LED test / review - XM-L3 U4 color kit E2 - basically more powerful 1st gen XM-L2, low Vf, high light flux, but not so nice light in optics

LED test / review


EN

Cree XLamp XM-L3 U4 color kit E2


2023-07-23

The XM-L2 has already been on the market for almost 10 years. It was first offered for sale in mid-2014 and then very quickly used in many flashlights. It offered impressive efficiency for the time, especially compared to its predecessor XM-L, and featured better electrical and optical properties thanks to a new LED chip.

At the end of 2019, a successor to the XM-L2 was announced. Some years earlier, Cree offered a new platform with significantly revised LED chips in the form of the XP-G3 and XP-L2, but continued to keep the XM-L2 unchanged in the product range, especially since it was still more frequently installed in flashlights and downlights.

Finally, in 2020, the XM-L3 was introduced to the market. The LED had changed characteristics; a lower forward voltage and high power binnings from the start were supposed to offer high efficiency, and the maximum forward current was also significantly increased, which was primarily supposed to result in the proclaimed “55% higher luminous flux” compared to the previous generation.

Here now is the test of this LED. As usual, the performance is tested and it is determined to what extent this LED can be used in flashlights with it’s typical optics.

Technical data


Tj 85 °C / 700 mA

Order code: XMLDWT-00-0000-0000U40E2

Type: single die, domed
Binning: U4 (typ. 340 lm)
Color group: color kit E2 (≈ 5700 K)
Rated voltage: typ. 2.75 V
Max. rated current: 5,000 mA
Max. peak current: —
Viewing angle: typ. 125°
Thermal resistance: typ. 2.2 °C/W
Max. Temperatur Tj: 150 °C

Official datasheet here (Cree, PDF)

First appearance


The XM-L3 differs significantly from its predecessor, the XM-L2. (It should be noted here that there is a new design of the XM-L2 since 2021, in which the LED has been completely revised, see my test of this updated XM-L2).

The biggest difference is the yellow coloring. While on the old XM-L2 the phosphor is only on the luminous surface itself, on the XM-L3 it is also distributed on the substrate around the LED chip. Cree’s logo has been applied to one corner, which also serves as the cathode marking here.

In contrast to XP-G3 / -L2, the LED chip is still connected with bonding wires. This indicates a classic lateral construction of the chip. Furthermore, it is noticeable that the bonding wires are thicker in the XM-L3, which should guarantee a higher current carrying capacity (this is also necessary with increased maximum current, since the XM-L2 was already destroyed at 5 A).


In its dimensions and general form factor, this LED corresponds to the XM-L2.

Due to the symmetrical design, centering rings made with lathe can be used without any problems. In general, accessories for XM-LEDs (called 5050) are fully compatible.


The footprint also corresponds to the XM format. The thermal pad is electrically isolated so that DTP boards of this format can be used without any problems.

LES and LED chip




As mentioned, the LED chip is connected with bonding wires. The shading of small areas of the LES by the bonding wires could affect the overall efficiency somewhat.

What is interesting here is that maybe a new LED chip is used. Small dots can be seen on the luminous surface, which are arranged more densely than on the XM-L2 (in real life more visible than on the picture). In addition, the luminous surface is somewhat smaller, which should have an influence on the luminance.

Some light is emitted from the sides. A faint glow is visible around the LED chip, like in the LEDs of the new platform. It is not clear if this is just from the phosphor applied to the side or if the LED chip also emits light from the side, as is the case with most newer LED chips.

The luminous surface is 7.75 mm² in size. It is also possible that other silicone with a different refraction index was used for the XM-L3 which causes the LES to be a little bit smaller as with the original XM-L2.

Power and overcurrent capabilities


Within official parameters:

  • at 5000 mA (official maximum current): 1952 lm @ 3.40 V
  • Power at official maximum: 17.00 W
  • Efficiency at 5000 mA: 114.8 lm/W

The official maximum current has been raised significantly compared to the XM-L2, which was at 3,000 mA. The specified power binning (typ. 340 lm @ 700 mA 85 °C) is met with this sample tested here.

The Vf is very low, especially for an LED classically connected via bonding wires. It can be seen that there are reserves for overcurrent at the specified maximum current.

In general, the XM-L3 has a very high efficiency, especially in the lower power range.

Overcurrent:

  • Maximum reached at 9.4 A, at this point 2798 lm @ 3.93 V
  • Power at maximum 36.94 W
  • Sweet spot at about 6.6 A (2349 lm @ 3.58 V)
  • Power at sweet spot 23.63 W
  • Efficiency at maximum 75.7 lm/W
  • Efficiency in the sweet spot 99.4 lm/W

Note: The overcurrent was stopped at 9.4 A to avoid destroying the bonding wires. By the increase of Vf it can be seen that the bonding wires are already heating up strongly and a failure at extremely high currents can therefore occur at any time!

The power increase is present. The Vf remains below 4 V. If FET-based (linear, or battery voltage dependent) drivers are used, a high luminous flux can be expected here, although this is always dependent on the quality (internal resistance, current and voltage behavior) of the cell used also the voltage and flux bin of the LED itself.


The LEDs shown here for comparison are very similar in their electrical properties. The luminous flux curves of the XM-L2 (old) and Luminus SST-40 are virtually identical, although the XM-L3 benefits directly from its better bonding. The flip chip XM-L2 (new version) benefits from its better heat dissipation as well as from the connection of the chip without bonding wires.

The extremely high Vf of the old XM-L2 is very unusual these days. In this respect, both the XM-L2 (new) and also the XM-L3 in classic design are a huge step forward, although the sample of the SST-40 tested here has an even lower Vf.

Luminance



The luminance is relatively high for an LED with a dome and a relatively large chip. The new version of the XM-L2 achieves a significantly lower luminance due to its optical properties, whereby the luminance gain is enormous with dedoming.

The luminance of the SST-40 is somewhat lower, whereby this is primarily due to the lower overall efficiency and this effect then no longer plays a major role with the same luminous flux as the XM-L3.

Light quality and color measurement



The LED tested here does not offer any special features. With a duv of 0.0040, the color locus is slightly above the BBL, whereby the green content is not as annoying and the light is absolutely usable in flashlights (at least if low color rendition and high CCT is not a problem).

With an Ra of 70 and an R9 of -33, this LED is a typical cool white LED in the range of typical 70 CRI.

It should be noted that this LED is only available with a CCT of 5000 to 6500 K and a CRI of typ. 70! Warm white light colors or higher color rendering are not offered for this LED!

Use in optics


Despite the yellow coating around the luminous surface, the light pattern is surprisingly good when reflectors are used.


Although a slight yellowish discoloration is visible around the spot, it is much less pronounced than was the case with XP-G3 or -L2. Even when using smoothen (SMO) reflectors, the light pattern is quite usable. In the spot, a minimally colder CCT is present, but this does not appear disturbing thanks to the likewise colder spill. In the case of textured reflectors (OP), the slightly yellowish area is even less visible.

When using zoom lenses, the light image is less nice, as the yellowish shimmer around the illuminated area is more visible. For such applications, however, an LED with a sharply defined luminous area, without a dome (domeless) and usually with a significantly higher luminance is recommended, such as the XP-P, Black Flat or “Yinding 5050”.

Conclusion


The XM-L3 is a surprise in that it is quite an interesting LED despite the complete redesign of the old XM-L2. The Vf is very low, the power is very high and the light image is usable even when using secondary optics. In addition, it offers high efficiency thanks to binning U4.

Basically, the XM-L3 is a performance-enhanced version of the original XM-L2. In construction, XM-L2 and -L3 are very similar, the main part of the better performance is primarily due to the thicker bonding wires.

I don’t like the yellow coating around the LED chip, which can cause slight color deviations in optics, and the lack of any CCTs and CRIs below 5000 K or above 70 CRI, which, however, may be due to the presence of a very similar LED (XM-L2 new version).

Pro

  • XM footprint
  • very high efficiency (U4 Bin)
  • very low Vf despite bonding wires
  • relatively high luminance despite dome

Neutral

    • Light image in secondary optics expandable, but still usable
    • Destruction of the emitter at extremely high current likely

Contra

  • No warm white and neutral white CCTs and high color rendering indices available
  • Poor color rendering

Thanks for reading the test! :slight_smile:

Many greetings, Dominik

10 Thanks

To me it seems like the new-gen XM-L2 with the flip chip design is superior in every aspect compared to the XM-L3

2 Thanks

Yes, and the availability of warmer CCTs with better CRI is also a reason not to choose the XM-L3. Why Cree released this upgraded XM-L2 1st gen, is beyond me. It makes no real sense, and even the price for these emitters is more or less identical.

I think it came out before the new gen XM-L2 didn’t it?

BTW, do you happen to have a big spreadsheet of all your tests?
Max amps, cd/mm², etc?

No, not to this date.

For me I am also not happy how these LED tests and therefore additional information are getting lost in this big sub-forum ‘modding’ (this is the reason why still no charts or other overview information exists), in german forum I have an own sub-forum for these tests where it’s possible to find all informations collected together in one sub-forum with great overview, but here…

2 Thanks

Thanks for the testing! I got a video up of the XML-3 vs the SST-40 in a C8 they are fairly similar. The XML-3 does produce higher lumens but not by a large amount and throw is almost exactly the same.

Awesome testing! Shame about the slightly disappointing results, but not every emitter can be the best.

Interesting how Cree chose to call the redesigned XML2 an XML2 instead of an XM-L4, seeing how the XM-L3 is inferior in almost every aspect. BTW, do you happen to know the phosphor area size of the dedomed XM-L2? Would be interesting to see how the beam compares to XP-L HI.

I tested the new design XM-L2 already, also with dedoming.

While fully dedomed, the LES of the new design XM-L2 is 3.9 mm² in size.

1 Thank

I’m happy to see you back in business koef3. :+1: