White LED lumen testing

[bizzybody]

Update: The colored LED efficiency charts are now separate files. They actually use less space than the one big file since I was able to convert most of them to 16 colors.

If the graphics app* you're using has a feature to convert to Indexed Color with Exact pallete, saving to .PNG or .GIF can make really small files of huge graphics.

Fer instance, I've been doing some texture maps for 3D models at a 2048x2048 pixels, in 2 colors. Saving to GIF with an exact pallet crunches the 12 megabyte image down to around 47-67 kilobytes with no loss of quality. More colors would of course increase that, but with the lossless compression on chart and graph type images with large areas of the same color, it wouldn't be much.

*Photoshop 7.01 and/or Photoshop CS2 (9.01) is my tool of choice.

 

[jtr1962]

I added the BestHongKong 5mm warm white LEDs to the list. You might need to refresh your browser to see the updated graphs. The relevant spreadsheets were added to the .zip file linked to in the first post of this thread.

 

[jtr1962]

Cree 7090 XR-E bin P4 (purchased November 2006)

I purchased these from CPF member Erasmus. These are tint bin WH and flux bin P4 (80.6 to 87.4 lumens at 350 mA). The color temp of these seems a close match for my 5000K fluorescents which makes the WH bin perfect for interior lighting although perhaps a bit warm for flashlights. I measured the 50% beam angle at 95.3° which is somewhat wider than the 75° given on the spec sheet. Despite the out of spec beam angle, the overall flux at 350 mA is 85.67 lumens which is within the range of the P4 bin (and also a verification of the accuracy of my methodology). Vf at 350 mA is 3.25V and overall efficiency is a very impressive 75.37 lm/W. Even at 700 mA efficiency remains above 60 lm/W. Luminous flux at 500 mA is about equal to the Lamina Ceramics BL-2000 arrays but the power input is only around 1.7 watts instead of the 4.7 or so watts used by the Lamina array. Although these obviously aren't a 20 mA part if they were they would break previous records by a huge margin, coming in at nearly 112 lm/W. There's really nothing more to say about these than hasn't already been said by me and many others. They represent a huge step forwards in LED technology.

Here is the efficiency versus current chart:

 

[EngrPaul]

Great information, Joseph. WOW the Cree can be run efficiently down at 25mA !

Thanks for your one-of-a-kind comprehensive testing!

 

[Erasmus]

Great results JTR!

 

[frenzee]

Thank you very much for this info. I wish all manufacturers would provide this data on their data sheets. Hopefully one of these days you will find the time and patience to do the same kind of testing on red and amber LEDs.

 

[ViReN]

Now That's Impressive

 

[jtr1962]

Thank you very much for this info. I wish all manufacturers would provide this data on their data sheets. Hopefully one of these days you will find the time and patience to do the same kind of testing on red and amber LEDs.

I did test some colored LEDs (results in post #19 of this thread). I decided to discontinue doing so after finding out how inaccurate my light meter was with colored LEDs in the Light Meter Benchmark Test. If I ever have suitable equipment available to accurately test colored LEDs I would certainly reconsider doing so.

 

[milkyspit]

Hey Joe, :wave:

Great data as always. You know I'm a big fan of your work! :bow: :bow:

Question: for the Cree, was 25mA the lowest current you tested, or did performance really fall off a cliff below that as shown in your graph?

 

[jtr1962]

Hey Joe, :wave:

Great data as always. You know I'm a big fan of your work! :bow: :bow:

Question: for the Cree, was 25mA the lowest current you tested, or did performance really fall off a cliff below that as shown in your graph?

Many thanks for the compliments! I'm just glad this data is useful to so many people.

Actually 20 mA was the lowest current I tested at. It looks like 25 mA on the graph thanks to the way Excel plots the data. For the same reason the performance also looks like it falls off a cliff.

I didn't bother testing under 20 mA because of the difficulties of measuring the intensity. The Cree is a wide angle LED. At 20 mA I was only getting 2.8 lux at 1 meter. Since my light meter only resolves to 0.1 lux on the lowest scale this meant at 20 mA my granularity was already close to 4%. Also, NewBie tested these at lower currents and from his data it seemed that efficiency more or less peaked at 20 mA and then dropped off at lower currents. 20 mA is a little under 6% of rated current. I've found that with 5mm LEDs I get similar results where the efficiency peaks at 5% or so of rated current although in this case rated is 20 mA. So for 5mm LEDs I usually test down to 1 mA but for power LEDs like the Cree I usually go down to around 5% of rated.

BTW, you can access the spreadsheets containing the raw data for my power LED tests here.

 

[EngrPaul]

I'm sitting here visually comparing the beam patterns of "BestHongKong 25,000 mcd 20° 5mm UWLC series white" BUWLC333W24BA25 against the "Jeled 50,000 mcd 20° 5mm white"

All I have to say is the BestHongKong has excellent distribution and overall color. If you are going to upgrade a 1-LED light, this would be a very good candidate.

On the other end of the spectrum is the JELED. It's splotchy, with regions of pi$$ yellow and ocean blue. The only place I found for them were behind the frosted bezel of LED nightlights and all mixed together in multiple-LED cheapie lights.

Neither of them have tapered heads, which prevents them from being a drop-in for many applications.

 

[3rd_shift]

Is there a way that you can remove the lense and retaining ring from a cree xre to measure how much light is getting out without those on it?

 

[Pinter]

JTR, thank you for the XR-E test. Overall lumen value is fitting great in the given bin range.

I measured the 50% beam angle at 95.3° which is somewhat wider than the 75° given on the spec sheet.

Just wondering that this huge difference is general issue, or just an individual misalignment of the embedded optic?

Do you have other Crees from this batch?

 

[Erasmus]

Here's another table belonging to the Cree XR-E tests :

Once I can get my hands on Q-bins, these will to straight to Jtr1962 so he can measure the output! Thanks Jtr!

 

[Opto-King]

Hmm, according to your testing and to SSC's datasheets for their P4 LEDs the P4 LED in U binning is better than the Cree LED.


How is this possible if they are both using the same die?

 

[SpeedEvil]

The die is a very important component - it's hardly the only one though.

You've then got the thermal connection to the heatsink, so it doesn't heat up too much, and stop working.
The electrical connections, which are a compromise in some ways between light output, ease of manufacture, and resistance.
The refractive index, and design of the bits just around the die, to optimise on extraction of light from the die.
Manufacturability (hence cost).
The phosphor coating formulation, thickness and evenness of that coating.
Then you've got the lens, and how it all ages over time, which is probably the hardest bit.

 

[NewBie]

Hmm, according to your testing and to SSC's datasheets for their P4 LEDs the P4 LED in U binning is better than the Cree LED.


How is this possible if they are both using the same die?

You will notice the CREE hit the lm/W target even with actual derating for temperature.

In other words, the test was conducted on the part with the die heated, as it would be in a flashlight.


It will be interesting to see how the Seoul Semiconductor part actually does, when ran at actual temperatures.

The LumiLEDs parts, you have to de-rate from their datasheet/binned numbers due to die heating to get to the actual real life lm/W.

 

[NewBie]

Cree 7090 XR-E bin P4 (purchased November 2006)

I purchased these from CPF member Erasmus. These are tint bin WH and flux bin P4 (80.6 to 87.4 lumens at 350 mA). The color temp of these seems a close match for my 5000K fluorescents which makes the WH bin perfect for interior lighting although perhaps a bit warm for flashlights. I measured the 50% beam angle at 95.3° which is somewhat wider than the 75° given on the spec sheet. Despite the out of spec beam angle, the overall flux at 350 mA is 85.67 lumens which is within the range of the P4 bin (and also a verification of the accuracy of my methodology). Vf at 350 mA is 3.25V and overall efficiency is a very impressive 75.37 lm/W. Even at 700 mA efficiency remains above 60 lm/W. Luminous flux at 500 mA is about equal to the Lamina Ceramics BL-2000 arrays but the power input is only around 1.7 watts instead of the 4.7 or so watts used by the Lamina array. Although these obviously aren't a 20 mA part if they were they would break previous records by a huge margin, coming in at nearly 112 lm/W. There's really nothing more to say about these than hasn't already been said by me and many others. They represent a huge step forwards in LED technology.

Here is the efficiency versus current chart:

Your testing on an actual part is still hitting CREE's claims even with the die not at 25C, very impressive.


Very nice, your P4 did even better than my P3:

My peak lm/W on the XR-E was 20mA.

 

[NewBie]

.

 

[jtr1962]

Is there a way that you can remove the lense and retaining ring from a cree xre to measure how much light is getting out without those on it?

Unfortunately I need all 6 pieces I purchased for projects (and I don't want to risk destroying one by removing the lens). However, if anyone wants to send me one without the lens I'll be happy to test it.

Just wondering that this huge difference is general issue, or just an individual misalignment of the embedded optic?

Do you have other Crees from this batch?

I haven't tested the other 5 I have. I can maybe do a rough check on the beam angle to see if the others are out of spec. In the end I don't think it matters much as far as total lumen output goes. These are factory tested, presumably in a small integrating sphere, so the lumen output will be within the bin specification regardless of beam angle.

Hmm, according to your testing and to SSC's datasheets for their P4 LEDs the P4 LED in U binning is better than the Cree LED.

How is this possible if they are both using the same die?

The P4 U bin isn't a whole lot different than the XR-E Q2/Q3 bins. Neither are out yet so I don't think SSC has accomplished something Cree hasn't. By the time P4 U bins are out we'll also have comparable XR-E Q2/Q3 bins.