LED test / review - Lumileds HL2X (1800 / 5700 K, typ. 80 / 90 CRI) - High power candle light LED!

LED Test / review EN


Lumileds HL2X (1800 80 CRI/ 5700 K 90 CRI)


2023-08-20



The Lumileds HL2X was launched in mid-2020. With a standard footprint of 3.45 x 3.45 mm and a special focus on a good light image in secondary optics and light quality, it is preparing to compete with established emitters. But how does it really perform, and which factors have to be compromised?

In this test, a 5700 K 90 CRI and an 1800 K 80 CRI, which was only released at the beginning of this year, are tested. The latter LED in particular could be ideal for dual-channel tintramping lamps when it comes to the widest possible range of adjustable CCT.

These emitters were regularly purchased from Mouser in early 2023 (5700 K) and July 2023 (1800 K).


Technical data


Tj 85 °C, If 700 mA
  • Type: single-die
  • Bin: J (180-190 lm, 1800 K) / Q (240-250 lm, 5700 K)
  • Color group: 5-step MacAdam ellipse, 1800 / 5700 K
  • CRI: typ. 80 (1800 K), 90 (5700 K)
  • Rated voltage: 2.83 V (min. 2.65 / max. 3.00 V) V
  • Max. Forward current: 2,000 mA
  • Max. Peak current: 2,500 mA
  • Viewing angle: 120 °
  • Thermal resistance: 2.4 K/W
  • Max. Temperature Tj: max. 135 °C

Datasheet can be downloaded here (Lumileds, 3.4 MB)


First appearance


At first glance, the HL2X could be mistaken for a Nichia 219C. The luminous surface is surrounded by white silicone, which in turn has been applied to a gray substrate. Shimmering golden material is visible at the edges.

The difference in size of the luminous surface between the two tested CCT types is interesting. A larger luminous area results in lower luminance and thus lower ranges when optics are used, as long as the larger luminous area is not compensated by correspondingly higher luminous flux.

Due to the transparent silicone dome, the luminous areas appear larger than they actually are.

Handling of these LEDs is no problem, the mechanical stability is very good and should also withstand a somewhat rougher treatment.



Unlike the previous Luxeon V from 2017, which still relied on a special format (4 x 4 mm) and made reflowing on XP boards very difficult, Lumileds has opted for the classic XP format (called 3535, dimensions 3.45 x 3.45 mm) for the HL2X. This allows free swapping of existing emitters and at the same time offers the advantage of maximum compatibility of the accessories available for XP footprint.



The footprint offers no surprises. The notch in the thermal pad marks the cathode. To enable lower contact resistances and avoid corrosion, the contacts have been vaporized with gold.


LES


The luminous surface is reminiscent of older Nichia emitters. Fine granulation is visible, but this should have no effect on the light pattern. Disturbing lateral radiation or color fringes are not visible.



The luminous area is 6.1 (1800 K) / 4.90 (5700 K) mm² in size. <The enormous difference in size between the two luminous surfaces is striking; it amounts to more than 20% here.

Lumileds refers to the technology of the LED chips as ‘CSP’ (Chip Scale Package), as was already the case with the Luxeon V, but this does not fit purely in terms of definition. JEDEC has defined this in the J-STD-012 standard. This standard states that the package must have an area that is not larger than 1.2 times the LED chip (of the die).

Presumably, this designation therefore refers to the optimization of heat dissipation or the mounting of the LED chip on the substrate, which here corresponds to the known flip-chip design.


Power and overcurrent capabilities



Within official parameters, as far as known:

  • at 2,000 mA (official maximum current): 503 (1800 K) / 672 (5700 K) lm @ 3.20 / 3.20 V
  • Power at official maximum: 6.4 / 6.4 W
  • Efficiency at 2,000 mA: 78.5 / 104.9 lm/W

The clearly lower performance of the 1800 K HL2X is clearly visible. This is mainly due to the lower base efficiency. The Vf is more or less identical for both types, although that of the HL2X is a few mV higher.


Overcurrent:

  • Maximum reached at 4.8 (1800 K) / 5.6 (5700 K) A, at this point 804 / 1244 lm @ 3.64 / 3.69 V
  • Power at maximum 17.5 / 20.7 W
  • Sweet spot at about 3.0 / 3.6 A (666 / 1031 lm @ 3.38 / 3.45 V)
  • Power at sweet spot 10.1 / 12.4 W
  • Efficiency at maximum 46.0 / 60.1 lm/W
  • Efficiency in the sweet spot 65.7 / 83.1 lm/W

Interestingly, the maximum possible operating currents of both LEDs differ somewhat. Possibly this is due to slightly different thermal resistance or the thicker phosphor layer, which reflects some light back to the LED chip.



To cut a long story short: the thermal resistance of the HL2X is simply too high. Both the Nichia 519A and the already older XP-G3 are clearly better in terms of overcurrenting. The 519A tested here reaches the maximum luminous flux of the HL2X already at 5.2 A, and can even exceed it by about 20% - and that with an emitter with higher R9 and warmer CCT. The XP-G3 in the diagram is out of competition due to its poor color rendering of only typ. 70 CRI, although it impressively shows that it is still a good emitter at least in electrical and thermal properties.

For the 1800K variant it looks even worse in terms of thermal resistance, apart from the extremely reduced efficiency. The lower maximum current shows that there are differences in thermal resistance or design of the LED. But to be fair, the 1800K HL2X is one of the very few LEDs on the market that can achieve a luminous flux of 800 lm at this extremely warm CCT.


Luminance


Values at 25 °C Tsp, at 85 °C Tsp values are 13 % lower


The luminance of the 5700K 90 CRI HL2X corresponds to the normal midfield for an XP LED with dome. The 519A has a larger luminous area and accordingly has a lower luminance. The 1800K HL2X has an even lower luminance, but this is due to the larger luminous area and especially the considerably lower maximum luminous flux.

Light quality and use in optics



The HL2X differs in the light pattern depending on the CCT. While the cool white HL2X has a faintly visible color shift with lateral radiation depending on the reflector, the 1800 K variant does not have this problem. The light image of the 1800 K variant is almost perfect and corresponds to that of the Nichia 519A. This allows the trouble-free use in flashlights and secondary optics.

The cool white variant has a faint green-yellow corona around the spot, which is more or less noticeable depending on the optics. However, this is not annoying at any time. At least in the Quad-TIR optics of the Convoy S21F (shown in the picture), this problem no longer occurs, presumably because of the texturing of the lens, which allows optimal mixing of the light from different exit angles.


Color and light quality


The cool white HL2X has a slight greenish tint, but this is not really annoying at any time. Outdoors, this is only really noticeable in direct comparison with an equivalent light source with a negative duv, which can also be seen in the pictures. Nevertheless, the color saturation is lower than that of the 5700 K 519A, and the difference is especially visible in a direct comparison of both LEDs.

The deviation of the CRI/Ra from the values stated by the manufacturer is not so good. A CRI of 85 for an LED with typ. 90 is too low, furthermore the R9 is also lower than would be expected for an LED with high color rendering. Presumably, this deviation is due to the high green content in the spectrum.

The 1800K variant, on the other hand, delivers a better color rendering index than specified. In general, this light appears subjectively perfect, the duv close to 0 and the extreme similarity to a tungsten-type light source / black body spectrum underlines this. The special thing about this LED is that it still has blue and green components despite the extremely warm CCT. This not only increases the CRI, but also makes the color perception appear much more natural, or enables a light that is otherwise only known from thermal spotlights.




  • Ra: 85 / 87
  • R9: 27 / 32
  • CCT: 1925 / 5784 K
  • duv: 0.0013 / 0.0118

Conclusion


Overall the HL2X are absolutely usable LEDs, which can be used freely thanks to the XP footprint.

However, the star here is clearly the 1800 K variant. With its good light pattern, the spectrum with enough blue content for a higher CRI and the relatively high performance (seen for the CCT), this offers a real unique selling point. For applications where this extremely warm light color is essential, this emitter is absolutely recommendable.

Depending on the reflector, the cool white HL2X can have a problem with color fringes around the spot, although this problem no longer occurs with textured optics. The disadvantage here is especially the color localization, which can be in the greenish range, but this is not as extreme as with other LEDs tested so far and is hardly disturbing, at least without direct comparison.

For me, the HL2X is definitely an LED that can be used well in flashlights. I therefore expect that this emitter (depending on the price development in wholesale) should be found in the future in one or the other series lamp.


Pro

  • 1800 K variant with good spectrum, which has blue component
  • 1800 K variant with good light pattern
  • very good tint with 1800 K variant
  • high performance for LED with ultra warm white CCT
  • XP footprint for maximum compatibility

Neutral

  • 5700 K variant with greenish tint
  • slight color fringing possible depending on optics

Contra

  • no narrow color binnings available (mainly relevant for neutral and cool white variants)
  • CRI of 90 not achieved (5700 K variant)

Thank you for reading the test. :slight_smile:

Greetings, Dominik


v1.0.1
8 Thanks

Thanks for all the info! Beam shots of the 1800k?

Maybe later, but tbh the beam is like 2700 K but a lot warmer. For outdoor I don’t recommend this 1800 K LED since everything is basically red and yellow, with only slight amounts of green and blue.

Hmm. I might give it a try! Since 1800k emitters are scarce.
I wonder what would happen if dedomed?!
1400k ??

Edit.
Mouser quoted me 35$ for shipping for a 1.5$ led… never mind lol!

Try to go over 50 euro/dollar per order (at least in the EU this saves you from these extreme high shipping fees)

The HL2X cannot be dedomed as easily since the phosphor is more or less ‘glued’ to the silicone dome. In case of removing the dome the phosphor is torn off the LES.
Slicing does not change the CCT.

For the XP-E2 in color kit ZA (2200 K) which I dedomed several years ago it goes down to around 1700 to 1800 K, so your assumption of 1400 K could be realistic. I think the spectrum looks more or less identical to the PC Amber type LEDs out there, with almost no blue in it, which makes the color rendition horrible…

3 Thanks

Thanks for the test.

I wouldn’t call 0.0118 slightly greenish, that’s not even ANSI white which is 0.0090 max at 5700K, 5-step being ~0.0085 max, so according to the measurements it’s outside the specs.

How do you take the readings ? Spot from a reflector ? Or is it integrated ?

1 Thank

Was looking for a more budget friendly solution. But thanks for the tip! I’ll keep it mind!

To bad they can’t be dedomed easily. I became spoiled by the easiness of the 519a of dedoming

Very promising results for the 1800k version, I wonder how it compares to the 1800k versions of XP-G3 and 219F, both are only Ra70 so they might perform better at the cost of tint. Might make for an interesting swap into a D4v2.

Looks like the high-cct versions really are nothing special though.

The XP-G3 did have the problem with extreme color fringing when using reflectors and also TIR lenses.

It is likely that the XP-G3 at 1800K will be less noticeable when used in optics. This XP-G3 at 70 CRI typ. should be about 15-20% more efficient, plus the advantage of better overcurrent capability comes into play (though the ultra-warm white types generally have lower overcurrent capability in my experience, but it should be still better than the HL2X tested here).

The 219F is hardly available in 1800 K, at least in the EU, because there are still no wholesalers in the EU where you can easily buy Nichia LEDs…

1 Thank

Always integrated.

Do you apply correction factors on the SPD measured in your sphere ?
Maukka used an halogen bulb (uniform source), measured outside and inside the sphere, even his professional sphere needs correction for tint measurements, because while barium sulfate reflectance is very good across visible wavelengths, it’s not 100% uniform.
(Dunno if your sphere has barium sulfate paint or not)

Edit : -53K, +0.0021 without correction. CRI is also affected.

IMO the most interesting aspect of them isn’t brightness, it would be tint, not much use for us to have a bright candlelight that is piss yellow in color.

I guess that would necessitate trying them out, since usually ultra-warm emitters end up pretty positive.

Maybe eurekatronix has some 1800k 219F, though shipping takes quite a while. A test of it, even if only tint and lumens would be extremely valuable IMO.

No, I used titanium dioxide based paint, since barium sulfate based paint is not really available for consumers.

Thank you for the suggestions to make the measurements even better. I am currently looking for a certified lab to get some calibrated light sources with official paperwork according to the international standards, so I am able to improve the accuracy of my measurements (light flux in general, and spectral data).

Maybe, but even the webpage eurekatronix.com is not reachable (" Looks like this domain isn’t connected to a website yet").

And since I got still no info for a group-buy of Nichias from Clemence (via a user from german flashlight forum) I don’t want to order something there since I cannot be sure to get these LEDs in a reasonable amount of time (max. 4-6 weeks). Tbf I don’t know if this is because of the user in german TLF but reasonable delivery times are crucial since I want to use the LEDs at some time.

IIRC the website is down but the email is still live, pretty weird but it’s one of the only Nichia dealers that sells emitters to community members. Store wasn’t working for a while so you had to buy through email anyways.

So like BaSO4, TiO2 reflectivity decreases at shorter wavelength, but it’s more pronounced.
You can easily verify it with a random incandescent light since it should deviate from the black body reference.

1 Thank

Is there a price list available (maybe as PDF)?
Is tracked shipping (for the much higher price, respectively) available? For getting the ordered parts in reasonable time the price is not the biggest issue for me.

Tbh, I am not really happy to order some parts via email and hope for the best…

If you know someone who sells ready-to-use barium sulfate based paint to consumers, I would appreciate it.
I can only work with the things I have access to.

I suspect that the accuracy and repeatability of the luxmeter plays more of a role here than the coating of the sphere. In the worst case, we’re talking about an 8-9 percent difference here (which means not the same amount of difference in light flux), which certainly has an influence, but I still don’t consider it to be that significant.

What difference should be expected there? Without a comparison value, I can hardly evaluate any differences…

Interesting topic, but I cannot open this thread quoted by you.

I don’t know, when I wanted to make my sphere I wanted to use readily available barium sulfate powder and an acrylic binder, I never finished it though…

Anyway i’m not saying you should use barium sulfate, but just apply a correction for the paint (as it would be needed as well with BaSO4)

The reference is the incandescent bulb measured outside the sphere, since it emits the same light it every direction, it doesn’t need to be integrated, then you compare the direct vs integrated measurement, calculate the correction factors for each wavelength and apply it when you do color measurements in the sphere.
CT&A doesn’t have a correction function, you’ll have to do it in a spreadsheet or with a script. ArgyllCMS spotread has a correction function.

It’s a private discussion, there isn’t more relevant info to quote.

Here’s the price list I was sent, it might not be up to date though but it seems approximately $1 per emitter for an order of 219F-V1 in an unusual CCT would be likely.

The default economy shipping ($15 Indonesia to Peru) was tracked, there’s probably also an option for faster (DHL, Fedex, UPS) shipping than regular mail if you request it.

About 11USD to Germany, tracked.