Doesn’t look like much, does it? It’s an ordinary blue LED that I used for the upper colon dot in a clock. Worked fine for a few dozen power-on hours, then it turned off a bit after 6:00 pm one day. Back on an hour later, more or less, then off again by the next morning, back on again, off again.
Might be a software error, as each colon LED is a separate TLC5916 display driver output. Might be a soldering problem, as my board doesn’t have plated-through holes. Might be (shudder) a burned-out transistor inside the TLC5916.
When it’s off, VCC appears on both sides, within a few tens of millivolts.
Resoldered the joints, after which it worked for a while. When it’s on, voltage measurements look normal: about 3.5 V drop across the diode and 1.5 V across the driver transistor.
No obvious code problems, but, then, code problems are never obvious.
Finally the thing stopped working for a few hours. I unsoldered it and there’s no continuity: it failed open. Peering deeply inside with a microscope shows nothing unusual: the flying gold wires look OK, the bonds look flat, and the chip has no burn marks.
Just a bad LED, I suppose. It’s surplus, of course, but that doesn’t mean much these days; there’s a lot of surplus going around.
Soldered in a replacement from the same batch and it’s all good.
So far, anyway.