On Sat, 19 Nov 2016 22:15:59 -0500, "J. Clarke"
wrote:

In article ,
says...

Simple - I used to design and test semiconductor material -
I started in the mid 70's and went pro in 1980.

I remember some of the first high speed CMOS HMOS and NMOS
that ran really fast but shorted power to ground in the output
transistors in the process. It seemed ok in theory - but after
millions of cycles the part melted the epoxy off the IC !

All semiconductor heats up. It is conducting current. Current flow
causes vibration heating.

But in actual tests is that the major failure
mode of commercial LED lighting?

Commercial or residential? I know it's the failure mechanism in
residential LED lighting. Commercial may be better. Or not.