A brief history of hardware epidemics
eclecticlight.co
eclecticlight.co
The fix back then was to bake your GPU in the oven for a while, essentially reflowing some of the cracked solder.
And I know of countless BMW M3s and M5s dying too soon because of early iterations of lead-free bearings.
I understand the toxicity of lead, but I wonder if the hand could've been more targeted. Does lead in bearings really show up in the environment?
The origin of the capacitor plague is so interesting:
https://en.wikipedia.org/wiki/Capacitor_plague
> A 2003 article in The Independent claimed that the cause of the faulty capacitors was due to a mis-copied formula. In 2001, a scientist working in the Rubycon Corporation in Japan stole a mis-copied formula for capacitors' electrolytes. He then took the faulty formula to the Luminous Town Electric company in China, where he had previously been employed. In the same year, the scientist's staff left China, stealing again the mis-copied formula and moving to Taiwan, where they created their own company, producing capacitors and propagating even more of this faulty formula of capacitor electrolytes.
Stolen and stolen again.
Part of the issue is in manufacturing. It might be hard to prevent exposure of employees to lead dust if they’re machining parts containing lead even if the final product isn’t too risky.
As far as I am aware, the act of soldering does not produce any sort of lead vapor or particulate either.
Er, no. Look up hazards of soldering fumes.
Oh that brings back bad memories! We were running a LAN centre, and our 7900GT graphics cards were failing left and right. We bought 23 8800GTS cards to replace/upgrade the lot. After a year or so they all started failing too. Reflowing, i.e. baking in a cheap little electric oven in the staff room, would give them an extra 6 months or so of life. After each subsequent baking, it would last less time than the previous. Having to replace so many graphics cards, after a much shorter that expected lifespan, was a lot of money for a tiny business. (Having said that, looking at how much it would cost now, I shouldn't complain.)
I read at the time that it was because of microscopic cracks in the solder, but hadn't realised before now that it was due to the removal of lead. We had no further problems after switching to AMD, but I never knew whether it was really an NVidia problem, or just those models, etc.
The GeForce FX 5900XTs, from the 2004 PCs (before the article's 2006 date of the start of the problem), were still working fine 10 years later, albeit in old PCs used for just web access and the occasional game of Bejeweled.
A related failure mode (which is closer to organ transplants I guess), is that when replacing a part with a faulty one, a defect in the new part can cause the other parts to bear more load. When a part fails abruptly you have a halted system and a lot of healthy parts, but when a part fails gradually, the whole system starts to degrade with it by sympathy.
And of course electrical networks are a classical example of faults expanding, there may be security devices to limit the failure to the device or even the local electrical network, but sometimes those failsafes fail, and that's what causes wide blackouts like the one in Spain recently.
Electric devices (including infrastructural), are connected by networks, electricity can "travel" downstream just as easily as it can travel upstream unless specific mechanisms are engineered to avoid that, sometimes those mechanisms fail.
A power surge generated by oversupply in a power plant can affect burn both "downstream" devices that consume its electricity, as well as lateral unrelated networks, since they are all connected in a network.
I'm sure I'm wrong in a lot of points, but pretty confident in the fact that failures can flow upwards and cause blackouts downstream.
The classic real example of actual transmissibility was the Zip drive click of death. A bad drive would damage disks, which would in turn damage another drive they were put in. The case was rarer than people thought but did happen. https://www.grc.com/tip/codfaq4.htm
Unfortunately, it was replaced with new hardware (identical RoHS solder problem), so it failed again relatively quickly (to crickets from Apple).
I didn't buy another Apple laptop until the M3 MacBook Air debuted (what an impressive machine — particularly battery life!). I already know this machine is "disposable" (i.e. delicate), but heat is definitely not going to pop any circuits (like the MBP from a decade+ ago — that thing got HOT).