Cree XP-L2 CRI70 + CRI90 color testing

Flashlight Modding and DIY Parts
1

Peter from luciferlights.net sent me some XP-L2s for testing. They are used in the new Zebralight SC64c for example. I haven’t gotten the SC64c yet (not available in Europe), so this should be interesting.

I’ve got here
3x 4000K CRI70 (XPLBWT-00-0000-000BV50E5)
3x 4000K CRI90 (XPLBWT-00-0000-000UU60E5)

All six are mounted on Peter’s own DTP boards he uses in his headlamps.

I tested the tint on the small reflector from the Reylight Pineapple. I’ve noticed that it manages to hide tint shift pretty well, maybe a heavily diffused TIR could be better, but after the S1 Mini testing, I’m not so sure.

I still haven’t fabricated a mount for the sphere, so no lumen testing. I did graph candelas against drive current though.

Here are the six samples with the reflector measured from the hotspot (1000mA)

The CRI90 ones are clearly greener and the variation is larger. So far looks very good for the CRI70 emitters!

Here’s the tint shift with varying current from 200 to 3000mA.

Clearly different behavior between the CRI bins. Both turn cooler and less green on higher currents, but the trajectory is different.

Tint shift within the beam (1000mA, hotspot via corona to spill)

This is probably the most interesting part of these emitters. The Cree rainbow is there, but it’s not as bad as the XP-G3. The CRI70 version does shift a bit more and is more visible. The strong green cast on the CRI90 spoils it though. This can be fixed with a minus green filter, but the shift is still there.

An ideal emitter would of course have the same tint all throughout the beam, but that’s not really realistic. Nichias do very well though and their shift is linear to only one direction without much tint shift towards green or magenta.

CRI data (1000mA)

Best samples embedded here, click links for the others. Nothing surprising here.

The rest:
CRI70 #1
CRI70 #2
CRI90 #2
CRI90 #3

Forward voltage and throw

The CRI70 is a bit more efficient but not too much (8-16% depending on current). It has a higher Vf but also more output at the same current.

2

A big thanks!
I was hoping for this :).

The 90CRI XP-L2 is special because it has the 3.5mm^2 die and is very efficient. Especially at higher power levels it is much brighter compared to the Nichia 219C R9050. It is a great general purpose emitter if one can live with the tint shift problems.

3

Thanks for this, maukka! I had been wondering if the tint shift on the high CRI variants were as bad as on my CRI 70 emitters. It appears to be better, but probably not worth going out and buying/swapping for my own lights. For those that care about CRI, I’m sure it’s a win. I’m not sure if I’m glad or sad that I can’t see any difference on high CRI lights (due to color blindness). :wink: Tints and tint shift however… I’m a bit more picky.

4

Thanks for the test and detailed figures. I have a Zebralight h53c with the same emitter and love it. It’s far from perfect as there is tint shift and yellow hotspot Corona, but in real use colors are so much more vibrant than the previous h52w using xm-l2 75cri LED. I think this LED is a good compromise for brightness and tint vs a Nichia 219c which is clinically much more consistent but not as bright.

I’m hoping MTN electronics starts selling the 4000k 90cri xp-l2 so I can put it in my S41 quad! :slight_smile:

5

maukka, thank you for the tests! But about ZL you are not correct. ZL used 2step xpl2 and bin should be XPLBWT-00-0000-000UU640H. Its avaliable on mouzer.com . 2step should be closer to BBL.

6

Thanks for the clarification. Let’s hope it’s better. Still doesn’t remove the shift though.

7

Is is actual for reflector only. ZL use xpl2 with frosted glass. So there is reason to measure average characteristics.

It is obvious that Hicri not for reflector lights)

8

SC64c doesn’t have frosted glass.

9

is there any beamshot, i dont like High_green-CRI from cree(like the one in olight S1 mini)

10

Nope, 2-step 4000K ist mostly above the BBL (Nichia also has this problem). 2-step for 4000K LEDs means that they are consistently slightly yellow-greenish. :frowning:

3-Step and 5-Step can be even worse, but it’s also possible to get more reddish tints.

11

Nevertheless, I used 2-step xpl2 and after that I have decided not to buy 219C (greenish) and 219B-v1 (reddish) any more. It has most beautiful tint I saw over 3В LEDs) That is my subjective opinion.

12

Has anybody tried a Lucifer headlamp?

13

I have a bunch of 90 CRI 4000K XP-L2’s and I removed them from my D4 because of the green tint + tint shift combination (the Luxeon V’s that I replaced them with have both better tint and less rainbow, albeith low CRI).

What I really do like are the 4000K 80 CRI XP-L2 that I got from Mouser. None of the green-ness, great tint, and the tint-shift over the beam seems less too, a nicer led IMO, and 80CRI is pretty nice for a flashlight (in my home I aim for 90CRI at critical locations like the kitchen and dinner table).

14

Output measurements for the 4000K CRI90 XP-L2.

15

Nice that you got the output tests sorted now, so with you and koef3 posting them on BLF, I can safely go on emitter test retirement. :innocent: (I still like to do UV leds though).

To get the complete light cone of a led up to 90 degrees inside my integrating sphere I had to add a copper pillar on my led-mount, which for very high power leds got in the way of perfect cooling. How does your mount sit in the sphere I wonder?

16

The heatsink is 9mm and the sphere opening is 10mm. I hold it in place with three screws. Here’s an image of the mount. It’s been adjusted so that the surface of the mcpcb is level with the sphere wall.

It’s of course been calibrated with taking into account the light that can escape through the heatsink gaps.