Semiman,
I am using an Ocean Optics Spectrometer coupled with an Optronic 6" Integrating Sphere:
Now for calibration of the system, I used a calibrated incan source from OceanOptics and its accompaning "lamp file". The means of introducing this calibration lamp to the IS no doubt introduced error and some light loss. As a result, I am confident that my absolute measurements are off but by how much, I have no idea. The spectral response as well as relative comparisons however should be very close to reality.
In the case of the LED's measured above, they were all mounted on my "E-Screw" heat sinks and introduced to the IS with the same adapter (shown on the right in the pic below) :
These heat sinks thread into the adapter and then the adapter drops down into the port on the IS. The IS is aluminium and the port is aluminium. I have not noticed any of the test pieces even getting really warm to the touch after some significant on time while mounted to the IS. Ambient temperature has been between 65 and 75 F for most of my testing and in the case above, it was probably in the high 60's for all tests.
That is a typo on the Seoul 2.5W LED. It should be 4.0 V
Now this test was interesting and I find it instructive to consider. However I want to point out that it is based on single samples of the various LED's and that variations in flux within a bin not to mention across the whole variation beyond specific bins significantly devalues these data points.
On an aside, this system was expensive to say the least and yet it would be considered by many to be a sensless toy and not worthy of use. These would be the folks that point out the shortcomings of $20,000 systems. I have found this humble system to be very instructive and the repeatability as well as predictible responses I have seen have supported my perception that I can glean valuable insight and information from its use. The integration sphere by its very nature allows for a much better and more comprehensive measure of light than say the meassures one gets from a lux meter. :shrug:
EDIT: I might as well throw in a table and graph I did on a UW0K sample which I find even more useful than the previous table.
I need to point out that my digital bench supply is limited in resolution and control of current to 10's of miliamps and this is why the 50 mA measure is skewed. The actual current to the LED could have been as low as 46 mA or as high as 54 mA. I have another BK Precision bench supply that has resolution to three places but although advertised as 0 - 6 amps, the lowest I can set it is about 165 mA as I recall! Man, I called them up and read them the riot act! How can you call it 0 to 6 amps when you can't get close to 0?!?! They should have rated it .200 to 6 amps!!