Slashdot Mirror
Orion Capsule Safely Recovered, Complete With 12-Year-Old Computer Guts
Lucas123 writes While NASA's Orion spacecraft, which blasted off on a successful test flight today, may be preparing for a first-of-its-kind mission to carry astronauts to Mars and other deep-space missions, the technology inside of it is no where near leading edge. In fact, its computers and its processors are 12 years old — making them ancient in tech years. The spacecraft, according to one NASA engineer, is built to be rugged and reliable in the face of G forces, massive amounts of radiation and the other rigors of space."Compared to the [Intel] Core i5 in your laptop, it's much slower — much less powerful. It's probably not any faster than your smartphone," Matt Lemke, NASA's deputy manager for Orion's avionics, power and software team, told Computerworld. Lemke said the spacecraft was built to be rugged and reliable — not necessarily smart. That's why there are two flight computers. Orion's main computer was built by Honeywell as a flight computer originally for Boeing's 787 jet airliner.
Not only was the launch itself successful, but the sensor-laden craft's splashdown was smooth ("bulls-eye," as NASA puts it), and NASA has now recovered the capsule. ABC News has some good photos, too.
I get sick to my stomach when I hear Ruby and JavaScript weenies go on and on about how they're "engineers".
No, you shitheads, you aren't "engineers". The people who work on Orion are engineers. Some high school dropout writing web apps in Ruby is not an engineer in any way!
The clock speed is the same but I guarantee even the lowest end AMD would destroy anything 12 years old in terms of work.
Only the State obtains its revenue by coercion. - Murray Rothbard
... just feels kinda weird. This is basically a scaled down repeat of an Apollo test mission done nearly 50 years ago. At least then the Saturn V launch rocket was being tested as well.
The more exciting mission comes later month with SpaceX attempting a powered soft landing of a first stage on a mission delivering cargo to orbit. Small chance of success on the first attempt. But if successful, that will be something never seen before and once thought to be impractical, if not impossible. It will also be a major step in greatly reducing the cost for access to space and something much more liable to impact the lives of everyday people.
In fact, its computers and its processors are 12 years old
They word it like NASA is dumpster diving for its flight computers these days. The CPU may be from what was new 12 years ago, but I seriously doubt the physical unit is actually 12 years old.
It's also hardened against radiation. I would be willing to bet that any processor in these systems will still be functional long after most newfangled home CPUs are long dead. These flight computers will be remain functional in an extremely harsh environment longer than any home CPU would last. Even with how pampered home processors are in comparison.
If you took your i5 into a high intensity radiation environment like space it would be more likely to have single event upsets whereas the processors that most space applications are hi-rel (hi reliability) and have been tested against radiation. A lot of the chips used in space are also built on silicon and in chip packages that are designed for these reasons. Guess what? If you are a chip designer and you want to build a radiation hardened chip, you usually don't get your hands on the latest designs and you don't get to fab a new version every 6 months. There are people still using 8051 chips that are 20 years old because they 1) Have been used before (really good if your spacecraft parts already have a history of working in space) 2) Have software already written for them from the last project (code that has worked before is good too). 3) can't easily find another part. On a cubesat mission that I helped design we did use a commercial chip that was not rad-hard because we were in a lower earth orbit with less radiation, although the spacecraft does lock up now and again. We almost went with an 8051, we used an FPGA for some of the critical stuff which are less susceptible in some ways to the spacetime environment.
Yeah IBM, It's still common to use this specific PPC core in some of their Custom Foundry parts, the Wii-U's processor also made by IBM is based on the 750FX.. Well it will soon to be Global Foundries making these as they're acquiring IBM's semiconductor business and all IP.. I currently work IBM's test and development, hopefully will have a job with Global Foundries.
12 year old software? No way. We need to fix that. There's no way we're going to Mars without rounded corners, infinite scrolling,and a tiled UI. If we don't launch in beta, all the other countries will think we're not hip. We won't get seated on the Trilby committee at the UN. Get some interns and fresh grads on this project, pronto.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
BAE is still making the RAD750. I worked with the predecessor that is in the RAD6000 computer board.
I recall that the CPU in my first computer (an RCA VIP, with an 1802 processor) was still being used in satellites and such years later. Why? The processor was fully static CMOS, could be run at extremely low power (as long as speed wasn't an issue), and was more tolerant of radiation. But I guess I'm showing my age ...
An 8086/8088 would be enough to get this thing into orbit.
I used to work in avionics. We never used anything as powerful as an 8086. It was all 8051s. They are rad-hard, can withstand lots of voltage jitter, and the logic has already been verified down to the gate and transistor level. The 8051 has certified compilers, assemblers, and linkers, that have been formally verified. They are also dual sourced, which is usually a requirement.
Looks like they are using an RTOS that is commonly used.
http://www.ghs.com/customers/n...
Pretty cool system
Cooler, yes, Faster, no. Clock speed compared to process size has NOTHING to do with "fast" vs "slow".
Process may allow higher clock speed, as well as many other advantages (fitting multiple cores, larger caches, etc on one die) - but without any other innovation the SAME architecture at the SAME clock speed with ANY process size will give you the same performance....
The CGI on the Control Room board was a "display" of what the craft was doing at the time. When the craft fired the thruster the CGI showed it. It was used on parts of the mission where there was no live feed from the craft to see what was going on. In some NASA TV feeds the CGI craft was a little out of sync with the real one; but you chould see/hear that it was doing the same thing as the craft.
GO NASA!
I noticed that Intersil still makes a rad-hard variant of the awful RCA 1802. (you know, the CPU in a COSMAC ELF).
When I saw that, I figured NASA and or the DoD probably give them enough money to make it worth their while... so they must use that antique for something.
Sent from my PDP-11
Having programmed the 8048 and 8051 in assembly language I can appreciate the tech. However I think the 12-year old technology label is probably referring to something like the RAD750. Its roughly a hardened PowerPC G3 at 200 Mhz, sort of comparable to what was in the original iMac. I think the RAD750 was used on some of the Mars missions.
Exactly.
Rocket science ain't ... uh... never mind. But it doesn't take a lot of computing power to navigate. But what it takes is computers that can withstand the stress. Extreme acceleration, radiation, possibly temperature changes, unreliable power supply and so on. When you only need one percent of the processing power of a modern CPU, you don't care about having only 10% of the current CPU power available. But it is very comforting to hear "works from 3-5V, logic accepts up to 6V on its I/Os without damage" instead of "if you're off by 0.2V, unpredictable behaviour might occur, be off by 0.5 and it's going poof".
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
My guess is that they have a truckload in storage already made. It would not make sense for them to not make them available for sale in exotic applications. It's a proven design being (that can use a minimum of other expensive rad hardened parts) used in other proven designs so they can pull them off the shelf and have something ready to fly quickly. As parent poster noted, for many applications 64 bits can be overkill. They could also being used for repair for things like military aircraft that used them in their manufacture in that era and are still flying.
Take the cheese to sickbay, the doctor should see it as soon as possible - B'Elanna Torres, "Learning Curve"
The MPC565 is pretty standard in Airospace. Has all the features you need and not more:
* Clock: in the low MHz range. Pretty easy to make transmission reliable, even if a PCB trace is damaged or the board deteriorates.
* No MMU: Why the hell would i put a MMU in a Controller which should perform identical operations over 5years-40years and has no additional unplanned tasks, and is running software which is somewhere between well tested (level D) and insane (level A). The complexity of a MMU is incompatible with ceritfying this thing as level A (critical) for any reasonable price.
* big SRAM on chip. Buffer the voltage to the processor well and it does not matter to you if the clock fluctuates wildly.
* Flash on chip. (for program storage). So you can be pretty sure that as long as your program runs, it will run well.
That being said it should be mentioned that a variant of TFTP (35years old) is the standard for Loading SW onto parts in Planes.
My home server is an old blue and white Apple G3 running Yellow Dog.
Sure I've added a Sata card and a GB Ethernet a while back and it's got 6 TB in a software raid.
But it's a server, why would a server ever need more processor, all it does is read/write stuff between Ethernet and disk. I'll keep running it until it finally dies.
Well, something like 8k of ROM and 2k of RAM was enough to go to the moon, land, take off, and come back, so...
That's what people have the disconnect on. Flight control software isn't stressing. It's maybe a dozen or two 6x6 matrix-vector operations which unroll into maybe a few hundred FLOPS (or they could be fixed point) that need to run maybe at 20 or 30 Hz (Apollo's major cycle was 10 Hz). This is stuff you could do with hand-wired 7400 IC's if you really wanted to (in fact they did the equivalent for the first submarine launched ballistic missiles in the 50's). Having a programmable computer that's fast enough to do it a few hundred times a second, and handle the control loops for some of the other stuff in the capsule is nice, but it isn't hard with a 10 MIPS processor, let alone the 200+MIPS they're flying in ORION.
In the 60's when they went to the moon, it was hard because there was no such thing as an off-the-shelf space-qualified programmable flight computer, so they had to invent it all from scratch, and there's this mistique that developed around it. But even by the 80's and 90's, the space hardware and avionics industries advanced to the point where the hard stuff was knowing what software to write, not finding a computer and inventing a compiler to run it on.
Which were built using 2800 integrated circuits interconnected by wire-wrapping, every single one of those ICs containing exactly two NOR gates.
What's awesome about that is it's like something you'd see at a maker fair today,except that hardly anyone knows about wire wrapping these days. Too bad; it's faster, more reliable and more repairable than soldering.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
Truth be told: I work at a *ahem!* major microprocessor manufacturer, and even the cutting-edge SoC's we're working on right now, that are not even close to being at a stepping ready to be released into the market, still have an embedded 8051 in them. Being myself old enough to have had a CDP1802-based computer, built entirely by hand on perfboard when I was back in high school, I just have to laugh at anyone who actually believes that you have to have multi-core, multi-gigahertz-clocked processors with gigabytes of memory for any given application. I laugh even harder at kids who think you have to have at least a microcontroller, if not a Raspberry Pi board, to make an LED blink on and off; they think I'm trolling them when I tell them that two bipolar transistors and a few passive components will do the same job.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
There are only a handful of space qualified microprocessors available. Most of them were designed 20+ years ago. In fact this is the case for most space qualified ICs of all types. Nothing that goes into space with the expectation of high reliability uses modern high speed circuitry because smaller features result in greater error rates and a shorter operational lifetime due to radiation effects. It is also cost prohibitive to develop a modern fab line to manufacture space grade parts so the industry is mostly stuck in the past using older designs largely due to reliability requirements.
The Java set despairs that they can't play in their perfect abstraction of a machine without gobs of memory and CPU cycles to blow away. People who know how to program bare metal can get by perfectly fine on a "slow" memory constrained device.
I am becoming gerund, destroyer of verbs.
There are two approaches to radiation tolerant computing. One is to make the processors hardened to radiation. These processors are usually slower, and use an architecture with fewer knowledgable computer programmers. This seems to be the approach on Orion.
In contrast, the Space X Dragon Capsule uses multiple processors operating simultaneously to create a fault tolerant system. To quote:
Dragon uses a "radiation-tolerant" design in the electronic hardware and software that make up its flight computers. The system uses three pairs of computers, each constantly checking on the others, to instantiate a fault-tolerant design. In the event of a radiation upset or soft error, one of the computer pairs will perform a soft reboot.[45] Including the six computers that make up the main flight computers, Dragon employs a total of 18 triple-processor computers.[45]
An advantage of this is that the processors are far faster. There are also many more trained programmers available for these more current architectures. Such systems arguably have similar (or better?) radiation tolerance to the older hardened processors.
This and no other is the root from which a tyrant springs; when first he appears as a protector - Plato (423 to 327 BC)
Hey... you insensitive clod...
One of the first computers I built (actually the 3rd) was a hand wire-wrapped 1802. Cutting edge CMOS for it's day.
Back then, it was anything but "awful".
Get off my lawn!
I don't read your sig. Why are you reading mine?
Don't laugh. NEVER laugh. You'd be bitching if they were out in the streets in gangs, stealing your stuff, vandalizing your property, or any of that other shit old people complain about. Instead, you have a youth who has a desire to learn and invent something which will hopefully lead them into being a productive member of society instead of serving 20 to life for cooking meth. Young folks, especially in the high school age range, are easily discouraged and you ridiculing them about being too hipster or whatever will only alienate them. When faced with this situation, I will show them how to blink that LED with their Arduino, then show them how to blink it over a USB port with their gaming rig, then show them how to blink it with a mechanical cam switch (old points distributors work great for that fyi). If I've got some lying around, I'll show them how to blink it with some vacuum tubes! What you have to do is find out why they chose to do it the way the are doing it (typically because that's all they know) and then show them all the choices available, why you'd use each one in a different application, and show them why they get to use that Arduino now instead of wire wrapping an 1802 on perfboard. The key to all of this, I've found, is maintaining the balance between lengthy enough to get the concept across but short enough to keep them from picking up their phone and tweeting their facepage.
/. is always posting stories about how we need more STEM graduates, more hardware hackers, and more programmers. A lot of folks here agree with that sentiment and perhaps you've said as much in the past, so please, don't ever laugh at them. They're sensitive.
Besides, grandpa, you should be happy the hipster Makers are doing what they do. Thanks to steampunk, vacuum tubes(especially nixies) are making a huge come back. I have a USB vacuum tube audio amp similar to this. You couldn't buy that shit when we were kids. You had that 100lb behemoth amp that made the house lights dim when the bass hit, and kept your room 80deg during the coldest of winters. And that was if you had a good bit of money. If you had a little money you might be able to buy a Heathkit. Otherwise, like me, you cobbled together some barely functional and noisy bullshit from an old guitar amp and a half working tube powered CB radio. Now you can get something handheld portable for a hundred bucks with the further satisfaction that your "dying" craft is actually living on and they'll need people like you to teach the new gen.
The overall point is, no matter how they get to the destination, what really matters is that they're taking the journey. Time itself will teach them when they need to blink that LED with an RPi or if they need to use a couple transistors.
DISCLAIMER: I am dense at times. If "laugh at" was just a figure of speech, please don't take the post personally- perhaps someone else can be inspired.
I built a wire-wrapped computer around an 1802. 2k of memory. Lots of fun. One advantage was that there was a chip that mapped the available memory to a video signal so it was very easy to get a TV display going. And it didn't require any particular clock frequency so you could actually single-step it very easily. I don't recall much about the instruction set, but I don't think it was particularly awful.
I learned wire-wrapping about 12 years ago in university courses. Prior to that, I would have assumed it was obsolete tech, but there were some very wise and experienced professors. It is much faster and cleaner than soldering. I still use it when debugging boards and I need a quick test lead attached to a pin header. Solder would ruin the connector, but wire-wrap is removable.
You are not a "real" programmer until you've done serial I/O by flipping an output toggle up and down by hand on the front panel! ;-)