LED test / review - Cree XLamp XP-L2 V5 40E & U6 40H 90 CRI – High light flux and low Vf, but very huge spread in performance

LED test / review

EN



Cree XLamp XP-L2

XPLBWT-00-0000-000BV540E (4000 K 70 CRI)
XPLBWT-00-0000-000UU640H (4000 K 90 CRI)

In September 2016, Cree released the next generation of XP-L, the XP-L2. The manufacturer claims that lumen density, voltage characteristics and reliability has to be improved.

August 2017 I tested this LED already, you can find this outdated test in the german TLF.

This test is different from my first test in the TLF. My measurements are more accurate than then and also I added luminance values for domed and dedomed XP-L2. Additionally I tested a XP-L2 U6 4000 K (min. 90 CRI) too, with really surprising results. From now on the test in the TLF is not valid and supported anymore!

Technical data

Tj 85 °C / If 1050 mA

Order codes:

XPLBWT-00-0000-000BV540E (4000 K 70 CRI)
XPLBWT-00-0000-000UU640H (4000 K 90 CRI)

Type: single die (Flip Chip)
Binning: V5 460 lm (505 lm @ 25 °C Tj), U6 380 lm (417 lm @ 25 °C)
Rated voltage: typ. 2.82 V (max. 3.15 V)
Max. forward current: 3,000 mA
Viewing angle: typ. 125°
Thermal resistance: typ. 2.2 K/W
Junction temp.: max. 150 °C

I purchased both emitter at Mouser a few months ago.

You can find the official datasheet here (Cree website, pdf)

First appearance

Like earlier Cree XP LEDs the XP-L2 is 3.45 x 3.45 mm (0.136 x 0.136 in) in size. In top view it can be easily discerned that the whole visible die including the outer surfaces is covered with yellow phosphor.

In off state the die is not visible well, unless the LED lights up with a few milliamps.

Like all Cree emitters of the newest generation the LED chip was mounted in Flip Chip design. Unless the classic Lateral design there are no visible bonding wires which could shield the LES because the contacts lying underneath the LED chip.

The dome is the same already known from the predecessor XP-L and also the general handling (reflow) hasn't changed from it's predecessor XP-L.

The only way to dedome this LED successfully is to shave it. Grab a razor blade, put a piece of center ring around the LED on the MCPCB and heat it slowly up. Then cut the dome on height with the center ring. Dedoming with chemical (gas or acetone) stuff ensures that the yellow phosphor peels of the LED chip so a bad light pattern and color is the result.

The LED was built in Flip Chip design. So the LES area is not sharply demarcated, which makes the determination of the die size much more difficult than other LEDs. On the picture above dedomed XP-L2 is right.

In original (domed) state the die is 10.41 mm² (0.0161 sq in) in size, in dedomed state 5.581 mm² (0.00865 sq in).

Overall, in dedomed state the LES of the XP-L2 is much bigger than of the XP-L, at approx. +40 %. The reason is that the LED chip which was designed in Flip Chip technique emits light to the side, not only to head-on. This is also the reason why the newest generation of Cree emitters produces discolored beam patterns both with reflectors and lenses.

How it looks if the phosphor layer was fully removed:

Here is also a comparison of XP-G2/G3 in side view to illustrate this design issue in a more practical manner.

Power and overcurrent capabilites

25 °C Tsp, unless noted otherwise.

Noticeable is the massive difference between both XP-L2 types I tested – mind you, the setup was not changed during this test. As part of this test, I additionally tested at least three XP-L2 V5 40E out of two different deliveries from Mouser. All three V5 delivers the same amount of light flux and shows the same performance behavior at very high current.

For the 90 CRI variant bin U6. Aaaaand… this is true. The bin was easily reached and – not only that. The Vf is MUCH lower than of all other V5 XP-L2, and the efficiency at the maximum current is almost the same!

This means also that all my tested XP-L2 V5 are NOT V5, but much less binning (V2/V3). I know this phenomena from the XHP70.2. For sure I double checked the results of my test setup with other LEDs of the same type (but also other manufacturers) to check the accuracy of my results. This is the reason why I also tested the U6 40H with 4000K CCT.

My opinion is that Cree has big problems to ensure the constant production yield and performance values. I have not seen such big differences between two LEDs of the same series before. Overall, the new LED chip technology seems not so finished and unproblematic as you would expect although this technology is on sale for at least a year now.

Within official parameters:

  • at 3,000 mA:
    • V5: 1,184.6 lm @ 3.34 V
    • U6: 1,057.9 lm @ 3.17 V
  • power at rated maximum:
    • V5: 10.02 W
    • U6: 9.51 W
  • efficiency at 3,000 mA:
    • V5: 118.2 lm/W
    • U6: 111.2 lm/W

Overcurrent:

  • maximum current:
    • V5: 12,000 mA - 2,606 lm @ 4.39 V
    • U6: 13,000 mA - 2,438 lm @ 3.91 V
  • power at maximum:
    • V5: 52.7 W
    • U6: 50.8 W
  • Sweet Spot at 7,500 mA:
    • V5: 2,147 lm @ 3.99 V
    • U6: 1,991.7 lm @ 3.57 V
  • power at sweet spot:
    • V5: 29.9 W
    • U6: 26.8 W
  • efficiency at maximum:
    • V5: 49.4 lm/W
    • U6: 48.0 lm/W
  • efficiency in sweet spot:
    • V5: 71.8 lm/W
    • U6: 74.3 lm/W

I define the sweet spot as the position in the diagram which gives a good average between light flux, current and efficiency. In most cases it lays roughly 15 to 35 percent lower than maximum possible light flux.

Interesting facts

  • The differences between the two LED types are very huge. The efficiency of the U6 is almost the same as V5, because of the much lower Vf.
  • The 'V5' is not V5, but V2/V3.
  • The XP-L2 U6 reaches almost the same maximum light flux as the V5 one.
  • At max. current the efficiency is very low, so I cannot recommend the using of maximum current.
  • The LED (V5) died at 18.4 Amperes (!) and 5.49 V, at a maximum power of 103.2 Watts (!!).
  • The Luxeon MZ 5700 K in 90 CRI Variant (3 V) has almost the same efficiency but delivers a lot more light at higher current - at a much lower Vf!

Luminance

I established a new method for determining luminance, especially to ensure more realistic values for 'real-life' conditions (flashlight use). The measurements are taken with a new original Convoy C8 reflector, but with same heatsink / setup as previously used in flux measurement.

Reflectivity for reflector 85 %

Transmission UCL glass 97,8 %

Values at 25 °C Tsp, for 85 °C Tsp values are 6 to 14 percent lower, depending on LED

LEDs marked with Warning sign uses old values are still taken with previous method (determining die size) due to problematic light distribution (donut holes etc).

In general the XP-L2 is not really suitable for thrower. Even in dedomed state it can't keep up with the predecessor XP-L HI. This also applies to the XP-G3 (dedomed). The new design of the next gen Cree emitters is not really suitable for reflectors and big aspheric lenses.

I have not updated the values for XP-L2 U6 yet, because I ran out of samples to test.

Light quality and use in optics


The newest generation Cree LEDs (XP-L2/G3, XHPxx.2) are similar to each other. They all create a yellow-tinted corona around the spot when it's used in conjunction with reflectors and clear lenses.


With this LED it’s the same. To show you this problem in a more practical manner I put the XP-L2 with same CCT as the LED tested here in a JAX Mini C8 with OP reflector and clear UCL glass. The textured reflector should improve the light pattern.

Like XP-G3, XHP50.2 and XHP70.2 the yellow-colored corona around the spot is visible and annoying. The spot is definitely cooler than the stated 4000 K CCT. This CCT could only reached in situations where the light is mixed like using of diffusors / matte lenses or ceiling bounce.

The XP-L2 90 CRI (2-step MacAdam ellipse binning) variant has a yellowish tint (sometimes also greenish), like the most High CRI Cree LEDs. In the most cases the tint stays above the BBL which guarantees a yellow color...

Conclusion

Like other LEDs built with the new LED chip technology primarily the XP-L2 is a significant leap in over current and power capabilities. I do not like the big differences and spread of the tested XP-L2 in different tints. In general, sorting of XP-L2 LEDs (‚pre-binning‘) is absolutely necessary if highest performance and highest efficiency is desired.

Nevertheless, the XP-L2 delivers an huge amount of light flux in XP footprint. Only the XHP35 delivers a higher light flux at the same package size, but needs 12 Volts instead of 3 V. Because of the XP footprint used the handling and reflow is very easy and you can rely on a large selection of MCPCBs.

Pro

  • Very high light flux possible
  • Very low Vf (U6 90 CRI variant only)
  • Fits on industry standard XP footprint
  • Easy handling

Con

  • Huge spread of voltage and light flux maximum with different XP-L2
  • Imperfect color consistency if used in clear optics
  • Bigger LES than it's predecessor
  • Rated CCT can only be achieved with mixing the light
  • V5 emitter didn’t reach the specified binning
  • not suitable for ultra-long-range flashlights

Thanks a lot for reading! :)

Greetings, Dominik (aka BLF member koef3)

Mistakes and suggestions are best sent via pm.

Terrific testing Dominik. The info here is invaluable to modders. Thanks. :beer:

Thanks for posting.

Thanks for the nice test. Those are weird differences!!

I can add that the 4000K 90 CRI XP-L2 leds that I have from Mouser are way too green to my taste, while the 4000K 80 CRI XP-L2, also from Mouser, has a wonderful yellow/rosy tint and even the corona is less annoying.

Great tests, koef. I’m glad you made your way over to BLF!

Which order code?
The yellow corona is caused by the yellow phosphor which covers the whole visible surface. I don’t know how the use of another XP-L2 makes the artifacts more bearable… Only because of the rosy tint?

@gchart: Thanks. :slight_smile:

Thanks for the info :+1: :beer:

Thanks for this valuable information, Dominik.

Same results I have noticed, I really like all the 80CRI second gen CREE LED’s I have tried, (xp-L2, xhp70.2, XP-G3) but the 90 CRI versions all seemed to have a green hue.

so 2000 lumen 90+ CRI single die led, interesting
if you go with triple setup, that means 6000lumen 90+ CRI and 22Amp
i wonder that what’s the problem with such low vf and DD driver, i’m scare it will go over “sweet spot” and broken

In most cases the FET on the driver can’t handle really high currents (10 Amps or more). They got very, very hot in operation (over 120 deg C), even in highest modes. With a standard A17DD_L I won’t go over 7 Amps current on LED.
Second the lifespan of the emitter will decrease, and the more current the more heat has to be dissipated.

You will have losses in light flux (10 to 30 percent) in every light, because of TIR optics, lens, reflectors, no perfect focus and so on.

Thanks for the nice test! :wink: The 90CRI Variant of this LED is also special because of the die size (other 90CRI LEDs with a single die are smaller).

Why is the dize sie different from your last test? You measured 3.55mm^2 for the de-domed variant in that test.

I can reccommend using a Lee quarter Minus green filter (or even better a Lee Zircon) to reduce the green part of the spectrum. It makes it much more bearable.

In my last test I measured the semiconductor (die) size only. But this is wrong, with applied phosphor layer the LES is slightly bigger, as you can see. :wink:

The LED chip itself is always smaller than the LES in the complete LED (with phosphor and silicone). As example, see a EZ1950-p Chip is significantly smaller than the LES of a shaved / dedomed XM-L2 (where this chip is used in).

I updated the LES size measurements and therefore the luminance values.
This affects all newest generation Cree LEDs.

You also added a very nice pic showing the emitters side!

TLF so sorry for your loss… Well I am not :smiling_imp:

Keof3 is the man! Beautiful review :+1:

Important notice:

I updated the luminance values for all LEDs shown in test chart including XP-L2 V5 40E. I could not update the luminance values for U6 40H because I ran out of samples.

To ensure more realistic values for 'real-life' conditions (flashlight use / secondary optics) I established a new testing method to determine the luminance with a SMO reflector. All upcoming and already existing tests will use this method!

Updating of my other tests will be done in the next hours / days.

Hi, wonderful review, does anyone know where to get the 90CRI variant on 16mm copper plate?

Is the xpl2 v5 still the highest bin? And has anyone used them in a triple setup what’s the current and output with a fet driver

I’m pretty sure V6 has been available for years.