It occurred to me that much of the variability people find, and many of the seemingly odd little failures, could be from the past history of the semiconductor parts.
If they have not been carefully handled at every step, they may be wounded.
I hope this doesn't start an argument that anything that doesn't kill you makes you stronger. Maybe that's true for some people. For electronics, it's clearly not. But our experience isn't a trustworthy guide on this.
I know I've met plenty of "electronics repair techs" who carelessly don't use electrostatic protection, believing that if the device still lights up, then they couldn't have hurt it by carelessly ignoring ESD protection. It's self-deception. Fools a lot of customers too.
Back when computer stores had repair benches, I'd always insist on seeing where they did their repair work before dropping something off. Far too often *cough*CompUSA*cough* the bench had no antistatic mat, there was no static wrist strap, and blank incomprehension when I asked about precautions.
Been burned, got shy.
Many of the devices I get mail order come in packaging that's not safe (ziplock clear plastic, or nothing).
A reference worth a look:
[PPT]SEMICONDUCTOR DEVICE SENSITIVITIES http://emp.byui.edu/fisherr/ESD/StaticSeminar.ppt [quote] Intermittent / Walking Wounded. Device is operational, but erratic and will cause additional service calls; Represents 90% of ESD failures. [/quote]