Results: Testing XM-L, MC-E, SST-50, and XP-G emitters **Updated**

**Update Jan27, 2012**

(Cleaned up the post a bit and added xp-g data)


Here are my results from testing the Cree XM-L emitter(s) up to 5 amps.

**Update Jan27, 2012**

(Cleaned up the post a bit and added xp-g data)


Here are my results from testing the Cree XM-L emitter( up to 5 amps. This is something I've wanted to do for awhile now, mostly because I haven't been able to find this info on the interwebs, and partly because of some debate and speculation going on in a few recent threads. This test took a heck of a lot longer to do than I first thought, so without further ado here are the results:

Test Set-Up:

* Emitter tested: Cree XM-L 1C mounted on a 20mm star

* Sample Size: 8 chosen randomly out of 80 (all I had the patience/time for)

* Test rig: Stars mounted on 1lb 25mm pure copper barstock set into a ~10lb block of pure aluminum. Arctic Silver 5 used on star.

* Temperature just under star monitored entire time. Maintained +/- 1 Degree throughout testing.

* Results taken 10sec after power on for any given level

* I know it's obvious, but lumen results are emitter lumens (bhp, not rwhp for you motorheads)

Here's a few pictures of the heatsink and test set-up:

And now for the big table and graph. The following represents the average results of the 8 different emitters for each data point (All emitters were within <1.1% of each other...another reason I stopped after :

Current (ma) Lumens
100 40.31
250 104.3
350 147
500 207.1
750 303.5
1000 392.1
1400 520.9
2000 686.2
2400 773.1
2800 852.2
3000 881.4
3200 905.9
3400 937.5
3600 956.5
3800 977.1
4000 988.9
4200 996.0
4400 996.8
4600 996.0
4800 988.1
5000 973.2

6-24-2011 Cree MC-E test added:

Mr. Oldbobk was kind enough to send me both an MC-E and an SST-50 to test. Test performed is exactly like the xm-l above, except only one emitter was tested. I will update again with sst-50 data once that test is complete.

Current in ma Lumens
200ma 73.5
400ma 142.3
600ma 207.1
800ma 267.2
1000ma 323.3
1200ma 376.2
1400ma 424.5
1600ma 470.4
1800ma 513
2000ma 553
2200ma 596.8
2400ma 630.8
2600ma 661.7
2800ma 693.3
3000ma 718.6
3200ma 746.2
3400ma 768.4
3600ma 788.1
3800ma 808.7
4000ma 826.1
4200ma 837.9
4400ma 849.8
4600ma 861.7
4800ma 868
5000ma 874.3
5200ma 869.6

6-30-2011 SST-50 Test added:

Again, same test set-up as for the xm-l. Here are my results up to 6.4 amps, which is as high as my bench power supply would go.

200ma 61.3L
400ma 122.8L
600ma 181.8L
800ma 239.5L


1200ma 348.6L
1400ma 400L
1600ma 449L
1800ma 497.2L
2000ma 543.1L
2200ma 587.4L
2400ma 630L
2600ma 670L
2800ma 708.3L
3000ma 746.2L
3200ma 790.5L
3400ma 825.3L
3600ma 857.7L
3800ma 889.3L
4000ma 921L
4200ma 948.6L
4400ma 975.5L
4600ma 1001.6L
4800ma 1026.9L
5000ma 1047.4L
5200ma 1056.4L
5400ma 1067.2L
5600ma 1087L
5800ma 1110.7L
6000ma 1126.5L
6200ma 1146.2L
6400ma 1160L

And there are the results. I didn't notice the drop around 5200ma until just now, but can attribute that to being distracted at the time and may have gone past my 10sec lumen snapshot. My data is rather more conservative than what is stated by the official datasheet, and that could be contributed to any number of variables. For example, even though my test heatsink is monstrous I feel that the 20mm star itself is the bottleneck for heat transfer. It is interesting to note that where the XM-L falls on its face at any current past 4.2amps, the good old SST-50 keeps chugging along. In certain high draw applications this would definitely be beneficial, not to mention the bond wires are much thicker and more numerous vs. the xm-l.

That's it for now...Considering all emitters were tested under the same conditions and with the same equipment, this should hopefully provide a good comparison between them. I hope some find this info useful.

Thanks for your time,

- Match

*Added Jan27, 2012*

Cree XP-G R5

200ma 80.9L
400ma 154.2L
600ma 218.2L
800ma 274.3L


1200ma 365.2L
1400ma 398.4L
1600ma 423.7L
1800ma 442.7L
2000ma 453L
2200ma 454.5L
2400ma 446.6L
2600ma 430.8L
2800ma 403.2L
3000ma 364.4L

For your viewing pleasure, the Cree XP-G. Test set-up was the same as the xm-l above (except only one emitter tested).

Special note for the morbidly curious: After the testing, I decided to sacrifice an xp-g by seeing at what particular amp level it would let out the magic smoke. I kept increasing current at 200ma increments past 3 amps (roughly 3 seconds between steps) until *poof*. The answer is 4.2 amps. I don't know about you, but I was rather surprised!



Update April 28, 2012

Many of you have requested that a test be run on the Cree XR-E R2. Thanks to Dthrckt who was kind enough to send me one, those results are now in. Again, testing procedure was performed like the others above.

Seeing as how the xr-e's are primarily used for throw, the above graph shows why such great sucess in this endevour is seen at greater than stock current draw. What my testing only indirectly shows is the increase in surface brightness that's critical for extreme throw.

Also note that this emitter was tested as shipped mounted on a 14mm(?) star. Life's been rather hectic as of late, but I plan in the near future to test this, along with the xm-l, directly mounted to copper. I believe it will be possible to get over 400lm out of the xr-e, along with a substantial increase in throw, in this configuration.

As always, thanks for reading.


Nice data, I appreciate the effort. Looks to me like all you need is a real 2.8 to 3.0A, after that the law of diminishing returns kicks in.

My big question is, does the draw that I measure at the tailcap actually equate to draw you get in these measurements?

Great work!

Okay Match, I for one am satisfied. My 980L is not over 1,000 lumens. (I'll be updating my review)

The only other question I had was the "overdriven" issue. Based on this it would seem 4.50 amps (which is what my 980L draws every single time I test it on high) is a waste. I would be perfectly happy with 3.5 amps and >900 lumens. Less heat and all that . . .

You are a real asset to this board Match and I applaud you.



Nice data, I appreciate the effort. Looks to me like all you need is a real 2.8 to 3.0A, after that the law of diminishing returns kicks in.


That's my feeling as well.


My big question is, does the draw that I measure at the tailcap actually equate to draw you get in these measurements?


Not unless your cell is directly connected to the emitter (DD). My test bench power supply was connected directly to the emitter, so the amps it put out was what I read and what the emitter truely saw. Anytime the power has to go through a driver there will be a loss.

So can you put a number on that loss Match? I have heard 20% thrown around, so at the tailcap we would need to see a number of roughly 3.5 or 3.6A to know that there is really 3.0A getting to the emitter?

Very nice Match ..

The only thing I would like to see is vF results ...

Can you vary voltage ? To the emitter ? say starting at 3.2v and maybe to 3.6v or 3.7v ? [ That would be interesting ]

Thank you sir for taking the time, effort and expense for conducting this experiment for the benefit of fellow enthusiasts around the world. Great job, you are the man.

It would be neat to test the 980L in the I.S. I'm guessing it would do ~850 OTF lumens. The real strength of the 980L lies not in the tailcap amps, but in the very well designed reflector (and the fact that it's a damn good looking light). I'd be willing to bet that it would very hard to tell the difference between a 980L with 2.8a SB driver and a stock one without having one in each hand and flashing them back and forth. So yea, less heat and better battery life.

Thanks for the nice comments guys. I'm glad to see folks find this info useful.

I had a extra fluke, but neglected to record vf as I didn't feel it was necessary at the time. The test bench power supply I used is either in CC or CV, depending on the setting, so no way of varying that. Previously (about a month ago) when I was playing around I set the current at 6a just so it wouldn't be a limiting factor, then proceded to vary the voltage starting at 2v and going up. Unfortunately I didn't write it down, was just messing around, but I do rememeber current rising quite rapidly due to small variances in voltage.

Another thing I just thought of is, it would seem the 3-mode UF XM-L is driven a little high too. Most of the time I get 3.5 to 4 amps in an L2P. Tonight, I just tested one of mine in a Solarforce L2. (Mr. Silver)

with protected Solarforce 2400:

high - 4.30 amps

medium - 1.18

low - .19

with protected AW IC 2900:

high - 4.36

medium - 1.17

low - .19

Here it is with the first black L2P I bought. Same exact drop-in, same exact batteries with the addition of the Trustflame.

with protected Solarforce 2400:

high - 4.20

medium - 1.14

low - .18

with protected Trust Fire flame 2400:

high - 4.00

medium - 1.17

low - .19

with protected AW IC 2900:

high - 4.10

medium - 1.15

low - .19

I'm just a little bored tonight so, sorry if this is irrelevant.


p.s. Match, if this is a thread hi-jack I will quietly delete.

I’ll have to bust out my AW 2900 vs Trustfire Flames graph when I get a chance. They’re pretty evenly matched. Actually, the AW 2600 is better at holding its voltage (resulting in higher amps) than either.

Now...the question is why the difference in amps between the L2P and L2?

EDIT: Yup... I just moved into a new place in Hijacksville. I'm such a hypocrite.

I was just wondering that myself. Gimee 10 minutes.


Sort of like putting an 850 holley double pumper on your honda civic ..

Foy if your flashlight is drawing that many amps instead of thinking of changing drivers to bring your amperage down why not just add two more xml's .??

I just updated my earlier post with the additional readings.

Well Boaz, in an attempt to at least somewhat stay on topic I guess my point is that all these XM-L drop-ins may be needlessly driven too high. Seems like I'm getting 10 or 20 more lumens at the expense of more heat and less run time.

2 more XM-Ls? Yeah, that's what I need; two more XM-Ls.


Wow Match, thanks for your effort these results are really interesting!

Great job! I really appreciate your work!

Really a great work Match. Would be really useful for everyone that approach a DIY wit an XML.

[quote= Flashlight Foy]

p.s. Match, if this is a thread hi-jack I will quietly delete.


No, not off-topic at all. In fact, this brings up a good point. After testing I took out my MF UF 3 mode drop-in and measured the current it was drawing off the battery ( measured without host. Used ampmeter leads to make negative contact). I measured @4.2a on high with a fresh battery. Then I hooked the same battery directly to an xm-l (DD) and took another reading. Guess what it read? @4.2a!

From what I've observed, with these new "xm-l drivers" is that they are DD in high mode and there is no current regulation happening at all.

I feel any current difference at high end is largely due to the individual cell. With that in mind, I would highly advice against running a low internal resistance cell like an IMR due to the high potential for excessive current. (If anyone has an 18650 IMR and is willing to test this, I'd be very interested to hear the results).

Great work Match , the results are what is expected given the manufacturer's data.

Trooplewis, the loss near or beyond 150 Celsius which is the temperature surely reached for those 'monsters' is around 30% therefore at this numbers a big chunk must be corrected , so I doubt you can get more than 700 L at 5A.

Also, roughly the tailcap current is the emitter current , what the driver burnt is the excess voltage Vin -Vf...

Foy; you are not getting 20 or so more Lumen . Temp corrected probably are getting as say above 700L ...that is what I mean in the other thread , dim , weak, less bright ...

The trick here is find the most you can push an emitter in a given body to get the most. Don't even dream you can put 4A successfully in a flashlight , not enough mass and a lot of thermal resistance.