News for hackers. Stuff that matters.
Re:Chip aging becomes design problem (2018) by boznz on Wednesday January 12, 2022 @06:46PM (#29914510)
I'm an EE and never had a chip fail when working inside its recommended operating parameters and properly protected and I have devices in the field over 20 years, I do see failures when poor soldering or electrolytic gone bad compromises the power supply. My main worry is some flash memory is only guarenteed minimum 25 years, Not sure how they determined this but yet to see any issues.

I would hate to be on a generational spaceship or the first settler on another planet with currently designed commercial stuff.

Re:Chip aging becomes design problem (2018) by Robotbeat on Wednesday January 12, 2022 @08:42PM (#29915760)
Flash is indeed a weak link. I use to put together embedded Linux systems out of commodity server components 15 years ago, and over the years we lost many flash chips (altho they would often continue functioning in a degraded state… read only) due to the high logging rate of Linux combined with the low write endurance of typical usb flash disks. You can get more expensive single level cell (or pseudo-single level cell) chips now that can last an extremely long time.

Old Tesla Model Ses had this same problem with the infotainment system flash, but Tesla will fix it for you for free now since the infotainment system contains some features NHTSA considers safety essential (wipers, etc).

Re:Chip aging becomes design problem (2018) by piyh on Wednesday January 12, 2022 @10:11PM (#29916353)
I'm only buying high endurance memory now. I've had enough cheap usb and sd cards die on me. Losing my data is not worth saving $10.
Re:Chip aging becomes design problem (2018) by ece on Thursday January 13, 2022 @12:47AM (#29917278)
Good flash atleast goes readonly (Sandisk USBs in my experience), bad flash, not so much. The controller is probably playing a part as well.