The idea of creating a wiki with information on parts compatibility is one proposal being looked into.
That is good idea for starting up with that. I would say there is 3 groups of parts: 1) Some parts are replaceable with COTS (Commercially Available of The Shelf); 2) some are very customized for particular model not COTS, but could be produced by third party in some batches if providing right parameters (metal gaskets, O-rings, apertures, where only physical dimensions would be required to be reverse engineered). 3) then there are components designed by Cameca and either produced at limited number at Cameca factories, or outsourced, but are very complicated designs with hidden inner working (needs firmware that is electronic boards and modules). We have some experience up-keeping our probes (SXFiveFE and SX100) (as lab in poor university having no enough budget for paying for service contracts), where most fixes were replacing parts from group 1 or 2. The problem after Cameca support will cease will be in groups 2 and 3.
Example of COTS parts (1st group):
Ion pump, Diffusion pump, video camera, VME PSU, passive components and non programmable CHIPS on electronic boards.
Possible problems: electronic component getting obsolete. Real-Life example: We were getting peeper color noise on our SX100 physical monitor. The problem was memory going bad in RGB composing chip and some bits in color table self switching state from 1 to 0, chip is no more in production. There is some shops with old stockpiled or refurbished (salvaged) parts two years ago I could manage to order 3 such chips. At that time it was showing that there was 15 such available. So One thing is to replace burnt resistor, inductor, damaged capacitor - these rather would not get obsolete, or can be replaced by something similar. Other thing is the more complicated Chips, where exactly same model should be used, as board was designed for such timing, voltage level and pin configuration. Obsolescence is particular problem for older SX100 electronic boards.
group 2:
O-rings, apertures, gaskets...
Apertures theoretically can be cleaned and reused. O-rings can age and could need replacement - the custom form could be ordered in batch - there are pretty many of vendors which produce it, metal gaskets are one-time use - and needs replacing with new every time. Tungsten cathodes can be refurbished. However there is no clear way to refurbish used separation and spectrometer windows - those are very custom Cameca parts and there is very important know-how of its production. Even if Detector windows would be made unnecessary replacing GFPC with solid state detectors, the separation windows still are needed as that is blocking the electrons from the chamber as there is no other mean to block i. Maybe we could come up with some custom replacement but that probably would be worse than what was developed by Cameca. The similar issue is with beam-regulating aperture - the production of it is customized by Cameca, and it would be hard to produce it by third part. Albeit it is much less hassle for Field emission SXFiveFE where (as far my experience gets) it performs worse than plain aperture.
Group 3.
Old electronic boards are quite easy fixable (to some point). Any volatile component (which state is not saved between on/off) or linear components are replaceable and new parts or salvaged/refurbished parts can be drop-in replaced (some components are socketed, other are soldered, but most of components are THT on the old boards). However some boards contain firmware - that is some of chips are programmed to define its behavior and even if replacement chip could be found to buy (which is still possible) - it would be worthless without programming/flashing-in the same logic. PAL (present in many of old electronic boards) and non-volatile FPGA's (for WDS) without knowing what should be flashed-with are worthless. PAL are simple - the logic could be reverse engineered while we have at least one working chip as they basically react to input and produce the corresponding output - so the behavior could be reverse engineered with careful scanning of output response to inputs (needs logic analyzers, which BTW I have and are familiar with). FPGA, however... It is close to not reverse engineer-able.
Then there is this new generation electronic boards, where multiple PAL devices are replaced by single big FPGA's. Those FPGA's are already past EOL (Altera(intel) Cyclone) (maybe it is additional issue for decision of production cessation, different FPGA would require different board design and so would complicate the software as there would be not just two groups of electronics (old/new) but 3 groups, and that would bring-in more complexity (also FGPA licenses for these biggies are very costly).
Few more parts which would be hard to replace or repair without Cameca:
HV tank for SX100. Theoretically it is very simple design, but it is filled with dialectric oil, any fixing would be dirty and it requires certified HV electrician skills, that is because possibility of going up to 50kV; up to 30kV (what most of SEM offers) would be dry HV generators. I.e. SXFiveFE contains separate (OEM) HV generator, and HV regulation board of Cameca is partly bypassed. So there should be some kind of possibility using custom HV generator replacement for SX100 at cost of giving up 50kV. Such generator would be easier to just drop-in-place in case of malfunction.
I was already conducting some initial investigation for reverse engineering of electronics of SX100 and SXFiveFE - I wanted to use the free slot on VME backplane (reserved for EDS connection) where VME bus and signals are exposed, and so I had acquired some tools for that (32 channel logic analyzer) and have some "wiretapping" for VME communication board half designed. I was doing that from curiosity and with a hope to overcome some difficulties, and have some better way of direct communication with micro-probe alongside working system. This situation, I think, will make me to pursue that much further as necessity.