Test parameters:
4 LEDs were tested, 2 Nichia NSPW500CS UB2 and 2 ETG/Cree 15degree "snow". In all tests, the Nichias were run at 60mA while the Crees were run at 30mA. The LEDs were mounted side by side on a PCB within 1 inch of each other. The entire rig was placed in a plastic enclosure to reduce any air currents over the test fixture.
The current levels were selected for these reasons: The current the LED is driven in either flashlight varies depending on items such as regulation scheme used, vf of the LED, battery voltage, temperature, etc. For the Arc-AAA, it averages about 45mA to the LED for both the Premium and the Standard versions. In some cases, this may be higher and as the battery voltage drops, this current will drop as well. I decided to use 60mA so as to make the durability of the LED we use in the Arc-AAA very clear and easy to understand. The 30mA for the Cree LED is a guess really. To my knowledge, Peak continues to keep that data to themselves. My observations of the few units I have seen show a wide current variation from to unit. Some could be only running at 25mA while others are over 40mA. Nailing this down is not my responsibility. I selected 30mA to serve as an example of how the LED would perform at a typical current for this type of flashlight. Even if Peak chose to publish a current spec, the best I could think they would be able to do is to provide a wide range of possible currents for a particular model.
Two tests of 2 LEDs each were run. The test fixture utilized a precision power supply feeding dual current regulators each driving an individual LED. LED output was measured using a CDS cell in series with a common power supply. For the second test, the LED brands were swapped in the test setup to factor out any variations in the test rig.
Remember, the response of a CDS cell is logarithmic. That means the differences between the plots are greater than what they might first appear.
Since the LEDs were tested in pairs, any variations in ambient temperature, power supply drift, etc should be normalized between both LED measurements.
This graph was generated in excel. Although the original plots were separated by differences in brightness between the various LEDs, all plots were normalized to the same start value to show relative changes more clearly. Remember this when you read the graphs. And as always, if you doubt the data, test it yourself.
Your comments are welcome of course.
Piper: Yes, using a 1.7v lithium will reduce the lifespan of the LED in both flashlights. However in the Arc, we have tested the effect of the lithium on our system and do not feel it will make an appreciable effect. We will continue to offer our standard replacement guarantee.
Solstice and others: I also want better LEDs. 200lm/w, perfect color coordinates, infinite lifespan, etc. In the meantime, I will find the best for my goals and use those.
Btw, not trying to stir up controversy or anything. The snow LED does provide some benefits. It is nice that the customer is free to decide what is important to their particular use. The lights I design are not for everyone. That means, not everyone is going to like them.
Peter