Even though I’m using what seem to be good-quality parts, one of the WS2812 RGB LEDs in a Glass Tile frame died:
It passed the Josh Sharpie Test:
After building the third Glass Tile unit, one of the LEDs didn’t light up due to an easily diagnosed problem:
A closer look:
Shortly thereafter, the Nissan Fog Lamp developed an obvious beam problem:
The WS2812 had the proper voltages / signals at all its pins and was still firmly stuck to the central “heatsink”:
It also passed the Josh Sharpie Test:
I’m particularly surprised by this one, because eleven of the twelve flex-PCB WS2812s in the Hard Drive Platter light have been running continuously for years with no additional failures.
The alert reader will note the common factor: no matter what substrate the LED is (supposed to be) soldered to, no matter when I bought it, no matter what it’s wired into, a WS2812 will fail.
They’re all back in operation:
Although nobody knows for how long …
Obviously, it’s time to refresh my programmable RGB LED stockpile!
The Smell of Molten Projects in the Morning 
I’ve been having better luck with the new WS2812E, though of course no long-term results yet. Less lens fractures during reflow
Yikes! At least you know which ones failed right away, cold comfort though that may be.
I don’t trust an item from the usual eBay sources to be anything in particular, but if it has a white LED inside, it’s probably not a “2812” module. Sheesh & similar remarks.