The Argon-16 flight computer. Real computers will kill you if they fall on your head
The Sep 2010 issue of IEEE Spectrum has a nice example of how determination beats technology. The article is “A Digital Soyuz” by James Oberg, and discusses how the Russians have upgraded their main manned spacecraft. They’ve replaced the main flight computer, the Argon-16, and five analog monitoring and telemetry systems with a single digital system, the TsVM-101. The Argon-16 has been in use for over 35 years (!). It contains 6 KB of RAM and can do 200 K adds/second. That’s K, not M or G. It weighs 70 kg, but oddly enough draws about as much power as a PC, 280 W.
“Backwards Russians,” you might think. “No wonder they lost their empire.” I take away the opposite message. They built something that worked and then stopped screwing with it. This machine has triple redundancy – three copies of every subsystem that vote as to who’s correct. It uses rad-hard TTL circuits. It has never failed in flight.
As of March 2010, the Soyuz series has had 104 flights (compared to 114 for the Shuttle), and has lifted 276 people into space (723 for the Shuttle). Its last failure was in 1971 on Soyuz-11 when the crew died on re-entry from loss of pressure. The Shuttle’s last failure, of course, was when Columbia broke up on re-entry in 2003. So that’s 39 years without a failure for Soyuz, and 7 for the Shuttle. The Shuttle only has a couple of flights left, so Soyuz will be the only way to get to the International Space Station for some time.
So why are they upgrading Soyuz now? Several reasons:
- The new computer is lighter than the Argon-16 and the analog systems, increasing cargo capacity by 70 kg.
- The analog systems were harder to calibrate and test, which made it take longer to prep the craft. It’ll be flying more often when the Shuttle stops.
- It only needs one pilot instead of two, so two non-pilot mission specialists can go up instead of one.
The only upcoming contender for US manned launches is the Dragon spacecraft from SpaceX Corp. It’s due to have its first launch this year, but it’ll be unmanned. SpaceX has had 3 failures in 6 launches total. The parachutes did not open on reentry for first Falcon 9 launch, but they called it a success anyway. That kind of spin is not encouraging. Perhaps the hyper-tech founder of SpaceX, Elon Musk, could learn a thing about reliability from the stodgy Russians.
[Juicy gossip addendum – Musk divorced his wife Justine in 2008, leaving her with 5 kids. He then married a much younger actress, Talulah Riley. He’s 39 and she’s 25, which just fails under the XKCD creepiness rule – you shouldn’t date anyone younger than your age divided by 2 + 7. He and Justine are wrangling in court, of course, but one thing she wants is a Tesla Roadster, glacier blue.]
A Niche in the Library of Babel 
Somewhat disingenuous. The Space Shuttle flies an upgrade of the venerable IBM AP-101 radiation hardened digital computer, which was designed in 1966. It has a whole megabyte of ferrite core memory, and is good for .82 MIPS.
And, of course, no Shuttle airframe has ever been lost due to a computer malfunction.
This post was factually incorrect to the point that I am now considering unsubscribing.
Good point about the Shuttle computer being extremely old technology as well. I wonder if these are the last operating machines to use core memory. There might still be air traffic control machines that are of that vintage, perhaps in Russia. Technology moves slowly when lives are at stake!
I looked into the Shuttle General Purpose Computer a little more. The first machine in this line, the IBM AP-1, was announced in 1966, but the machine on the Shuttle, the AP-101b, came out in 1972. It was picked because of its compatibility with the IBM 360, and its software tool set. It did use ferrite core memory, but only ran at 0.4 MIPS, not 0.8. It was replaced in 1991 by the AP-101s, which replaced the core with 64Kx1 SRAM. There’s an interesting study at http://klabs.org/DEI/Processor/shuttle/oneill_94.pdf which looked at the error rates in this SRAM due to radiation. Sometimes a particle would actually drill through two separate chips! Fortunately, their ECC has corrected all detected errors. The machine’s biggest failing was lack of address space; it was initially limited to 64K 32b words, later extended to 256K by the use of segment registers. It was a struggle to make the code fit.
Anyway, it looks like core memory isn’t even used on the Shuttle any more.
With regards to the SpaceX falcon 9.
It made it to orbit on the first try. Do you know how many rockets do that on the first try? Not many. It was a test flight so give it some credit.
Pingback: Russians Throw at the Basket From Mid-Court « A Niche in the Library of Babel
A Soyuz spacecraft needs to have a crew consists of one pilot, one flight engineer as minimum (not two pilots). It’s not true that the new Soyuz to be flown by a single professional cosmonaut despite of the new avionics upgrade. Regarding the methods of Soyuz training, there are no huge differences between TMA and TMA-M simulators. Always a space mission will need two active crew members as minimum and the role of flight engineer is very important! Look here that cosmonaut Aleksandr Kaleri said:”Mainly new changes took place in avionics. So we will have new on-board computer system, we will have a control and informational bus joining different computers in the complex on-board, the computer complex. So we will have some more redundancies in this case. We will have a new telemetry system. So it’s mainly in avionics and that is the new crew interfaces, like displays, like some information on displays, like some lights, some switches and, so it’s an interesting version and I think that the improvement of Soyuz will get us new possibilities to improve it more and more, because the main issue is that now we will have an open architecture of the computer system on board, The TMA version, like in the older versions, we had the closed architecture; now we will have an open architecture and we can improve and improve it. “(http://www.nasa.gov/mission_pages/station/expeditions/expedition25/kaleri_interview.html ) and also Oleg Skripochka :”The main modification is that they have installed the new on-board computer, therefore we have new software. And the new motion control system for the vehicle, so there are more navigational capabilities, more control capabilities. The vehicle will continue supplying data throughout the entire flight: we will know where the vehicle’s located, etc. There are many things that I can tell you but the main idea here is that the motion control system of the vehicle has been changed. ” and is asked on “Do these changes make it easier for the crew members to fly, or make it more robust in its performance? ” he responds :”I would say both. We have a new interface which makes it easier to perform the informational exchange with the computer. We will be receiving more complete information from the computer regarding the on-board systems, and that will make the process of controlling the vehicle and approaching the station more simple for the crew, I hope. “
Thanks for the info! I hope that the recent crash had nothing to do with all these avionics changes.
Absolutely! This new computer is not used for the launch vehicle guiding, the launch vehicle has a proper avionics witch has no connection with Soyuz spacecraft or Progress spacecraft. During launch the spacecraft avionics is not operating because all the the flight procedures is done only by the launch vehicle avionics. After the detaching of the launch vehicle third stage the spacecraft on-board computer enter in operation. About the TsVM-101 computer, it’s true that is more advanced than Argon 16 because is more rapid and ergonomic but that doesn’t mean to eliminate the flight engineer. It’ true that the spacecraft may be piloted by only the commander, or as an unmanned vehicle like progress spacecraft. Otherwise, in terms of flight operations, there are no differences between these spacecrafts, but with crew on board mission runs much better and that is much easy for the flight controllers. During the spaceflight history have occurred such situations, for example: in 1979 Soyuz 34 flew automated to take back the resident crew from the Salyut 6 space station because of a malfunction of Soyuz 33; the crew of Soyuz TM-6 (returned with TM-5)consisted of one professional and two non-professional and also a similar case was Soyuz TM-13 visiting crew; Soyuz TMA-1 was considered an automated spacecraft with crew on board (all three crew members was professionals)<>. In conclusion is not something new that a Soyuz spacecraft can be piloted by only one professional cosmonaut (or without crew) but the participation of two professional cosmonauts during the flight operations is a great plus,especially when it comes to rendezvous and docking procedures(very critical tasks). Most of emergency situations was resolved because on board the spacecraft was two professional cosmonauts. To make an analogy, we take a case of the Airbus A380 airliner. It has a very advanced avionics but this did not lead to elimination of the second pilot.
I wish you all the best!
P.S. Please send me your e-mail address on cezar_enache@yahoo.com to send you some documents about russian space programme!
Do you think the manned Soyuz can get down to not needing a pilot at all? The cargo flights don’t seem to need them. As it is, it can only take up one mission specialist at a time. It would be easier to keep the ISS staffed if the passenger capacity of the Soyuz was higher.
Yes, to get down the Soyuz, re-entry data can be uploaded into the computer by the Moscow Mission Control via command radio link. Nominally, even with the crew on board, these data are introduced by flight controllers from mission control. Both mission control and the crew monitor the data entering procedure. The crew is authorized to change the setting data only on the mission control instruction. Also the crew can enters these data without any instruction but this is a rule and it not be violated. In 1979 Soyuz 32 capsule landed without crew. About cargo, early 90’s Progress was equipped with Raduga capsule that was used for returning materials from space station Mir. Unlike Soyuz capsule, used ballistic return (not semi-ballistic). Nevertheless, a spaceflight with a complete crew (2 professionals as minimum) is the preferred choice and chances that the mission to be accomplished are much higher. About a passenger higher capacity spacecraft I know anything. To learn more, please send me you e-mail address on cezar_enache@yahoo.com to send to you some documents about that!
I wish you all the best!
Pingback: The Oldest Active Computer « A Niche in the Library of Babel