3 watt Luxeon lumens output?

About a week ago, a friend of mine who is a SureFire fanatic, did a modification for me, of an old L1 bezel he had in a used parts box. His original bezel had a 1 watt Luxeon with a Batwing configuration. He replaced it with a new 3 watt Luxeon with a Lambertian configuration.......I somehow forgot to ask him the bin rating. My measurements of this bezel, taken with my ammeter across the battery negative and the body of the light, were 850mA at 3.0V.

In the page below, the spec for this LED is 65 lumens @ 3.70V and 700mA, or 80 lumens @ 3.90V and 1000mA. When I compare the output of my original bezel with the bezel he modified, I see a slight increase in brightness and a very white beam, instead of the slightly purple one indicative of the 1 watt Luxeon used in the L1. The L1 is stated to emit about 20 lumens in the high beam. Why isn't the output of the new bezel, similar to the figures on the Luxeon webpage below?
http://www.luxeonstar.com/item.php?id=381&link_str=90&partno=LXHL-PW09

I know that there is some circuitry trick preventing the output from reaching 65 lumens, but I don't know what that is or why this was done? Can anybody clear this up for me?

I am not entirely certain of what is happening. But I can reason a couple of things out.

The lithium cell you are using is giving 850mA @ 3.0V. This translates to about 2550mW and will translate back to about 689mA @3.7V. Obviously this is already below the spec of 700mA @ 3.7V, plus we have not yet accounted for the efficiency losses in the converter circuit.

So just to start out with, you are unable to drive the LuxIII at spec. The circuit itself probably has some drive limit to prevent thermal runaway.

Thank you. I understand your calculations. But how can the output be around 20 lumens instead of 65 lumens, if the current is only 11mA less than spec?
Why apply so much regulation, that a Luxeon that is capable of 65 lumens, is actually delivering 20 lumens? This may be a question for Lumileds, right?

Your original configuration has a driver board set up to run a 1W LED. Putting a 3W in doesn't change anything on the driver board. The driver board doesn't know you changed the LED, so it still will drive the LED to the original LEDs specs.

If you don't change anything on the driver board, you'll still be at approximately the original lumen output specs no matter what LED you put in there.

One good thing, at least now you have nice white light and no more purple. /ubbthreads/images/graemlins/smile.gif

you are measuring the current draw of the battery and not the current going to the LED. As the converter has some inefficiencies AND has to ramp up the voltage the current to the LED will be less, it should be about 360mA for the L1 which is constant current regulated in high mode.
The L1 is made for a Lux1 and the typical current there is 350mA. That's where the 20lm come from.
Now with the HD LED with maybe a slight increase and efficiency AND a different beam pattern and a much better color rendition than the LD you have a bit more brightness, but still 350mA to the LED.
bernhard

Thank you both, MoonRise and Bernhard. You gave me a huge education about the way SureFire designs their lights. It now makes sense why people always prefer the regulation components to be in the bezel, instead of in the body of the light. In this way, a person can benefit from the enhanced Luxeon LED in an LED Conversion head, and not be limited to the design parameters of the original light. For you Bernhard, I will rig up my meter with the bezel detached from the body and the tailcap all the way in, in such a way as to check the current and voltage going to the lamp in high mode. I will report back to you, STAT! Even if the output is still around 20 lumens, I am enjoying that the light is pure white and has that wonderful high dome beam pattern. The Lambertian configuration is a vast improvement over Batwing configuration. Things are looking up at Lumileds..............

And IIRC, the Luxeon data says that this is for 20°C die temperature or something like that.
Which you cannot establish in a real life light.
Does anybody have the exact data at hand?