From: Dave Emery <die@pig.die.com>
Date: Thu, 28 May 1998 20:10:26 -0700 (PDT)
To: Tim May <tcmay@got.net>
Subject: Re: Satellite processing (was Re: DRUDGE-REPORT-EXCLUSIVE pablum)
In-Reply-To: <7iF8Pe1w165w@decode.com>
Message-ID: <19980528231016.A16286@die.com>
MIME-Version: 1.0
Content-Type: text/plain


On Thu, May 28, 1998 at 06:22:16PM -0700, Tim May wrote:

> 
> ROMs currently have no sensitivity to cosmic rays or alpha particles from
> natural radioactivity in the environment, and are not expected to. Single
> event upsets typically happen with cross-coupled static RAM cells or with
> dynamic RAMs. _Some_ errors happen in logic circuits, including
> microprocessors, but not many.
> 
> No doubt some examples of reported errors in ROM circuits can be found, but
> I suspect the errors are either in the decode logic or elsewhere, and
> certainly not in the mask-programmable bits or the blown fuse sections.
> Lots of ROM types exist, so I don't want to try to list all major types
> here.
> 
> Some EAROMs (Electrically Alterable) or EEPROMs (Electrically Erasable) are
> sensitive to some radiation events, but not nearly as sensitive as RAM
> chips are. The IEEE Transactions on Nuclear Science, the December issue of
> every year, is devoted to reporting on actual sensitivity levels and
> physical mechanisms.

	I certainly defer to your expertise, as I have no particular
knowlage of radiation effects aside from reading an occasional technical
paper or article touching on the subject tangentially.   What I should
have said in any case is damage to  Flash EPROM or other electrically
programmed EPROM, as several people have pointed out that fuse and mask
programmed ROMs are not at all subject to single event radiation damage.
I suppose that  ROMs on board a satellite could easily afford to be
masked types, but of course there still are sound engineering reasons to
keep ROM'd code as minimal, simple and straighforward as possible, especially
when it cannot ever be changed and hundreds of millions of dollars of
satellite depends on it.



-- 
	Dave Emery N1PRE,  die@die.com  DIE Consulting, Weston, Mass. 
PGP fingerprint = 2047/4D7B08D1 DE 6E E1 CC 1F 1D 96 E2  5D 27 BD B0 24 88 C3 18