Intel486dx33 wrote on 2020-02-21, 18:39:

Why do you think old computer are still working today ?
Why does PCI and AGP and PCIe cards fail ?

Hi, I think one reason is heath and the size of the silicon.
Old microchips had larger circuits that were more sturdy (in in relation to modern tech).
And the fluctuations in temperature were more little (no engery saving mechanisms that cause hot-cold-hot cycles).
The manufacturing process also was different. NMOS technology, a successor of TTL, for example, as used by 8086/80286 CPUS,
used resistors that were power-wasting but more robust also.
https://en.wikipedia.org/wiki/Transistor-transistor_logic
https://en.wikipedia.org/wiki/NMOS_logic

Speaking of CPUs, a very very interesting one was the COSMAC CDP1802.
It used advanced Static-CMOS design and existed in a rare, radiation hardend version (Silicon on Sapphire design), too.
Due to its elegance and highly variable power supply voltage (4 to 10,5 volts),
it was used in amateur satellites running IPS (OSCAR-10,13, 40 etc) and the CHIP-8 computers.
https://en.wikipedia.org/wiki/RCA_1802
http://www.cosmacelf.com/publications/data-sh … cdp1802-rca.pdf
http://www.retrotechnology.com/memship/amsat.html

Edit: That reminds me of the Voyager probes. Their old technology is ancient, but robust (and include redundancy, ie. there are multiple circuits doing same task).
Or in other words, 70s tech was best. just imagine they are like 40+ year old washing machines that still work
(and will fail in near future only due to power loss because of the RTGs' limited live span). Purely awesome. 😎
https://science.slashdot.org/story/07/09/01/2 … ager-technology

Intel486dx33 wrote on 2020-02-21, 18:39:

Why do you think 8-bit and ISA, and VLB cards are still working today ( Some 30 years later ).
Where PCI, AGP, and PCIe cards fail.

I'm betting the capacitor plague is responsible for at least some of those failures.

Joseph_Joestar wrote on 2020-02-21, 19:47:

Intel486dx33 wrote on 2020-02-21, 18:39:

Why do you think 8-bit and ISA, and VLB cards are still working today ( Some 30 years later ).
Where PCI, AGP, and PCIe cards fail.

I'm betting the capacitor plague is responsible for at least some of those failures.

Interesting. Reminds me of the transistor disease of old Germanium transistors (as made by Philips etc).
Because of some unknown contamination, some fur grew within the metal caps that made the transistor unusable for RF.
Which in turn was really mean, because bipolar transistor testers didn't help in detecting that (they only measured current).
So instead, for real testing, a test oscillator rather had to be built. Re: Preventing bad caps

Edit: Speaking of the capacitor plague, there's something that makes me wonder.
Why do capacitors have no date code ? Without one, how do you know that your electronics seller keeps his caps range "fresh"
rather than keeping his New Old Stock (NOS) on sale for 20 years and more until it is sold out ?

I haven't had any issues with PCI cards, except some Audigy 1 cards that arrived DOA from eBay. Other than that, all of my PCI cards that ever worked, still work. Same for AGP and PCIe, although, I do expect hot-rod graphics cards to fail earlier than their contemporaries. They run hot, they heat-cycle with load, they're often pushed to the limit as-is. The margins on medium-bandwidth components (I/O, network, sound, and cost-conscious video cards) are much looser, and so things can degrade over a period of time and still be within tolerance.