Emitter Test Results pt4: XML2, XP-E2, XP-G2, and Nichia219! (along with sinkpad tests)

Folks,

Long time no see... Since most of you are now aware of my testing procedure, I'll be brief here. The following is a test of the Cree XML2 on a standard aluminum 20mm star, along with another xml2 mounted on a copper sinkpad. (member vestureofblood was kind enough to send me out these for testing).

Graphs speak louder than words, so here they are. The first one is the XML2 on a standard star, followed by a comparison graph between a standard aluminum star and a copper sinkpad. The third represents the associated vf of each.

There you have it. Max lumens on regular aluminum star came out to 1259, while it was 1770 for the sinkpad ( at much higher current). The sinkpads do well at dissipating the heat - During an extended run at 5 amps, lumens only dropped by 50 before stabilizing out. Also, while the output is quite impressive for the copper mounted XML2 (and xml2's in general), please note that there's a good chance max lumens won't be realized on a single li-ion due to the high vf.

UPDATE 3/10/2013

The last round of tests are complete and the data has been collected. Here, we'll see graphs for the XP-E2, XP-G2, and Nichia219. The two Cree's have been tested on a standard aluminum star and on a copper sinkpad, and the graphs will reflect that data.

There you have it. I'll let the data speak for itself, but suffice to say that with the availability of these sinkpads, I've milled my last copper pcb. I did get a chance to do a direct comparison between the same emitter (XPG2) directly mounted to a large milled copper pcb and then on a sinkpad. The difference between the two was anywhere from 5-10 lumens in favor of the milled pcb. Not enough difference to be noticeable in anything less than an integrating sphere, much less real world use (and definitely not worth the added effort).

My final addition to this emitter testing Opus will be to make one large graph with all of the new emitters present for easy comparison. After that, I'll be retiring my testing setup for it's a big pita and I'm finding I don't enjoy it as much as I used to.

Enjoy the data,

-Match

Great work Match, so now it looks like the vf of the XM-L2 is very similar to that of the Nichia 219. :(

Thanks Match,I love these "where the rubber hits the road" tests.The sinkpad really makes a difference and is clearly the winner over the aluminum mcpcb.It seems XML-2 is best suited in a multiple cell configuration because of the higher vf.Very valuable info for modders,and out of curiosity what is the bin of XML-2 in this test?

Yes, but isn’t Match is sinking copper to big chunk ’o copper? Results would be less spectacular if the copper sinkpad was sinking to a chunk ’o aluminium like in most flashlight bodies.

Thanks for the valuable information Match.
I guess amc7135 drivers are out at high currents. The battery wouldn’t be able to stay above vf of the led and the .12v overhead for the amc7135’s, at least not for very long anyways.

Great work and Thank You Match :slight_smile:

Thanks match, been waiting for a test on these.
Would the higher vf on these be accounting for the 2.25a draw using a 3.5a driver? Or is there something else coming in to play here?

Thanks again, your tests are invaluable to us here

Thanks for the testing match. Interesting results.

Are you using a single cell driver?
And if so, what battery are using?
Update: Pulsar I found your thread with the mag, I’ll take the discussion there.

Great information as usual, thanks!

Sweet mother of pearl. As always—a big “thank you” for the data, match! I’m looking forward to how the XP-E2 fares. Will it have a higher Vf than its predecessor, like the XM-L sequel? Or will it be notably lower like XP-G2 vs. G1?

I know. It looks like high currents will call for 4xNiMH in a Mag. All the better for heatsinking, at least.
On the upside, the XM-L2 hovers around 62 lumens/watt at nearly thirty watts, >1.5A/mm^2, and delivers >400 lumens/mm^2. That’s crazy!

Excellent work Match! This puts my ‘XM-L2 on copper SinkPads BTU Shocker’ mod very high on my list :wink:

Pretty Awesome that you can see the increase in Vf (roughly 0.15V at 3.0A) with the emitter on the Copper vs the emitter on the Aluminum. This is almost certainly attributed to the lower operating temperature of the emitter on the copper. Some of the difference, however, is likely just emitter to emitter variation. I’d love to see what the difference in the Vf to Current curve would look like for the same emitter soldered to an Al vs. Cu star.

You should, in fact, be able to work backward and figure out what the actual die temperature difference is between Al and Cu. It’s easy to calculate - 1.6mV difference per degree C.

PPtk

+1
I agree; Sinkpad’s design is definitely doing it’s job! I’m impressed.

Thanks for sharing…

This is how I would rate the upgrade by looking at your graph.
Up to 1,5amp, no difference
1,5-2,5 minimal difference
2,5-3,0 some difference, about 10% at 3A - or 100 lumen gained compared to aluminium.
3,0-4,0 defiantly some gains. Especially when getting close to 4A. About 200 lumen gained compared to aluminium star @4A. Or 15% increase.
4-6, huge gains

1500+ lumens @ 5A. NICE!

But, gains are mostly when having high vf.
Can someone explain to me how real life gains would be when using a 4-5 amp driver and only 1 li-ion like a Panasonic NCR18650B Protected in combination with a linear driver… Would there actually be any gains at all?

Using one protected 18650 in combination with a boost driver, would there be any issues?

If using only one battery, I am assuming larger , low resistance, no protection would be helpful. But pushing 5A for some time would not be realistic even with one “high ouput” 26650, or??

Exactly my plans. Got all my parts ordered. Just have to wait for these dang XM-l2’s to eventually come in.
And thanks match for the awesome tests!

Even the best 18650’s aren’t going to push the XM-L2 harder than about 3.6 to 3.8 Amps on a linear or switching buck driver.

And even if it does make that suggested current it would not stay there very long maybe a couple of minutes at best. A IMR cell would probably stand the best chance at it. Possibly 8 to 10 min judging by HKJ’s IMR graph.

Yep, the XM-L2 really wants two LiIon cells and a buck driver. That would be ideal.

PPtk

I had a feeling that would be the case…