Lou Scheffer wrote:

(Lou Scheffer) wrote in message om...
(Derek Lyons) wrote in message ...
h (Rand Simberg) wrote:

"But that doesn't explain why similar systems designed for marine use
[...] can be built for orders of magnitude less."

Please provide a cite of a marine system of the equivalent complexity
and mission of the weather sat with a design lifetime of years or
decades. (Or any marine system with those kinds of lifetime
requirements.)

Transatlantic cables come close. Before fiber optics, these had
repeaters every few km. Since they were in series, ALL of the
repeaters must work. If I remember correctly, they were designed so
the cable as a whole had a 20 year lifetime. Using good old
conservative engineering, the even made this work with vacuum
tubes(!). Needless to say this was one of the first applications of
transistors.


Oops - this was NOT one of the first uses of transistors. Another
article states that they were still built with vacuum tubes in 1964,
10 years after transistors were used for other applications, since
transistors were not yet proven to be more reliable. In retrospect
this makes perfect sense - any reliability conscious field will not
rush to adapt new technology, no matter how promising, if the old
technology is working at least OK.

Lou Scheffer


When a shuttle crew did a satellite upgrade (not sure if it was Solar
Max, or Hubble), it was noted that the replacement units were 386
processor based. Some people didn't understand why...you basically just
answered them. (Espically when you add in uncertainty as to how well a
new processor design will tolerate long term ionizing radiation
exposure)