26 Aug
2011
26 Aug
'11
11:26 p.m.
On 2011 Aug 26, at 6:35 PM, William Donzelli wrote:
The issue of the transistors did occur to me, if you are aware of
problems with particular batches IBM used, that would be an issue.
Speaking generally though, I'm not convinced Ge transistors are as much
of a problem as they are sometimes made out to be (..kind of like
electrolytic caps), although yes, they are certainly more of a problem
than later/Si.
As an example, the three discrete-transistor calculators I have
comprise about 2000 Ge transistors. 6 of those 2000 transistors have
needed replacement (3 out of over 5000 diodes have also been replaced).
These are from 1966-68 so perhaps a few years later than a 7090, with
the potential for some manufacturing improvements, but probably of a
lower grade than IBM used.
The CHM writeup on the PDP-1 states "To the team's surprise, the PDP-1
worked almost perfectly right away", after the power supplies were
done.
Are you
speculating or do you know of some particular age-related
flaw in
said machines?
The transistors. There are some batches that IBM used in the SMS era
that have leakage issues, like many of the ca. 1960 transistors.
Combing through 50000 devices would be a challenge that never ends.