Re: Effect of radiation on computers

["R. G. Oesterling" <rgo@srv.net>]

At 18:29 2/22/96 -0600, Bob Flood wrote:
>I'm not sure if I missed a reply or 2, or if this aspect just hasn't been
>discussed, but there is a side of potential radiation-induced damaged that I
>don't believe has been addressed yet. While Radsafers have contrinuted their
>knowledge about radiation-caused damage to silicon and other computer
>materials, they haven't discussed the possibility of radiation-induced
>changes of zeroes and ones. All computer chips store whatever the chip is
>made for as zeroes and ones, and each one has the potential to be changed
>(flipped to its opposite) by ionizing radiation. In fact, there was
>sufficient interest in this phenomenon to begin research into a computer
>chip-based dosimeter a few years ago. I never heard any results, nor do I
>have any references onthe subject. Perhaps other may.
>
>This aspect may be of greatest interest to users. Nuclear plants that place
>PCs in posted radiation areas for access control and other purposes may find
>intermittent operating problems (apparently random) that may go undiagnosed
>as radiation induced. The probability of a bit-changing event would be
>compounded by the probability of changing a bit of any real importance
>(fatal vs non-fatal errors) in a device of real importance to the computers
>and its applications. Bit changes may be happening regularly on some
>machines, but generally go un-noticed because the effects of most bit
>changes are trivial (changing the spelling of a word in a menu, the color of
>a pixel on a screen, etc.). Only those changes that cause a detecable error
>or failure to function would attract attention.
>Bob Flood
>Unless otherwise noted, all opinions are mine alone.
>(415) 926-3793
>bflood@slac.stanford.edu

Some early portables I observed to have anomalies in the BIOS after being
passed through a security x-ray.  The failure mechanism has been postulated
to be radiation damage in the CMOS, but there hasn't been any verification
because we couldn't afford it.  Units built recently don't seem to be
bothered by being passed through the x-ray devices.

Other software is generally so buggy that one would be hard-pressed to
filter out the noise to get to the errors caused by radiation damage.

R. G. Oesterling