I dont have that equipment, nor the expertise to use them, but i have used a microscope before to examine dies. You can probably just use a regular microscope for this scale, and The LED even has a clear lens to see right into it. I still wont be bothering to do that. Instead I built the thing, and have been running it as a test to see how long it will take to fail, and it has shown no signs of failure yet.
This is not something you can do with an optical microscope. You need a scanning electron microscope to find out about the conductivity of the material. The effect of excessive current in an LED is that there is a depletion of charge carriers which reduces the light output over time. Waiting for a total failure with one sample is going to tell you absolutely nothing about your design. Most people do not understand about statistics and probabilities which is why the Casino owners are so rich and poor people continue to buy lottery tickets.
Its been months with no sign of dimming, ......
OK so lets see how you have tested that statement. I assume you have simply looked at it, I assume you have not taken any measurements of the light output. The eye's response is non linear, it is almost impossible to tell the difference in brightness of 25% over a month. That is why we have instruments to measure things. So I am not impressed with your test. If however you have conducted a real test I would be most interested.
Now dear henry:-
Engineering isn't about "you cannot do that", or "that's wrong". Engineering is about knowing why you cannot do it so you know when / where you can do it.
You may be surprised to here that I absolutely agree with that statement. However there are several points that you keep forgetting.
The biggest error you make is this forum is for beginners, they need to keep things as simple as possible because they know little and they are easily confused with esoteric arguments especially when you leave out all the caveats that your arguments normally involve to make them work.
Lets look at one of your off trotted out remarks:-
Leds, especially high brightness leds, behave far more like resistors (at high current levels) than like diodes,
First off here we are not talking about high brightness LEDs, and second any curve if you look at a small enough section of it looks like a straight line. It is not a matter of common sense it is a matter of knowledge of electronics, you seem to have a bit of the latter but non what so ever of the former.
the resistor really is there to provide some negative current feedback
Nearly right. In fact the current produces a negative voltage feedback to in effect reduce the applied voltage.
the datasheet lays it out clearly for you.
Indeed it does, things like internal impedance of the driver will limit the current to a non infinite value, but it is odd that you are willing to believe the data sheet over some things and totally reject the information over others.
Engineering is about being in control of what ever it is you are engineering, be it a jumbo jet, the world's highest building or a humble LED. It is about making decisions and trade offs.
With today's planned obsolescence culture then you indeed might want to design a circuit that only has a life slightly longer than the warranty you give it. You might however want to design a circuit that has a life significantly longer. In those cases you would not even drive the component at the rating given in the data sheet you would derate the device by 80% or even more.
If you want UL approval on anything then the capacitors in the power supply have to run at less that 80% of their voltage ratings even though you might make the "engineering decision" to run it at the maximum operating voltage.
Engineering is about making those informed decisions. So yes by all means run an LED with no current limiting resistor after all they do cost $0.001 each. The LED and driver will be in land fill sooner but hey the economy ticks over faster because people have to buy more stuff. But people have to know this stuff. Like simple Hippynerd, he has absolutely no idea what he is doing and is happy if his one sample hasn't stopped working after a year, but you can't say he is making an engineering decision, he is being fooled into a simplistic way of thinking. Someone has done a good job on him too as he clings to his misguided beliefs.
So dear henry, while you like to think you are much cleverer than most people here confusion and obfuscation are no substitute for real engineering.
This link tells about the fundamentals of reliability:-
http://www.ece.cmu.edu/~koopman/des_s99/electronic_electrical/This tells a little about how LEDs fail:-
http://en.wikipedia.org/wiki/List_of_LED_failure_modes