Re: Don't Fix It if it is Not Broken (was Looking at Macs...)

In article <nowhere-5990BF.07245824082005@xxxxxxxxxxxxxxxxxxxx>,
Travelinman <nowhere@xxxxxxxxxx> wrote:

> In article <230820052031060855%NoSpamDammit@xxxxxxxxxxx>,
> Mark Conrad <NoSpamDammit@xxxxxxxxxxx> wrote:
>
> > In article <nowhere-28594A.19425823082005@xxxxxxxxxxxxxxxxxxxx>,
> > Travelinman <nowhere@xxxxxxxxxx> wrote:
> >
> > > > Don't even try operating the same computer in an aircraft at 60,000
> > > > feet, because the soft error rate increases to 6,400 per month, or
> > > > appromately one soft error every 8 minutes of steady operation.
> > > >
> > >
> > > Since people operate computers continually for months at sea level and
> > > for hours at a time at 40,000 feet, don't you think your numbers are
> > > suspicious?
> >
> > Not at all. When I was working at Hughes research in my working days,
> > it was not at all uncommon for me to run tests on the old "Surveyor"
> > series of robot spacecraft (components) for months.
> >
> > Heck, 60,000 feet did not even require that I fire up the oil diffusion
> > pumps, the "Roots" mechanical backup pump was sufficient to reach that
> > relatively modest altitude.
> >
> > Of course, the completely assembled robot "Surveyor" had tighter time
> > constraints regarding its testing, so I could not test it in the vacuum
> > chamber for months like I did with the components and sub-assemblies.
> >
> > Generally, a week was the longest I could tie up the robot with tests,
> > before other people started yelling at me.
> >
> > The best we could do vacuum wise in those old days was about .0000001
> > torr, which is approx' equivilent to the pressure about 200 miles above
> > the Earth's surface.
> >
> > A small low priced vacuum chamber with a "Roots" mechanical pump is
> > aufficient to reach .1 to .01 torr. (approx' 100,000 feet)
>
> What does any of that have to do with your assertion that a computer
> would have a RAM failure every few hours at sea level and every 8
> minutes at 60,000 feet?
>
> Your numbers are absurd and a moment's reflection on real life would
> show that. Your silly diversion into air pressure doesn't change that.

I wrote a white paper on soft-errors for Actel Corporation about four
years ago called "Understanding Soft and Firm Errors in Semiconductor
Devices". It seems to me that I remember writing that soft errors
(caused by cosmic rays or photons knocking neutrons loose from gas atoms
at high altitude or by alpha particles emitted by thousands of different
things in our daily lives including the packaging materials used in
making semiconductors themselves) are 100-800 times worse at commercial
aircraft cruising altitudes than they are at sea level. This is rarely a
problem with personal computers because the memory cells are
cross-coupled and therefore unlikely to be "flipped" by a random alpha
particle or a neutron. However, SRAM based devices are much more
susceptible to soft errors because the capacitor size is small and SRAM
cells are generally only one or two transistors and are NOT cross
coupled.
.