OK geeks - regale me with the trials and tribulations associated with discovering the relative lighting efficiency of various modern bicycle or flash lights?

I have very limited experience - and I am sure that someone has compared their lights with respect to "brightness" and current draw. (battery run time)

My own contribution is in discovering that my Stella 120 "AA" tries to grab 700mA from four "AA"s and continues to appear bright until batteries run down to delivering 500mA.

This "test" has been repeated three times - using three brands of batteries and (2450-2600) they all power the light at these levels up to three hours. (+/- 15 minutes)

Is this consistent with other "pretty bright" flashlights or bike lights? What have you discovered?

Your Stella is exhibiting that behavior because it is regulated. If you're seeking the highest luminous efficacy, look for lights that have the Cree XP-G LED. These produce the highest amount of light per power input. As reference, per Cree's spec sheet, the XP-G LED is rated to produce about as much light as your Stella (120 lumens) at about half the current (350 mA), which means you'll likely see almost double the runtime.

But lumens isn't really all that matters. Like the MHz/GHz wars in CPUs, lumens isn't really a good measure of how useful a particular light is. It is useful in comparisons between similar lights; but to be useful to a cyclist, the LED must be paired with the correct optics that can shape that light energy into a useful beam. For example, a concentric and symmetrical beam may be useful for mountain biking where you need to watch out for low-lying tree branches; however, it is less useful for general road riding where an asymmetrical beam with vertical cutoffs is better because this beam concentrates the light energy lower and onto the road.

Try browsing the Candlepower forums. There's a lot of good information there if you're a light geek.

I have a River Rock 6V headlamp that uses 2 cr123 lithiums. 136 lumens. Will run about 4 hours or so (no dimming options via switch) before it goes into a much dimmer mode (probably about 40 lumens or so) designed for safety with the fussy lithiums. It will run in that mode for about 2 hours before it finally dies.

The brighter the LED, the less efficient it is, in general. Lumens per watt goes down as lumen output goes up.

Thanks for the info. The combination of rainy weather and underemployment leaves Richard Cranium with far too much time on his hands to investigate these matters.

The purpose of my investigations was to engineer a cheap "ball cap" led light with a custom remote battery pack that would run it at near "full gate" all night.

In this other case, I substituted 4xAA @1.2V NiMH for 3xAAA 1.5V Alkiline to see what would happen. The Energizer light is supposed to be 1 watt led and appears to produce something near 40-60 lumens. I base this on the fact that it has a bright spot near the intensity of my Stella 120.

However, as one of you notes, this light is all jazzed up with a PC board and extra options for other lights on its body. The main led draws over 30 mA the flood and night modes a little less. Anyway, I figured if I sub in "AA"s I would get better run time. I'll check out the CP forums.