What is the first computer? - test-drb@ccmp.vtda.org

18 Jun 1998

William Donzelli quoted Tony Duell as having written:

...
   Agreed. You
can understand an ASR33 or similar by turning it by hand and 
 tripping linkages, etc. I did it years ago. Doing the same to a video 
 terminal, even a simple dumb terminal, is a lot harder (Done that as well!) 
 I have actually done that with a chunk of one of these computers - the 
 turn of one gear would cause a hundred others to move.

  The other nice thing is that, with the possible
exception of custom cams, 
 it's generally pretty obvious what broken parts should be like. And then 
 it's possible to make them without too much equipment. But a dead custom 
 chip is almost impossible to figure out, and hard to reproduce. 
 That is very true - a stripped gear is a stripped gear is a stripped gear!  
Ooh! Not so fast!  Is it straight, helical, worm or hypoid?  Or an 
eliptical gear even?  But yes, a gear is easy to rconstruct from its 
wreckage.  Some cams and levers, though, have quite tight tolerance 
spikes and notches which it is quite difficult to get right from seeing 
the bent/broken ones.

...
  Fixing one of these mechanical computers, however, is
completely different 
 from the familiar digital ones. In the latter, all problems (except video) 
 are go/no go. In the former, problems manifest themselves in the outputs 
 giving out-of-spec or erroneous results (except with a catasprophic 
 failure!). 
Bearing in mind Tony's, Sam's and others' comments on intermittent 
faults and the like, yes, up to a point.  Video is not the only 
exception, though - other things (e.g. disk drives) can suffer 
similarly.

And finally, remember digital mechanical computers do exist - I have a 
Facit mechanical calculator with some quite sophisticated algorithms 
(optimised multiplication, non-restoring division) done digitally, 
entirely with mechanical parts.  (The only electrical bits are a motor 
and a switch.)

Philip.