I'm trying to figure out how strong an LED I salvaged is. It looks like a garden variety 1 watt, but the board it was on was marked XS-116. I'm betting it is a Cree XS series, though I never found that part number. If I was to try to find its optimal drive current, how would I know I'm near my rating? Could I go by voltage across the LED? What voltage would I go for, 3.6 volts?

 

The ultimate limit is junction temperature but that's difficult if not impossible to measure so the manufacturer relates it to something that is measurable such as case temperature or solder point temperature. Look this information up for a design or designs that match the form factor and apparent thermal mass of your unknown LED, then monitor the temperature while ramping the current up. This should result in an experimentally derived current value that's close to one of the data sheet values. Use that closest matching data.

 

LED brightness is determined by the driving current.
But LED can be easily damaged by over-voltage or excessive heat (as a result of excessive current).

A relative inexpensive tester for LED :

http://thumbs4.ebaystatic.com/d/l225/m/mFKZvGqkILPs-AEPx5aJe2w.jpg

 

LEDs are not voltage devices, it is only current that matters.

 

LEDs are not voltage devices, it is only current that matters.

While I knew LEDs were dictated by current, most diodes I have bothered would still have a voltage that correlates to its current, although not directly via Ohm's Law.

Would the package tell me all I need to know about the LED? All I have seen are good for a junction temperature of 150 degrees celcius.

 

The ultimate limit is junction temperature but that's difficult if not impossible to measure so the manufacturer relates it to something that is measurable such as case temperature or solder point temperature. Look this information up for a design or designs that match the form factor and apparent thermal mass of your unknown LED, then monitor the temperature while ramping the current up. This should result in an experimentally derived current value that's close to one of the data sheet values. Use that closest matching data.

A lot of LEDs have fallen victim to that method, although I have used a big heatsink in the experiment. Do I need to go without the heatsink and measure the back?

 

If the type of LED in question normally employs a heat sink then use one. The method I've described is for experimentally characterizing an unknown LED, not a method for long-term operation as you seem to misunderstand.

 

as stated above..
Run it at a current (use your best judgement if you think its a 1 watt follow recommended current for those..probably 350mA) for minimum 30 minutes to allow it to stabilize and measure temps the whole time as close to LED as possible add 20 or so deg C and assume thats junction temp.. Keep it under 85 deg C for long life.
If its way under temps.. crank up the current
If its too HOT ..crank it down.

But its a single low cost LED.. not worth the time to even attempt to characterize it IMO..

Or just post the picture.. More than likely you can tell by just looking at it..