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: | 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 ---