J. Presper Eckert
Eckert was a grad student studying at the Moore School in 1943. There, he
met John Mauchly, who spoke about the idea of an electronic version of the
Ganged Calculator. Mauchley needed someone to oversee the design and construction
of the ENIAC, and it was J. Presper Eckert who rose to the challenge.
Because vacuum tubes were well-known to burn out without warning, one of
the largest engineering challenges facing the ENIAC team was how to make
the tubes more reliable. Eckert's design addressed this problem in three
ways.
First, he designed the circuits of the ENIAC so that the tubes only needed
to represent "on" or "off" instead of any numerical
value. This design feature meant that even if a tube varied from its specification,
it would still be correct in terms of being on or off.
Second, the tubes themselves led "pampered" lives, for vacuum
tubes. Eckert knew that tubes usually fail early or late in their lives,
so he made sure the ENIAC used only tubes drawn from a special 'burnt-in"
selection. These were tubes that had been running long enough to be known
to be good, but still young enough to have many useful hours left. Also,
the circuits were only energized with about one quarter of the normal voltage
for the tubes. This way, the tubes ran much cooler and thus lasted longer.
Third, Eckert designed the ENIAC using a "component" system. By
organizing the parts into discreet components, Eckert simplified the tasks
of maintenance and troubleshooting. Whenever there was a problem, it was
a simple matter to replace the broken component.
These three design inovations, along with a general robust, over-engineered
approach to the problem, helped the ENIAC achieve a Mean Time Between Failure
(MTBF) that was much better then the skeptics had predicted. The sucess
of the ENIAC encouraged other computer pioneers to build even bigger machines.
Continue on to the Need for Speed...
Back to the ENIAC at Penn...