Failure modes of old hardware (in general) :
 
1)      Oxidizing contacts at any connector: (IC sockets, power-supply
        connectors, and many other places.)  Can usually be cleaned and
        fixed for a while.  A good contact-cleaning substance is a benefit
        here. (Back when it was still made, I swore by Cramolin.  They now
        have some other replacement for that which seems not as potent (but
        I feel better about spraying it indoors. :-)
 
2)      EPROM Alzheimer's:  EPROMs store information by charging isolated
        areas in the IC.  No insulation is perfect, and over the years the
        charges decay enough to generate reliability problems (perhaps
        temperature sensitive behavior), or totally inhibit operation.  The
        only fix here is to read the contents of the EPROMs *before* they
        give problems, and either burn copies, or keep the information on
        disk/tape which can be read later to re-create the EPROMs when they
        fail.  As I remember, the manufacturers were suggesting that you
        could expect a 5-10 year period of no problems, and later would get
        more and more problems.
 
3)      Electrolytic Capacitors:  These tend to dry up and fail --
        especially if not powered up most of the time.  Replacement with new
        ones is the best solution.
 
4)      Chip failures: This one is a factor of heat cycles (the only other
        one is the connector problems above, which are made worse by the
        thermal cycling which will walk chips out of some sockets over time.)
 
        However, *real* chip failures follow a probablilty curve.  The first
        few months will see the most failures in the early years.  Usually
        it is a problem with cracks in the edges of the chips (from the
        dicing step during manufacture).  These cracks grow with every heat
        cycle (just as winshield cracks grow in winter).  If the crack is
        already curving back out without having hit a trace or active area,
        it probably will do no harm.  However, if it is still pointed into
        the active area of the chip, it will probably grow with each power
        (and heat) cycle, eventually interrupting some circuit function.
        The chips which last beyond this stage will probably last for quite
        a few years.  Eventually, there will be enough migration of dopant
        to render the chips non-functional.  The migration rate of dopant is
        something complex like perhap the temperature to the fourth power,
        so a little temperature difference can make a major difference in
        long-term life.  This will probably be the eventual failure mode if
        the chips are not killed by rouge power supply voltages.
 
        At one time, chip makers were having a problem called "purple
        plague", which was (I believe) an interaction between the aluminum
        'vias' (conducting paths) and the gold wire used to bond to the IC
        package pads which lead out to the chip pins.
 
        I'm sure that I've overlooked some failure modes, and may have also
gotten some other details wrong, but this will give you some idea as to what
can happen to aging machines.
 
        Good Luck,
                DoN.
-- 
 Email:   <dnichols@d-and-d.com>  |  Donald Nichols (DoN.) 
 Voice Days: (703) 704-2280       | Eves: (703) 938-4564
 My Concertina web page:          | http://www.d-and-d.com/dnichols/DoN.html
        --- Black Holes are where God is dividing by zero ---