Grumpy_Mike:
- A shortening of the life of the component, in other words it will fail sooner than it otherwise would. This is due to the mechanical stress on the chip's bonding wires causing eventual fracturing and depletion of charged carriers in the semiconductor substrate. There is also a problem of localized heating when the thermal conduction will not allow enough heat to escape to prevent over temperature of the junction.
Yep, this is one of the expected outcomes. As I mentioned above even a lifespan of 30-40 flashes would have been fine with me for my purposes.
Grumpy_Mike:
2) A reduction in the maximum brightness of the LED. I know you will say it did not reduce in brightness but if your only measure is your perception of how bright it was then that is hardly very objective. It is well known that people can't remember the brightness of a light over a few seconds let alone several months. To properly test this you would have to make carefully calibrated measurements with specialist equipment.No one is saying you have to have a resistor but you have to have some form of current limiting even on a very narrow pulse. It is better if that limiting is under control rather than the vagaries of a particular component.
I didn't mention it, but in my experiment I was measuring the light output of the pulses, and yes there was about a 20% reduction after 10,000 flashes. As I mentioned, if the lights were bright enough for a small enough number of them then I would have been ecstatic with that outcome. I fully expect that as the bulbs get brighter and even cheaper I will eventually replace my existing air spark gap on quartz glass tubes (home made). Indeed as they become bright enough, I will certainly be able to use a better design that can limit current and lengthen the lifespan. But as I said, the current state of the devices isn't there yet.