One of the most complicated machines ever built was the US space shuttle (technically, the STS or Space Transportation System). Despite the title, we doubt anyone is going to duplicate it. However, one of the most interesting things about the shuttle’s avionics — the electronics that operate the machine — is that being a government project there is a ridiculous amount of material available about how it works. NASA has a page that gathers up a description of the vehicle’s avionics. If you are more interested in the actual rocket science, just back up a few levels.
We will warn you, though, that if you’ve never worked on space hardware, some of the design choices will seem strange. There are two reasons for that. First, the environment is very strange. You have to deal with high acceleration, shock, vibration, and radiation, among other things. The other reason is that the amount of time between design and deployment is so long due to testing and just plain red tape that you will almost certainly be deploying with technology that is nearly out of date if not obsolete.
A good example is the orbiter’s GPC or General Purpose Computers. There are five of them — at least three have to agree before they’ll do anything. Early versions of the GPC used magnetic core memory. That technology was old even when it was put in the design but it was the best way during the design phase to ensure the memory would not be upset by radiation effects. In 1991, a major upgrade did replace the core with semiconductor-based memory.
The information on the NASA site is a bit high-level but still detailed. If you want some real hardcore discussions about the shuttle’s avionics, the KLabs site lost NASA funding before moving to a private web server to remain operational. If you ever wanted to try your hand at programming in HAL/S as the shuttle programmers did, you can get the documentation there. There’s also a booklet with a lot of information that you can download from NASA and a video overview you might enjoy, below.
While we doubt anyone will be trying to build their own shuttle, this is still a great wealth of information. Of course, the Russians did try, back in the day and arguably made a better system. If you do build a copy, there’s already a shop manual from Haynes.
I recall when they first started bringing notebook computers into space, but I never heard how they performed.
As far as obsolete tech goes, some newspapers were still using paper tape up until they replaced the linotypes entirely. I would not be surprised if some manufacturing still used paper tape. It is paid for, if you only have to load something in once in a while it is not awful, and it is robust, even on a shop floor. Old tech is not bad. In a lot of cases it is quite good. It is proven. There is a lot to be said for having as much proven technology under you as possible when you have 2.2 million pounds of fuel in the gas tank and your starter is s fuse.
Backup a bit here. The Linotype machine you’re describing was in fact their last big one before sitting down and designing phototype setting machines. The originals of which there was one main design with several orderable features concerning the pot for the metal was actually one such. And the keyboard was a special non-orderable design outright.
And the hardware that the shuttle used for their computers, take a look at what ran the AWACS birds for example and even the Gemini space craft.
IBM designed all of that. It’s all based on the IBM S/370 mainframe.
And HAL/S is related to the languages that include PL/M for example.
And here’s where it gets interesting: Papertape was used by the printing market for the photo machines almost constantly until recently. Even after the machines got their jobs via the computer itself, the papertape was used to configure the machine for fonts and some special features.
Yes Al I’ve been there and got the T-Shirt someplace.
Paper (or rather: cardboard) tape is still being used by looms all over the world.
Notebook computers in space worked great, and continue to work great. The Shuttle PGSCs ran important rendezvous tools and interfaced with the orbiter’s rendezvous radar and LIDAR (TCS), among other tasks. The Space Station has kept up the tradition with the SSCs (Space Station Computers), which recently got upgraded to HP ZBooks. (I happen to develop software for them.)
I’m really hoping there’ll be a shuttle simulation someday that includes the portable computers. I *think* there are full disk images archived for the shuttles on a per-flight basis (though I know for a fact the ancestor of the software I work on got updated on the pad once, with one of my friends standing in the erected orbiter cockpit, hoping not to step on anything untoward…)
SpaceX has made a very interesting shift from this Shuttle-Era paradigm of using older known reliable tech in that they use relatively modern Consumer-Grade Multi-GHz Multi-Core Intel x86 processors in their avionics.
In order to offset the potential reliability issues they leverage the lower cost, lower size, lower power advantage gained by running the x86 processor in an odd mode where each core runs the same code at the same time, with a second physical CPU on the same board also running the same code, with multiple boards also running the same code.
The Falcon Heavy / Roadster Upper Stage deliberately lingered in the Van Allen belts far longer than necessary in order to validate to NASA/USAF that the approach of massively-parallel consumer grade electronics would work in such a harsh environment.
“SpaceX has made a very interesting shift from this Shuttle-Era paradigm of using older known reliable tech in that they use relatively modern Consumer-Grade Multi-GHz Multi-Core Intel x86 processors in their avionics.”
Google and others (cloud) kind of fostered the idea that cheap, and large quantity, would be better than few, expensive. If at first one doesn’t succeed, throw more hardware at the problem.
It’s not a new idea. I worked on a telco switch in the 80s that used a pair of 8086 processors running the same code, from different memory devices. One talked to the rest of the hardware and all the outputs from both were connected to comparators. If there was any mismatch in the signals coming from the two processors, control was switched to an identical board in a hot standby configuration.
I wasn’t there for the original design, but this was one of the first applications of x86, predating the IBM PC.
Why do they need all that massive cpu horsepower on a simple booster?
Those boosters have more CPU capability than a squadron of B1-Bs or a couple B-2’s. Hell you could power the entire Shuttle fleet with one of them.
My understanding is modern booster tech is Not at all simple by any means, to suggest a small incomplete set; multiple feedback paths with combinatorial complexity in respect of control strategies ie many variables not just one dimensional flow rate and pressure, instrumentation channels, influence of coding approaches with potential for bloat in that case a risk assessment issue re time taken to make changes in relation to audit trail in implementing changes and fall back positions – in that respect I imagine the fall backs complex far beyond simple state machine exploration. In any case the differential between current vs lower capacity CPUs could sacrifice augmentation potential for more complex strategies. Also consider supplier’s role in approving product for that highlg certified environment ie. They have the tech and commercial depth – perhaps choosing lower capacity makes that commercial tactic more problematic. Booster’s as a system not simple especially when reliability and controls necessarily advanced. I’m sure others can offer more…
Why wouldn’t you ? Common x86 cores are extremely well tested and have a wide range of mature development tools.
“Those boosters have more CPU capability than a squadron of B1-Bs or a couple B-2’s”
I suspect that if B1s or B2s were autonomously flying missions that included accelerating past mach 10 and landing vertically, they might also have significantly more CPU power than they do now.
+1
Indeed, I forgot that key aspect and it aint trivial with all the controls and levels of comms too…
Having worked on this problem for the competition, we always were afraid of SWaP (size weight and power) implications. Sure everything is getting smaller but if I need 20 of them… the real problem with the other approach is when you wind up having go commission your own 80386 runs because the NOS market for old parts have dried up and you need 5 more.
Is there any page with further details about SpaceX’s avionic hardware?
Only what Space-X chooses to release, and since they are a private company, I am willing to bet that it is in the range of fuck-all (Plus / minus 0.01 fucks).
This is a complaint I’ve had over the decision for NASA to contract with SpaceX for launches. Yes, it is cheaper, but the technologies and patents are now owned by SpaceX, rather than NASA. So if NASA wants a vertical-landing rocket, they’ll either have to pay SpaceX *or* they’d have to pay to re-develop the technology so as to not infringe on SpaceX’s patents. Before, they would own the technology and can use whomever they wanted to build it (typically whichever company was the most cost effective).
Patents expire after 20 years or less. Given the long time that it takes to develop something like that, I would imagine that the patents would expire before the new hardware is ready.
SpaceX is using tons of decades old NASA ideas that no one bothered with. Also they use a lot of COTS items. They were just willing to try going in a design and philosophy direction no one else though would pay off like it has. Doing a quick search indicates SpaceX is NOT patenting anything really. https://www.quora.com/Has-Space-X-filed-any-patents-in-the-USA-or-elsewhere-If-so-what-are-they
Copenhagen Suborbitals use their own ruggedised Arduino boards to control their rockets. Although they only use individual boards for each task, but their rockets are currently small and un-manned.
Actually I think they might actually be AMD CPUs and only on Dragon for high level processing of sensor data for low level control is done by various mostly PPC based micro controllers . Falcon 9 itself uses Power PC based controllers from the commercial aviation and automotive industry.
I’m surprised they didn’t use an nVidia Tegra based system from Tesla for the high level control on Dragon.
Ha! There is a whole series of Haynes ‘manuals’ for everything from the SR-71 to the Mars rovers. Having spent my fair share of time under the hood with many of those, I can’t help but crack a big smile.
Bet they still refer you to the dealer for the simplest of anti-lock brake problems though.
The older ones didn’t, I have a Haynes manual for 1980-1984 Subarus, that was printed in I think 1989, and the automatic transmission section covers changing the fluid, filter, and valve body, but refers everything else to a qualified mechanic, the manual transmission section basically shows you everything down to rebuilding it. I have a later print of the same book and the manual transmission section is fluid change and troubleshooting, and says everything else is “beyond the scope of the home mechanic”. The older one even shows you how to rebuild the alternator and the starter, which the newer book says to just replace.
I know, older manuals were great, newer ones seem to be basic service that’s it. Though they go into great detail about engine rebuilding still which I doubt most home mechanics would bother with even if they own a hoist, rather than get a decent wrecker motor with 3 month warranty for a few hundred. So I’d prefer they stuck all the engine rebuilds in separate books, like chrysler V6es or etc, instead of reprinting that wad in 5 or 6 different model editions. Also since about 2000 up, I’m coming across many errors and bad advice in them, that makes me doubt that they have actually done the job they described.
Chilton’s were better!
B^)
(IMO, it was a sad day when Haynes bought Chilton’s)
For things such as automatic transmissions you have to buy another manual that covers just that subject.
I actually have not purchased one of these since about the time ‘The Internet’ started getting popular. Before that they usually told me exactly what I needed to know, or at least gave me a reference to try to correct my own mistakes. That information was rather hard to come by otherwise. They probably saved me a couple thousand dollars back when my time was much less valuable than my money was..
I usually had the Chiltons also so I could compare and try to spot mistakes, so my memory could be overlapping a bit between the two references. I know I’ve been able to do complete rebuilds using nothing else though on vehicles up until about 1993.
Hello, I’m interested in building my own orbiter :) Where can I find documentation visible on film – with all these schematics and pictures? This linked NASA site seems text-only.
Buy them from the Government Printing Office like the Soviets did when they were designing Buran.
Hmm, one wonders if more advanced tech with tighter engineering management might do better and with much lower cost too, these days with capital at lower utility cost – what obstacles and comparative issues with shuttle reduce the chance this could succeed…
https://www.youtube.com/watch?v=mOqInOe5g8k
I left a comment on that YouTube thread a few mins ago about saving takeoff fuel with a suitable acceleration bogey…
Being propelled along the surface of the Earth in a tube full of rocket fuel without guaranteed thrust sounds like a good way to have a bad day.
Why ?
Because if it doesn’t fire reliably for a certain period you kill the engines and use the bogey to brake. Whereas without bogey and it falters before takeoff you only have the craft’s on-board landing gear brakes – which eveil would you choose if your partner’s life was on the line And which scenario has the greatest survivability value; The craft with less or more fuel available one aloft for a necessary go-around – that is – if you exceed runway length whilst having some altitude- bearing in mind this is a specified runway length for the craft with a healthy risk-assessment margin for this particular scenario…
As well as that you have the easy potential to fir unmanned smaller versions via bogey and autonomous without risking pilots which still offers advantage of more fuel on board…
And if the acceleration bogey has a human controlling it…
Will they be known as “The Bogey Man”?
(rimshot)
If the UK government had some sense they would fund that as it would leap frog right over the vehicles Spacex and Blue Origin have in the works putting the UK as a leader in space flight.
A HOTL has a lot of safety advantages over a VTOL such as it won’t fall out of the air if thrust dropped below 1 to 1, you can dead stick land it , and do rejected take off like an airliner.
Skylon probably actually could be used for suborbital point to point or launch payload into space daily.
“we doubt anyone is going to duplicate it”: The Russians did…
Just saw you linked in the russian attempt at the bottom *facepalm*
Wake me when someone duplicates the Saturn V. It carried real people to another world! The space shuttle was only really designed for carrying stuff into orbit for industry and the military. The military mostly lost interest once they realized how much easier it is to just lob their missiles at people using plain rockets. Unfortunately this was not before they managed to force redesigns on the shuttle that pretty much ensured it would be impractical as a tool of the manned space program.
Even the shuttle’s own pilots called it the flying brick and 40 years later our people have visited exactly 0 other worlds. That is what a shuttle is really good for, killing the dreams of a whole species.
I think you need a stargate if you’re going to be visiting other worlds.
+1
Rockets won’t take us anywhere worth visiting.
depends. wait another 3 decades. the moon might seem a veritable paradise
That reminds me of the Futurama episode where Fry finally got to go to The Moon, and found out it had become a cheesy amusement park!
As I recall the Shuttle carried seven astronauts compared to Apollo’s three. The shuttle spacecraft launched many times Apollos only once each. The shuttle launched most of the international space station and many important sats such as the Galileo Telescope. So in a way the space shuttle’s legacy is still flying.
The orbiter and it’s landing mode was not actually all that dangerous.
What was dangerous was the vehicle stack interaction and having the orbiter in the path of debris from the launch stack and having SRBs on that stack.
What was impressive about the Buran was that it was basically a radio-controlled vehicle that went into orbit and then landed safely without any humans onboard. There are web sites that show what has become if it and it is sad…rotting away in a hangar.
I “think” the Shuttle was capable of unmanned flight, but the astronauts have a pretty strong union.
B^)
Funny actually, in capitalist running dog America the “union” wouldn’t let them get away without a pilot, but in socialist workers “paradise” they ran it remote controlled. LOL