home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!ames!data.nas.nasa.gov!taligent!apple!kip2-41.apple.com!user
- From: keithley@apple.com (Craig Keithley)
- Newsgroups: sci.space
- Subject: Re: Shuttle computers
- Message-ID: <keithley-211192203501@kip2-41.apple.com>
- Date: 22 Nov 92 04:51:14 GMT
- References: <BxsMA3.D2o.1@cs.cmu.edu> <Bxvr1K.36L@zoo.toronto.edu>
- Sender: usenet@Apple.COM
- Followup-To: sci.space
- Organization: Apple Computer, Inc.
- Lines: 34
-
- In article <Bxvr1K.36L@zoo.toronto.edu>, henry@zoo.toronto.edu (Henry
- Spencer) wrote:
- >
- >
- > It's not a cycle-by-cycle lockstep like some redundant systems. Every
- > couple of milliseconds, the four computers in the main redundant set
- > compare notes; if one disagrees with the others twice in a row, the
- > others declare it to have failed.
- >
- >
-
- I seem to recall that this is correct; that the four computers running the
- same code are on something like a 40ms main event loop, and that they check
- their results at that point. I've also read/heard that the avionics
- sensors transfer (using some forerunner to the MIL-STD-1553 communications
- bus) their data at roughly the same time. This has something to do with
- the incredible number of sensors throughout the spacecraft, and that the
- data has to be multiplexed before the CPUs process it.
-
- So at the beginning of every 40ms loop, the CPUs acquire data from the
- multiplexed data stream, evaluate it, reach a decision, prepare to output
- new commands to the servors/thrusters/etc., vote on those decisions, and
- the winning decision is feed back out to the spacecraft systems.
-
- Newer fly-by-wire systems need a faster communications bus and a shorter
- main event loop.
-
- I wish I could remember where I've read this...
-
-
- Craig Keithley
- Apple Computer, Inc.
- keithley@apple.com
- Anything not forbidden is mandatory.
-