Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!tut.cis.ohio-state.edu!mailrus!shadooby!accuvax.nwu.edu!delta.eecs.nwu.edu!phil From: phil@delta.eecs.nwu.edu (Bill LeFebvre) Newsgroups: sci.space.shuttle Subject: Re: space news from May 15 AW&ST (replacing computers) Message-ID: <819@accuvax.nwu.edu> Date: 3 Jul 89 18:03:11 GMT References: <1989Jul2.054432.5054@utzoo.uucp> <3827@phri.UUCP> <1989Jul2.210944.15387@utzoo.uucp> Sender: news@accuvax.nwu.edu Reply-To: phil@delta.eecs.nwu.edu (Bill LeFebvre) Organization: Northwestern U, Evanston IL, USA Lines: 54 [ Note: I've removed sci.space from the Newsgroups. ] In article <1989Jul2.210944.15387@utzoo.uucp> henry@utzoo.uucp (Henry Spencer) writes: >In article <3827@phri.UUCP> roy@phri.UUCP (Roy Smith) writes: >>> the Atlantis crew replaced one of the orbiter's five general-purpose >>> computers after it failed ... >> I don't understand. I thought that the various computers were all >>constantly checking each other and if one failed, it would be taken off line >>automatically and the others would continue to operate without it. Why the >>need to rummage around behind the furniture for 4 hours? >Almost certainly they could have completed the mission without doing so. s/Almost //p They certainly could have completed the mission without that GPC (General Purpose Computer). >However, having one computer dead *before* reentry means having that much >less redundancy available against the possibility of further failures. >So shuttle mission rules say that if time and parts are available, the >repair will be done. Exactly correct. And it is in the flight rules. The basic philosophy behind that rule, and many other flight rules governing re-entry criteria, is that you want to re-enter with as much redundancy as possible. If you can get more redundancy, then the rules say that you should do so before re-entry. The spare GPC has not always been carried on board. I don't know exactly when they started carrying it, and I do know that some people thought it was a silly idea at the time. Probably because it takes so long to do a replacement. >The real botch here is that the computers are so inaccessible that it takes >four hours to replace one. My understanding is that they weren't really designed with in-flight replacement in mind. They were designed to be replaceable, yes, but primarily as part of on-the-ground servicing, and NOT as IFM (In-Flight Maintenance). The idea of carrying a spare on board and replacing a broken GPC was pretty much an afterthought. Some engineer said "hey, we can actually do this and it's a pretty neat idea, so why don't we?" My wife was on console in the DPS mission support room when it failed.... and it was her last flight, too! It was interesting to hear her description of the event. And in case anyone is interested, the specific problem that caused the GPC to "quit" was a memory parity error. William LeFebvre Department of Electrical Engineering and Computer Science Northwestern University