What, me worry?
Slowly, slowly, the charge on the floating gates of the EPROM (SE/30 Video ROM) is leaking away. A backup of the ROM contents would be good. A very small fraction of devices would fail in the 20 year design life originally targeted by Intel (described in Electronics, Aug 14, 1980, pp 132-141).
The metal vapor deposits inside the blown fuse area of the PALs can react with trace moisture and electric fields to regrow back a blown bit.
The plating on the sockets is porous, and chlorine and other reactive chemicals reaching the base metal can cause an insulating mushroom shaped corrosion product to grow out thru the pores and lift the socket spring contact away from the DIP IC lead.
The molded plastic IC encapsulant material, although much improved over the earlier materials of the 1970's, still is slightly permeable to water vapor. This in itself is not a problem unless the penetrating water vapor combines with trapped contaminants at the die surface at a vulnerable spot like metallization or oxide cracks, where corrosion will proceed.
The molded encapsulant has the leadframe, die and wirebonds in compression. With each thermal cycle associated with both ordinary use and with adverse storage conditions, the cyclic mechanical stress can cause crack propagation, eventually causing leakage paths for contaminants along the leadframe, lifted wirebonds, and cracks in the die and die attach propagating inwards from the scribe lines.
Similarly, the base lead seals of the CRT are under stress which varies with extremes of environmental thermal cycles. Cracks can propagate inwards destroying the vacuum seal, this is how tubes can go bad in storage and end up with that all white getter material suddenly.
When in use, the electrons are tearing away at the thin metallization of the ICs, the so-called electromigration effect. Usually not an issue for STTL and CMOS, but can happen if power metal has crack that reduces effective cross section.
The insulation of the magnet wire in the yoke and flyback is crazing particularly at the points of minimum bend radius. If adjacent turn insulation cracks nearly align and impregnation is porous, the electric field will assist moisture and contaminants in gradually creating first a leak then a shorted turn.
In the fabrication of the fiberglass strands for the multilayer PC board, a bubble occasionally occurs within a glass strand, which when drawn becomes a tube. During the fabrication of the PCB via holes are drilled and then the board is immersed for plated thru hole fabrication. If plating chemicals are trapped in a hollow fiber running between any two vias of dissimilar potential, this sets the stage for later development of conductive filaments between vias as electric field and available moisture combines to create via shorts mischief deep within the glass epoxy board material itself.
Twenty years is a long time for complex electronics to continue working. Many failure modes exist (more than the above) and will happen to the logic parts in the next twenty. And, I fear that the newer current hot running chips with lead free solder will be worse still. For the antique 30-40 year old electronic test instruments I collect and maintain, most but not all have failed at least once in the last 15 years requiring my diagnosis and repair of substrates, transistors, diodes, resistors and caps. If you are really paranoid, store your spares in the dark packed with dessicant and dry nitrogen at moderate and constant temperature. A vacuum could work too but will further speed up the dry out of any aluminum electrolytic caps and cause some additional evaporation of disk lubricants. Plastic cabinet spares would probably benefit from being bagged in a manner that avoids bag contact with paint (to avoid stains) but excludes light, oxygen and ozone. However I'm not up to being this rigorous, I would rather look for newer stuff to collect when time finally claims the oldest stuff in my collection.
] ]'>
