The big failure arguably is not testing the firmware sufficiently (or building in recovery algorithms) to necessitate a reset procedure in the first place.
The big failure arguably is not testing the firmware sufficiently (or building in recovery algorithms) to necessitate a reset procedure in the first place.
As to firmware, cosmic rays for example randomly flip bits. Which means for any large scale deployments you want a reset option.
Edit: a source https://www.researchgate.net/publication/4082030_Flash_memor...
any solution has to cost less than a penny ...
A capacitor that is kept charged during normal operation, and that powers the reset circuit which you close with your metal pin once you unscrew the bulb. The circuit flips some bits in persistent memory of the bulb.
It’s a heat sink that gets hot to to the touch, though not that quickly as your talking ~10w of waste heat. Sill several ways to do a reset switch just after power is off.