Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site petrus.UUCP
Path: utzoo!watmath!clyde!cbosgd!ihnp4!mhuxn!mhuxr!ulysses!gamma!epsilon!zeta!sabre!petrus!karn
From: karn@petrus.UUCP (Phil R. Karn)
Newsgroups: net.analog
Subject: Re: "Nuclear Event Detector"
Message-ID: <775@petrus.UUCP>
Date: Sun, 22-Dec-85 01:11:08 EST
Article-I.D.: petrus.775
Posted: Sun Dec 22 01:11:08 1985
Date-Received: Mon, 23-Dec-85 04:28:23 EST
References: <799@brl-tgr.ARPA> <634@sfmin.UUCP>
Distribution: net
Organization: Bell Communications Research, Inc
Lines: 31

Another (more constructive) application where radiation is a problem is in
spacecraft. Compared with nuclear weapons, which emit mostly gamma radiation
during the actual explosion, space radiation is generally particulate (a
wide range of electron and proton energies, with an occasional heavy ion).

The mix and intensity of particles depends on the orbit. The inner Van-Allen
belt is mostly protons and the outer one electrons, both consisting
primarily of the solar wind trapped in the earth's magnetic field.  The
particle energies vary over an enormous range; most are easily stopped, but
it's impossible to get it all (especially the primary cosmic flux).

The amateur radio satellite AMSAT-OSCAR-10 is flying in an elliptical 4000 x
35000 km 26 deg orbit, which takes it through both belts. The onboard
computer memory is 12 Mostek 4116 DRAMS with ECC. We generally see 2 soft
errors per perigee passage (where the radiation is most intense), and a few
weeks ago (after about 2.5 years in space) we had our first hard failure.
Two physically adjacent bits in one chip failed, but this isn't too serious
because the rest of the chip seems okay. In any event the ECC takes care of
it, as long as the same locations don't fail in another chip.  The theory is
that a single heavy particle took out the two cells.

We tried to shield the hell out of those memories, since they are the only
radiation-susceptible semiconductors on the spacecraft (all else is RCA
CD4000B series CMOS, which is rated to 1 megarad). A block of brass is
bolted around the chips on the PC board, and it seems to have kept things to
a manageable level so far. However, it would obviously be better to find
chips that are inherently more resistant to radiation than DRAMS.  Does
anyone have any suggestions? The last time I looked at this problem, either
the chips were too tiny (e.g., 1K x 1) or way too expensive.

Phil