[comp.misc] Core memory and the Shuttle

kurt@tc.fluke.COM (Kurt Guntheroth) (07/20/88)

I believe the memories on the shuttle are plated wire memories, which are
like core memories in function but are made with a somewhat more modern
mechanical process.

Core memories are intrinsically rad hard and nonvolatile, which is quite
useful in space, but the real reason they are used instead of some semi-
conductor memory is that the core memories and the ancient computers that
contain them are already rated for manned space flight.  The paperwork for
certifying a more modern device is so byzantine that it is simply not worth
the effort.

Let me give you an example of what I mean.  My brother worked for Rockewll
on the space shuttle.  They had a rule that it was worth $10,000 a pound to
remove a pound of weight from the shuttle.  For instnace, the walls of the
crew cabin were some sort of exotic alloy sheet about as thin as TV dinner
trays, and stamped into a waffle pattern for structural strength.  Onto
this wall was bolted a fire extingusher on a soft iron bracket(!).
My brother pointed immediately to the bracket and said, "Let me whip out an
aluminum bracket for this sucker."  Well, it turns out that the fire
extinguisher had been rated WITH the bracket, and would not qualify without
it.  The cost of re-rating the extinguisher alone was many times $10,000, so
the idea had to be abandoned.

So, you won't see semiconductor memories or modern computers in space unless
the added power and capacity becomes mission critical.  By the way, this
same sort of silliness goes on in the military, which is what leads to $500
hammers and so forth.  The contractor knows a $4 hammer would be cheaper,
but is constrained from buying it.  And then when the government tries to
lessen the regulation, the contractor generally goes out and buys a $.50
hammer that breaks on the first nail.  Makes you proud to be an American, huh?

jwatts@hpihoah.HP.COM (Jon Watts) (07/21/88)

mec@ardent.UUCP (Michael Chastain) writes:
>I've read that NASA successfully recovered the Challenger core
>memories.  Talk about a violent "core dump"!

This is quite likely true when I was working on core memories for the F-16 we
recovered data from several memories that spent two weeks in salt water in the
North Sea.  The cores don't care only the drive electronics have trouble with
it.
 
>sometime in 1987.  Ampex makes core memories for, among other things,
>shuttles.

I may be mistaken but I believe Ampex only makes the cores themselves for the
shuttle and not the complete memory assembly.

+-------------------------------------------+
|These opinions do not neccessarily reflect |
|those of my employer, my country or anyone |        -Jon Watts
|else if fact they aren't even opinions, in |
|fact, they don't even exist, neither do I  |
|your hallucinating again. Is this Kansas?  |
+-------------------------------------------+

bms@bdt.UUCP (Chris Rhodin) (07/28/88)

Well all this talk about core memories got me thinking.  I have
an old Ampex 16k byte Multibus Core memory card ( circa 1975 ) .
    This was bought from a surplus dealer and was used in my first
"computer" which had an Intel 80/10 cpu card.  I was the envy of 
all my friends, imagine I did not have to use the teletype or 
paper tape anymore.
    I took the box out of the closet yesterday, dusted it off, 
turned it on,  low and behold it still had the the same code in
it that I wrote back in 1978!  Ten years, seven moves, and no
power. I can see why Nasa likes core.

Vance at Berkeley Microsystems.