights to the Palmdale plant were unnecessary, and that all modification and maintenance work could be done in Florida.


Adam

Hale_Bopp
2003-Sep-26, 02:27 AM
I just talked about this in my astronomy class. My students were dismayed to hear how antiquated the computers on Hubble are. They hadn't even HEARD of a 486! I went into a lot of the reasons for this, but they still don't seem impressed.

Hey, if you don't have to run Windows XP, you can get by with a lot less :D

Rob