SunLike 3500K 95+ CRI midpower led from Seoul Semiconductors tested

Flashlight Modding and DIY Parts
1

The SunLike leds are an ambitious project between Seoul Semiconductors and Toshiba, creating extreme high CRI leds without cyan-dip, based on deep purple leds. They fall in the same category as the Nichia Optisolis but were from what I recall announced much earlier. I forgot about them because even if they would hit the market I did not expect them easily available soon.

But to my surprise 2 weeks ago I came across them again on leds.de (Lumitronics). They have them on their house-brand led strips, and the shortest one, with 26 leds, was with 15 euro within my hobby budget. I ordered a strip with 3500K SunLike leds, perhaps not the most interesting CCT but for me the most attractive. I unsoldered a few on the heat block for testing on a good quality non-DTP board (my custom djozz-board) and here are the results.

The SunLike leds are industry-standard 3030 size and footprint leds (like also Optisolis) rated for 150 mA and around 25 lumen output, and minimal 95 CRI. The SunLike cimes in 6500K, 5700K, 5000K, 4000K, 3000K and 2700K. The datasheet can be found here
.

As you can see in the picture above, with the led lighted at 1 mA, just like the Optisolis, the led seems to have 2 dies, but unlike Optisolis, the dies are next to each other without a gap and they are centered. So optically at least there is some potential using a single led in a (low power) flashlight.

Letā€™s first see if they live up to their high CRI specifications. The led at 150mA from straight above the led:

And at about 60 degrees:

The spectrum is a bit more wobbly than the Optisolis spectra, but the CRI and R9, R10, R11 are more than promised, and the tint is at both angles close to the BBL. The base led seems around 420nm, just as the cooler Optisolis leds, the warmer Optisolis use a 450nm base led so Seoul took a different decision for the SunLike. Looking very good sofar :slight_smile:

The voltage/output test.
The test is done like all my led tests, essentially like described first in my XP-L test: Cree XP-L V6 2C led, tested against XM-L2 T6 3C, edit23/7: I repeated the test and measured a dedomed XP-L . One led is tested. The ledboard was a non-DTP but good performing 16mm copper board. I added my Nichia Optisolis and Yuji VTC D50 5730 tests in the chart for comparison, so all the ultrahigh CRI midpower leds tested sofar are in there.

The Sunlike performs clearly better than the Yuji led, but also does not compare bad to the praised Optisolis leds, with similar performance up to 400 mA (which is already well outside the specifications), with almost a half Volt higher voltage. Above 500 mA the Optisolis keeps performing better while this SunLike led tops off. The SunLike led died between 950 and 1000 mA.

Conclusion: the SunLike led seems a very well performing extreme high CRI led that has a colour performance very similar to the Nichia Optisolis but it does not match the Optisolis in toughness. But it comes close enough to be an interesting alternative, especially for the 5700K colour temperature that the Optisolis led does not come in. I have no idea about the price per led, but that also could make them more attractive than the Optisolis leds.

2 Thanks
2

I modded a little Tool Ti with this led. These flashlights on an alkaline or Eneloop usually run the led at around 300mA, which is a nice current for the SunLike. I reflowed the led on the stock XP board (non-DTP), on which the led sat itself a mm off-center. I grinded the side of the board to make the led still sit in the center of the flashlight head. I did not get it completely centered, but it looked like the beam was about as good as it could be.

Although the colour performance of this SunLike led is good, both from above and to the side, the behaviour in a reflector has more to it, it also depends on the spatial distribution of brightness and tint within the led. And there is lots of variation over the die, cooler tint and brighter in the center, and more orange and dimmer to the side. And this shows in the beam: as least as much tint shift as a XP-G3, but hey, the corona is not green or yellow but a pleasant 2700K peach colour :partying_face: , with a 4000K in the center of the hotspot, and 3300K in the spill. A quite fun beam actually, not boring at least, and hotspot, peach corona and spill all still measure over 97 CRI with 90+ R9.
Hereā€™s the beam of the SunLike (bottom right) compared to 80CRI 3500K Luxeon Q (bottom left, lots of tint shift too), and 90CRI 4000K 219C (top). The picture does not show the true tints but it gives an idea.

3

I guess there isnā€™t much chance of hurting it in a 1S Li-Ion powered flashlight, since the Vf gets out of range of the cell before the LED gets to be in real danger. But the light output is ridiculously low, even at the peak output. {EDIT: It looks to be around 65 lm/W at the peak.} What would be the use case for this LED in a flashlight?

4

Although I like how it turned out in the Tool mod (indeed about 50 lumen on high) the best way it can be used in a flashlight is the same way as the Optisolis: multi-led and lots of diffusing.

5

How do you get to the LED in a Tool? I tried it in the EDC01 which looks like itā€™s set up the same and ripped out the driver :person_facepalming:
PM me if you donā€™t want to take this thread off topic.

On topic thoughā€¦ i would use this LED just for the lack of cyan dip (though ideally iā€™d like a little more violet, but thatā€™s just me being fussy). I rarely use more than 50 lumens, and it might work better behind the pebbled TIR in the EDC01 :smiley:

6

The Tool Ti is a bit different, it has a brass pill that does not screw into the head, but on one side the battery tube is screwed on the pill, and on the other side the bezel is screwed on. Easy to mod (if your led refows centered that is :confounded: ).

The beam will certainly be cleaner from a pebbled TIR optic.

7

Thanks for the test Djozz, I saw this LED earlier. The symmetrical dual dies beam looks a tad better than Optisolis indeed. I canā€™t see it from your pictures but I bet there must be cheerful rainbow party in the beam. Any TIR optic with heavy micro lensed surface will cure the beam easily, just donā€™t expect clean narrow beam.
If only the voltage is at least the same as Optisolis, which already too high for most EDC lights, this could be a better alternative than Optisolis. Output wise, itā€™s very good compared to warm Optisolis.
These Sunlike and Optisolis are (still) the misfits in our flashlighting world. Their native territory are those in general lighting as groups or clusters behind diffused optics.

[Clemence]

8

No they are not suitable for flashlights, just the extreme good CRI makes it tempting to try to fit them anyway. But I like the Tool Ti as it is now with peach-coloured corona. I may make a 4000 lumen quad flashlight out of these as well, with OP reflector and heavily frosted lens.

Waiting for a 219-Optisolisā€¦

9

Thanks for the info Djozz, my copper Tool looks to be the same as my EDC01, iā€™d rather they were like your Ti Tool!
I really like the look of these LEDs, appreciate the review. Do you think having the LED extend into the reflector rather than sit behind it would make much difference?

10

It has a rather defined hotspot actually so the focus seems about right.

If you do this mod you must realise that the led-minus pad is soldered onto the thermal pad of the XP-board and you have to make electrical contact between thermal pad and the led-minus of the ledboard. I did that in a rather clumsy way that I do not recommend (a little wire that of course got in the way of the center piece.)

11

I see that Soul now offers a more powerful SunLike variant, STW9C2SB-S.

Itā€™s rated to 200mA/6V max. Has thermal resistance of 10 K/W which is among the best that Iā€™ve seen so far among MP LEDs.

Too bad that itā€™s 6V but overall it may be goodā€¦

12

Hm. Looks like yolk in the pot :smiley:

13

:person_facepalming: I had to click it, and youā€™re not wrongā€¦

14

I looked at the color spectrum in the datasheet. I still see a peak at 450nm (blue wavelength). I thought these use violet pump instead of the typical blue. But still the amount of blue light emitted is incredibly low for the CCT. Great for use before sleep.

15

The pump is nearly violet though, looks like about 420 nm, just like the cool Optisolis CCTā€™s. The peak at 450nm I would expect to be the first fluorescence peak.

16

Hello everyone.

If anyone knows where to buy STW9C2SB-S CCT6500 please let know.

17

I ordered some short Sunlike strips from LEDrise.eu. Not sure if they have what you are looking for though.

18

Those are 0,2W led but i am looking for 1w. Searched the whole net but cpuld not find:(
Must be somewhere in Asia market for sure.

19

If you read the graph in the OP, you see that the led works perfectly fine at 400mA/3.6V, which is 1.44 W.

20

Thanks for the test!

The Seoul P4 high-cri (the first commonly availbale high Power high-cri LED) also had an orange corona back in the day.