PPRuNe Forums - View Single Post - Computers in Aircraft
View Single Post
Old 14th Oct 2008, 23:13
  #10 (permalink)  
sprocket check
 
Join Date: May 2007
Location: Sydney
Posts: 394
Likes: 0
Received 0 Likes on 0 Posts
Correction:

Computers will THEORETICALLY do what they are told to do. Silicon is not 100% reliable and chips do degrade and fail. Not very much if they are high grade as in aviation (mostly), but they are still subject to aging and heat stresses, electrical surges and spikes, etc. Even the simple switching on and off causes stresses to an electrical circuit.

The two major problems with computer reliability:

Humans program them. As the systems get more complex, so does the software required to run it. The variables needed to be tested increase exponentially and there comes a point where the testing for all possibilities becomes impossible or so prohibitively expensive that it is put into extremely unlikely category of failure and simply not done. Only the engineers may ever know what the potential faults are.

I call them 'undocumented features'.

The second is that computers have deep and dark souls. With many systems, the startup diagnostic is a compromise between testing functionality and time. To thoroughly test a hardware set on boot up could take minutes, even hours. We get frustrated when the GPS isn't on instantly, right? Thus one will never know if a particular RAM chip is not quite remembering everything or a particular transistor or transistor set is giving the correct output.

Overenthusiastic reliance on computers is quite unwise, IMHO.

sc
sprocket check is offline