[ExI] Why space tech isn't cutting edge

[Adrian Tymes]

On Mon, Nov 19, 2012 at 11:30 AM, spike <spike66 at att.net> wrote:
> Adrian, do invent such a coating, please sir.

All I can do right now is research and bounce ideas around.
That said...

>>...Composite armor, just like tanks use.  Outer layer converts massive
> attacks into a spray of smaller attacks.  Inner layer optimized to stop a
> bunch of smaller attacks...
>
> Ja most current research in that field goes that way.  The problem is that
> cosmic rays have such high energy we are lucky we have both an atmosphere
> and a magnetic field around us.  A long time ago I did a calculation to
> estimate the lead shielding equivalent of an atmosphere, and as I vaguely
> recall it was substantial, on the order of a couple of feet of lead.  I
> don't recall the answer, but I do recall concluding it is not a practical
> means of protecting electronics.

Only a couple feet, for single-material (lead)?  That
sounds like it could be optimized down to practicality.

Would, say, tungsten work better than lead for an
external coating?  (Or osmium, but that might get
expensive.)  Since I gather a high density is needed
for initial capture, and thus is the most desirable
property for the outermost coating.

What properties are best for deflecting a massive
number of secondaries, that might not be necessary
(or even desired) in the outermost coating?

>>...Not that huge, surely?  Considering how tiny CPUs are, and thus the tiny
> volume that would need protection...
>
> You hit upon a promising approach: figure out a way to do massive
> redundancy, so that the RISEU errors can be outvoted by several parallel
> processors.  This is done now to some extent.  We have a control system that
> uses three parallel processors.  The output is not used unless two
> processors agree on an answer.

I note that the trend in modern CPUs is multi-core,
more than higher CPU clock rate.  Hexa-core
processors are commercially available; that can
emulate dual-core with 2-out-of-3 processing.  Or
alternately, do quad-core, reserving 1 of the cores
for backup if a core gives consistently bad results
(and if that's necessary, it flags a warning to the
humans that one core is judged dead).