I hadn't seen this listed in the archives. I hope everyone can pull something useful from it.
http://workmanship.nasa.gov/wppr.jsp
http://workmanship.nasa.gov/
NASA workmanship site
Author
Message
Good ammunition for "doing it right the first time". Of course, I don't expect any of us to be doing space-worthy hardware, but the failures are still failures and end-users get plenty pissed off when things don't work the way they are supposed to.
Jim
ka7ehk wrote:
Good ammunition for "doing it right the first time". Of course, I don't expect any of us to be doing space-worthy hardware, but the failures are still failures and end-users get plenty pissed off when things don't work the way they are supposed to.Jim
I doubt all that information is relevant to space hardware unless they deadbug parts, use wirewrapping, and use jumperwires on the spaceshuttle. Loads of great information though. I never in my life would have thought there was a proper way to attach jumpers and additional parts onto a cicruit board. I always though it was some type of kludge. I've also never in my life heard a circuit board called a wiring board. Is that some type of British notation?
Adam, you would be surprised how old the Spaceshuttle computers are. For example it uses a flight control computer (well, five of them, of which four work in sync) called the AP-101. That one was build in the '70. It was indeed once, well, upgraded. After the Challenger accident it got the enormous amount of 256K (yes K) memory, and the CPU was finally integrated into one single chip, instead of being build from TTL chips. The TTL CPU alone weighted 57 pounds, and an AP-101 consumed 600W (take this, PC gamers :-)). With the upgrade the power consumption was reduced to 550W.
ArnoldB wrote:
I knew how old they. That doesn't surprise me in the least. What would surprise me is if the parts were attached like in some of the documentaion on the website (Barebug, wirewrapping, and hay/bluewire). Then again the one spec for wirewrapping does mention spaceflight.
Adam, you would be surprised how old the Spaceshuttle computers are. For example it uses a flight control computer (well, five of them, of which four work in sync) called the AP-101. That one was build in the '70. It was indeed once, well, upgraded. After the Challenger accident it got the enormous amount of 256K (yes K) memory, and the CPU was finally integrated into one single chip, instead of being build from TTL chips. The TTL CPU alone weighted 57 pounds, and an AP-101 consumed 600W (take this, PC gamers :-)). With the upgrade the power consumption was reduced to 550W.
Level: Raving Lunatic
Joined: Mon. Dec 15, 2003
Posts: 4313 View posts
Location: Slovakia, Bratislava
Quote:
Wait, are you saying that an ancient piece of obsolete hardware is STILL being used? Wouldn't it be more power/weight efficient to use something modern? When I read the specs now, it seemed like the NGW100 might do a similar job just as well... Tell me they don't use punch cards or tapes... please...
That doesn't surprise me in the least.
daqq wrote:
Quote:Wait, are you saying that an ancient piece of obsolete hardware is STILL being used? Wouldn't it be more power/weight efficient to use something modern? When I read the specs now, it seemed like the NGW100 might do a similar job just as well... Tell me they don't use punch cards or tapes... please...That doesn't surprise me in the least.
I'll be completly honest. I can't tell if you are being sarcastic to me or are being completely serious. The only reason why I wasn't surprised is because that factoid has been around for a long time. The statistic I've always known is the original Gameboy had more processing power. On top of I have no idea what specification the processors run at but I do know Atmel's Industrial probably will not work.
Quote:
Wait, are you saying that an ancient piece of obsolete hardware is STILL being used?
If this is true, I'd suspect that the philosophy is: If it works - don't fix it. It is a proven piece of hardware/software.
OTOH the Shuttle (or rather "The Orbiter", which is the correct term for that winged piece of capsule, payload bay and engines that does the whole journey and returns, excluding the solid rocket boosters and the external tank) has had a complete cockpit overhaul at least once. I suspect there is more or less nothing in front of the commander and pilot that is the same as in the early eighties.
EDIT/PS: Check this out, in case you've missed it: http://www.nasa.gov/55644main_NA... . It's running maximized on my secondary monitor as I type.
daqq, just to rub it in. The AP-101, called the AP-101S after the upgrade, is based on a CPU design from 1966.
It's not the size that matters, it's how you handle your tool(s).
JohanEkdahl wrote:
Quote:
Wait, are you saying that an ancient piece of obsolete hardware is STILL being used?
If this is true, I'd suspect that the philosophy is: If it works - don't fix it. It is a proven piece of hardware/software.OTOH the Shuttle (or rather "The Orbiter", which is the correct term for that winged piece of capsule, payload bay and engines that does the whole journey and returns, excluding the solid rocket boosters and the external tank) has had a complete cockpit overhaul at least once. I suspect there is more or less nothing in front of the commander and pilot that is the same as in the early eighties.
EDIT/PS: Check this out, in case you've missed it: http://www.nasa.gov/55644main_NA... . It's running maximized on my secondary monitor as I type.
Just out of curiosity what about radiation hardening?
I was under the impression that the Orbiter's primary flight computers were 386's.
I was also under the impression that one of the reasons for not moving to newer chips is the fact that with the smaller architecture of the newer chips there is a greater chance of the (significant) background radiation changing the state of bits.
Unfortunately I do not have any contacts through which to confirm the above.
JC
I've heard similar; also a rumour that the 486 is the last part with a geometry big enough to be cosmic ray tolerant. It is rumoured that NASA trawls eBay for 'new-old-stock' parts.
It's been a few years since I read what follows, and I have no link anymore, but...
Apart from the "proven hardware" argument for any computer going with the orbiter or to Station Alpha there is one more reason that bleeding edge models are not used: It is usually a long period of time from start of preparations to the actual mission, and of-course NASA are not switching the hardware in mid preparations just for the fun of it. So if they are starting preps today for a mission 2011, they would select computers now that are not bleeding edge today, but proven to be reliable. Let's say a single core processor at around 1.5 GHz and 1 GiB of RAM. I bet that in 2011 we'll all laugh at such a machine! So what goes up today was bleeding edge at maybe Y2K, and mainstream at 2004 or so.
This would go for mission-critic computers, and that includes much more than just the flight computers. Eg. the machines controlling the robotic arms failing would ruin the whole on-going shuttle mission. The machines used for eMailing the family is another thing, but you don't need QuadCore 2.5 GHz and 8GiB RAM for that so why risk it.
Remember that Apollo 13 was manouvered homeward, to hit the re-entry window, more or less with a computer with ATtiny-something capabilities, and when that was switched off (to preserve very precios amp-hours) manually with the aid of simple opto-mechanical star-navigation aids IIRC.
Although both the shuttle and space station programs are highly advanced I'm not sure that basic manouvering, flight-control etc is dependent on extremely massive computing power "up there". I'd suspect that keeping the life-sustaining systems, electrical systems and high-bandwidth earth-comms to require much more ooomph(tm) , and some of the scientific experiments to be the most demanding. (Then again, I'm just speculating over a glass of wine...)
Drifting slightly: Now that the ESA module is in place, and the Japanese module eventually being added also, the space station is not mainly a fun building project but a platform to do space sciense. I'ts been a while since I surfed the NASA site extensively, but it seems to me that the planned ability to house 7 (seven) astronauts/scientists on the space station simultaneously is little talked about. Early in the station program I saw at least two different plans for a living qwuarters module. Anyone know if that part has been scrapped, or else how they are going to house that many people on the station?
Sorry for the long post, but it's a really good wine! :D
The 5 computers on board are the original AP-101's with the upgrade. THe kicker is that the software in them is custom written for EACH mission. no two programs are the same at all. The language is unique to the shuttle and is not used anywhere else. Talk about job security!!
I personally think they use the old computers is because of several factors, Heat, cooling etc. that modern processors face. Could you picture the shuttle in deep do-do up there because of a fan failure? Who do they call for tech support?
Jim
Level: Raving Lunatic
Joined: Mon. Dec 15, 2003
Posts: 4313 View posts
Location: Slovakia, Bratislava
Jim: If it drains 600W, the 600W must go somewhere... probadly heat? So, wouldn't it be better if they used some processor that does not eat 600W?
Daqq
I hear where you are comming from believe me.
At the same time the shuttle project is nearing the end of it's life so why bother as well?
Jim
Heh... I just picked up January's Circuit Cellar and it had someone from NASA writing how to radiation harden circuitry.
Quote:
Jim: If it drains 600W, the 600W must go somewhere... probadly heat? So, wouldn't it be better if they used some processor that does not eat 600W?
Think of it this way. That's 600 watts of power the heaters don't have to provide.:)
Adam wrote:
Think of it this way. That's 600 watts of power the heaters don't have to provide.
I'm not sure they actually need that heat.
I have no figures as to how much heat they need to "vent" off the orbiter, but the inside of the payload bay doors have radiators to dissipate excess heat. AFAIK this is one reason they open the payload bay doors, even on missions when they don't need too for eg. moving things out of the payload bay or dock to the space station.
(On the space station truss, apart from solar cells for electrical power, there are (4?) large radiators to disspate excess heat.)
Interestingly, the lunar module computer was built by ttl chips which were connected by wire-wrap, is it's a more mechanically robust connection which provides a higher pressure on the chip pins. Fascinating.