From: nuny@bid.nes   
      
   On Thursday, June 16, 2016 at 3:41:28 PM UTC-7, Thomas Koenig wrote:   
   > One of the more ambitious aspects of SF tech are machines that   
   > continue to work for a _very_ long time.   
   >    
   > The thought that you can build a generation ship that lasts hundreds   
   > of years, or that you can find alien machinery that still works   
   > millions of years after their builders have left it, turns up   
   > quite often.    
      
     I like Niven's stasis field solution better. If that's unavailable...   
      
   > How would this work for our current level of technology?   
      
    Depends on what it's supposed to be able to do, how often it has to do it,   
   how long it has to be able to do it, and what hazards it will be exposed to in   
   the meantime.   
      
   > Glass should be OK over a very long timeframe.   
      
     As long as it isn't subject to thermal/mechanical shocks.   
      
     Also, it depends on what you want it to do after a very long time. AFAIK   
   we're pretty sure glass isn't a liquid, but certain ranges of certain   
   impurities make glass more likely to crystallize over time. If your glass is a   
   telescope lens the refractive    
   index will change a bit, probably manageably so, but if it's a final amplifier   
   disc in a high-powered laser system this is a serious KABOOM issue.   
      
   > Metals can last a long time in vacuum, or under a (really)   
   > inert atmosphere, so that is feasible, unless water or   
   > oxygen are present.    
      
     Can it do whatever it's supposed to do in there for practical ever?   
      
   > Lubricants for metal parts fare much worse - they would   
   > evaporate, or turn to coke, or...   
      
     There's only so much to be done about friction; more contact area means more   
   friction but less loading. Ball bearings are a good tradeoff, but the balls'   
   hardness has to match both the borne and bearing parts of a rotating machine   
   (roller bearings for    
   slides) otherwise the softer will suffer permanent deformation or crack   
   because all metals are malleable and or brittle.   
      
     There's magnetic bearings but "permanent" magnets really aren't unless you   
   keep them really, really cold.   
      
     Keeping things in general cold is also good to help minimize diffusion,   
   creep, evaporation of fuels, lubricants, and reactants, but too cold is as bad   
   as too hot because material properties change a lot when they get really cold-   
   think superconductors,    
   superfluids, and supersolids not to mention embrittlement and re   
   rystallization in metals.   
      
     Ceramics can do many things metals currently do equally as well or even   
   better but need some work to replace all metals/alloys.   
      
   > Electronics will degrade over time due to a multitude of   
   > effects such as thermal diffusion of doping atoms, cosmic   
   > rays, recrystallization etc.  The dielectric constants   
   > of capacitors can break down, etc.   
      
     So avoid semiconductors for conventional electronics and use vacuum tubes,   
   glass- or vacuum-dielectric caps, ceramic coil formers and whatnot instead.   
      
     High-gate-count stuff like memory and processors made of such means any A.   
   I.s will occupy cubic miles though.   
      
     One known way to mitigate radiation damage in semiconductors is to make all   
   of the elements e. g. gates bigger. This also makes catastrophic diffusion of   
   dopants over time take longer. It means your A. I might take up cubic yards   
   instead of miles.   
      
     I've seen blue-sky claims that quantum computers can be built without using   
   semiconductors but it isn't quite current tech.   
      
   > Polymers are much worse - they can depolymerize due to   
   > radicals, yellow etc.  Their additives can migrate, or   
   > they can break down.  This can have serious consequences   
   > for elctronic or electric devices as well, if insulation   
   > breaks down.   
      
     Insulation is indeed a long-term problem.   
      
     For high voltages, distance is the only truly reliable insulation even in   
   the short term- for that e. g. glass standoffs etc. are fine for a few   
   centuries at least. For lower voltages bare freestanding point-to-point wiring   
   as used to connect    
   semiconductor chip pads to their package pins works, but such thin wires will   
   sag after a few millennia and it isn't exactly conducive to miniaturization   
   when you have to do entire circuit blocks with it. Gold-plated oxygen-free   
   copper on glass boards    
   kept at a very constant temperature should also last a couple centuries.   
      
     One other serious issue is very long term power storage. Known batteries are   
   Right Out- secondary batteries lose charge capacity with successive   
   charge-discharge cycles over time for various reasons and self-discharge even   
   when not supplying power.    
   Primary batteries also self-discharge when not being used.   
      
     Any active power supply system that has to be called into action after a   
   long sleep (from diesel to fusion reactor) will likely have moving parts and   
   fuel/lubricants/reactants of one sort or another that can leak away or degrade   
   as well.   
      
     Spring-powered clockwork? Springs lose their springiness over time.   
      
     Externally-supplied power from solar cells or ambient RF rectification   
   (crystal radio-fashion) isn't very reliable or very intense (unless your   
   machine is in close orbit of a star, but then it's hot), but there's none of   
   the inherent issues of    
   batteries or generators/reactors either.   
      
   > So - really long-lasting technology needs constant maintenance,   
   > unless you want to postulate materials that are far different   
   > from what we have today.   
      
     I'd say occasional rather than constant on the scale of a century or teo   
   between house calls.   
      
     I see no real reason a really carefully designed minimally computerized   
   machine made using current tech can't last unattended a few centuries in say   
   Neptune orbit. Millennia is way harder.   
      
      
     Mark L. Fergerson   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)