Output and voltage test of 5000K and 3000K Nichia Optisolis leds

This an output and voltage test of the new ultra-high CRI Nichia Optisolis leds. Thanks to Clemence who was able to sell some very early samples of these leds for very little money. A discussion thread on these leds was here. This post is strange enough not about the most important aspects of the Optosolis leds: the tint, CRI and colour reproduction, simply because I do not have the equipment for measuring that. I hope that maukka at some point can do tests on tint and spectrum, especially on spectrum changes at increasing (out of specs) current. All I can say is that the tint looks great and that I do not see obvious tint changes when current is varied from 20 to 1000mA.

These are midpower leds that come in currently 2 colour temperatures: 5000K and 3000K. They are completely new developed types of leds, the 5000K Optsolis is based on a new 420nm blue led, the 3000K Optisolis on a (new?) 445nm blue led. They both do not suffer from the cyan dip and the visible part of their spectra come very close to the standard illuminants for their colour temperature.

The only existing leds (AFAIK) that can match the colour reproduction performance of the Optisolis are the Yuji VTC-series leds, of which I tested the 'D50' version of the 5730 variant a while ago here: https://budgetlightforum.com/t/-/45612. Here are the spectra of this Yuji led and the 5000K Optisolis led next to each other, as you can see they are very similar apart from their primary led peak:

My questions are:

-can the Optisolis leds beat the meagre performance of the Yuji 5730 D50 led?

-Do they have at least a more uniform illuminated die?

To start with the last question: no, their die uniformity is even worse because embedded in the led are two separate dies that are very a-centric. If only Nichia had chosen for 4 dies evenly distributed over the led... . Here's the two leds at very low 0.5 mA current:

The light green specs in the 5000K die I have never seen before

For the output tests I reflowed the leds onto my copper core non-DTP two-pad 'Nichia 119' boards.

The output test was done like all my more recent emitter tests. I described it in detail in my XP-L test. , with two minor differences that should not matter significantly for the results: I used my Integrating sphere no. II instead of no. I, and for the current I used a clamp meter, which appears to measure 0.1A lower than the power supply current-reading that I used before.

In summary: 1) just one led was tested, reflowed on a non-DTP copper board, 2) I used my large version II integrating sphere with high quality luxmeter, 3) the output numbers and voltages were measured with the led close to 'steady state' for each current, so warmed up and settled, you should be able to get these numbers in a well heatsinked flashlight. Mind that these are output numbers of the bare led, in a flashlight there will be losses from light obstructions, lens and optic, 4) output is in 'djozz-lumen' defined as 1/550 of the output of my Sunwayman D40A on high setting, which I hope is close to the real lumen, but at least is consistent over all my emitter tests done in integrating spheres. 5) my output tests usually show great performance at currents outside the maxium current of the datasheet, but I have no data on the longterm survival of the leds at these currents, the lifetime may well be significantly reduced.

Here's the results:

(error correction, under the graph: (A) should be (mA) )

What can be seen?:

*the Yuji led is beaten by a great margin, both in output and voltage. In fact up to 400mA the Optisolis leds are surprisingly efficient, for the 5000K Optisolis at 60mA I calculate 126 lm/W (I compensated for that my djozz-lumen is most likely 9% high compared to the real lumen)

*the output of both Optisolis leds are very stable, during the test I saw no output dropping at any current, even not at 1A. This implies that the temperature sag at least at these for BLF-standards low currents, is low.

*the Optisolis leds are rated for max 65mA 'pulse current', (edit: newer specs just posted by Clemence show max 100mA for the 5000K and 150mA for the 3000K Optisolis) but when mounted well like in my test they easily survive 1A. During the 5000K Optisolis test I kept the led at 500mA for 40 minutes to see any damage, it stayed within 0.5% of the initial output !

*the voltage stays low at all currents, in case you want to use these leds in a single li-ion flashlight, make your current regulated!


As mentioned at the start of this post, this post is not about the tint, but at least the performance is way better than the datasheet promises.So much in fact that these leds unexpectedly are actually usable in a flashlight

Encouraged by the good perfromance of the Optisolis leds, I was thinking of how to build a flashlight with the 5000K one. The obvious form for a led with such non-uniform emitting pattern would be a mule, or a fixed convex lens close to the led (like the Klarus Mi1C or the Sunwayman T26C), but I'm not too fond of their beam patterns. So my goal is a very floody flashlight that still has a more or less directional beam. I got away with an XHP50 once by filing the underside of the OP-reflector of a Convoy S8 away so that the protruding ring was gone and the underside of the reflector was completely flat, the led was too far into the reflector and that mostly smoothed out the black cross, at the cost of throw. The idea was to try this with an S2+, and if any irregularity was leftover in the beam I would solve that by frosting the inside of the lens with sanding paper. And because the beam coming out of this led would be rubbish anyway, I decided on a quad to obtain at least a light with close to a decent output. And so it happened.

I used a grey S2+ with short tube and sanded the rim on the underside of the reflector completely away. Then I opened up the reflector hole to 10mm with a reamer and cleaned it all with running water and soap. I put two layers of Kapton tape on the underside to prevent shorts with the led wires.

The leds were reflowed on a 16mm DTP XM-board. That fits reasonably well, but I did scratch some solder mask away (not in picture yet) at the outside of the electrical pads.

Now both electrical pads are led+ and the central pad (and core of the board) is led- . So I connected both wire-pads with a piece of wire (that also served as a raised barrier for the reflector to push against, obviously a plastic centerpiece does not work, and is not needed), and created a new minus-wire solderpad by scratching some solder mask away to the core and put some solder on top.

This way the large pads that are directly under the dies are on the DTP middle pad of the board. Good for heatsinking but bad for the electrical connection: the setup now suffers from the 'Black Flat' problem of a non-electrically insulated thermal pad .

As I found out the hard way, the only way to get this done is glueing the ledboard on its shelf by Arctic Alumina Adhesive, but not before both surfaces are completely flattened. So I sawed the rim off the upside of the S2+ pill and then sanded it very very flat with the disk sander and then even smoother with fine sandpaper on my dedicated flattended iron block. I made the ledboard just as flat as the pill and then glued it with a thin layer of AAA. So now I had good thermal contact with electrical insulation. and hopefully it could withstand the pressure of the reflector when screwing in the pill .

The driver is a 6x7135 (so 2.1A) biscotti driver, with thanks to Steve (MRsDNF) :THUMBS-UP: for sending me a couple of those with the correct firmware.

And happy enough it all worked and I got a healthy 460 lumen out of the flashlight on highest setting, stable with hardly any thermal sag.

Now for getting the beam right. The reflector did not really center the leds which does not look nice but may actually be better for the beam.

But the beam, although it had no obvious donut holes was not what I want for a nice flashlight like this (beams are at 2.5 meter from the wall):

So I frosted the inside of the lens wit fine sandpaper, with the added advantage of blurring the visibility of those non-centered leds a bit :-) :

That improved the beam a lot, but it was still not really smooth:

So now some heavy frosting with courser sandpaper (180 grit):

And lucky me this time when screwing in the pill, even the leds were centered much better (I have not touched it since to not try my luck again :-) )

With light on low:

And the beam is great now, floody but still directional, with very smooth brightness transition from center to edge:

The output has decreased a bit by the frosting, went from 460 lumen to 424 lumen (lost 8%) but it has become a real gem of a little flashlight.

And now I'm going to enjoy that great tint!

The table is probably incorrectly marked. Current through the led (A)? 240lm at 1000 Ampere?

Yep, it should be mA, I will correct it tonight :slight_smile:

Very promising. Never thought they would survive beyond 200mA!
Nicely done Jos. Thanks for testing

- Clemence

Well well, a quad would yield enough output for a work headlamp and should be compatible with at least some stock optics.
Thanks djozz!

Excellent results.I ordered one of each to try in a DQG Fairy 10180 keychain light on your 119 board,cut down,if it can be.I figured 60 or 70 lumens was doable.Now it looks like 4 optisolis’s on a VR16 board 3v,would be great for a general headlamp,500 to 800 lm on max.The same order has 4 E21’s on VR16 as well.
This was an unexpected,and pleasant surprise.Thanks Djozz and Clemence.
Edit:Agro beat me to it!

Thanks Djozz, very useful info!

What did you mean by this: “The light green specs in the 5000K die I have never seen before ”?

I have been looking at a fair amount of different leds through the microscope and the phosfor materials vary and contain a variety of coloured specs, and this green coloured spec is new to me.

I see, thank you

I added the S2+ build in post #2, really happy with that :slight_smile:

That was a real fast built Jos. You must be very excited to get things done in less than half a day!

- Clemence

Monday is my day off work :-), that is until 3 o’clock when I have to pick up my son from school. But the build did go very well, that is not always so.


I just noticed something in your final built pictures. I’m really curious and somewhat itchy here….
There are… (counting)… five solder balls in between the LED chips. :laughing:

- Clemence

Do some side-by-side testing with another flashlight pointing at some coloured objects, then we can see if the colour rendering is significantly better :slight_smile:

I tried to pull one off with tweezers and it was stuck, so I just left them there.

I can try something like that but I doubt any difference shows up compared to Nichia R9050/R9080 leds. Even with my eyes I do not see anything special if I do not look for it, these differences are subtle! What is at least clear is that qua tint this led does not have the typical Nichia rosy-ness.

Btw, talking about ultrahigh CRI leds, does anyone know a source for Seoul Semiconductors SunLike 3030 leds? I know that they are on the market because they appear in lighting products, but I do not see bare leds anywhere.

Try future electronics

Thanks for the test! Also, a nice mod!

I wish there was a version with only a single die in the center. That would be great for my old 2xAAA Minimag with a Terralux Drop-In. It has a really old, similar looking mid-power LED and a suitable smo reflector.