So what is going on with the Nichia219 on copper?

I had high hopes for some improvement using the 219 on copper, but I did not get more than 1.3amps from a 18650 battery through the led, despite the use of a 2200mA Nanjg driver. Vestureofblood got a bit more from a 2800mA driver, but still under 2A.

So despite of me not being an electronics expert, I did an attempt at testing the Nichia219 on a 16mm Sinkpad to see what is going wrong. And I must admit again (as in my recent MT-G2 measurements) that I did not do extensive measures to calibrate my equipment, so the same disclaimer is valid as in the post about the MT-G2 measurements: the numbers in this post are given as is, without any claim of great accuracy (and I truly admire those that really can do those accurate tests ). But I am confident of the overall trend of how the 219 performs on copper.

Oh, and I did these measurements because I myself was curious about this particular led, it is not that i am planning many more of these tests, I do not have proper equipment for that, and lack the will to do all the checks and controls that go with real accurate testing.

This is the set-up:

This is the led board, well screwed down on lots of aluminium, with thin layer of Arctic Silver in between:

And to be able to measure the output without a Integrating Sphere I placed a reflector on top with a glass lens (this time a cheap AR-coated one from FT), so a ceiling bounce lumen estimate could be obtained that can also be regarded as a value that can actually be achieved in a flashlight build.

The led was directly coupled to the power supply with short thick leads, I used both the current and voltage readings that were given by the power supply. I suspect at this time that the voltage readings from my power supply are about 0.15V higher than the real Vf of the led so please substract 0.15V from the Voltage values in the table and graph to get an idea of the Vf, but more about that later.

So what are my thoughts of this?

1)on an alu board the 219 output levels out at 1600mA, on copper it is around 3A

2)the max output is going up considerably as well, the above values are 'simulated OTF' numbers but I guess the maximum bare led output would be close to 500 lumens (compared to max 300 on aluminium, see match's tests of the 219)

3)this well heatsinked the led can withstand 4.4 amps! (but for the maximum output you don't want more than 3)

4)the Vf is going up fast!, and at max output it is more than 4.2 volts, so even with a single li-ion cell at direct drive you do not get there. (So where is that single li-ion 3 amp boost driver??? )

5)on the other hand, if you do not care about constant current or modes, just skip the driver and run the 219 directly from a li-ion battery (mag incan-style) , how easy is that?

6)you can see some irregularities in the graph, apart from lousy measuring, one reason is that not all values were taken in one session turning the power up, it seems that after the first session at high current, the current at a already measured voltage has increased somewhat (or the other way around: that the Vf at a certain current is somewhat lowered). This supports indeed the idea that you can 'burn in' this led to lower the Vf a bit (no hard evidence though).

So getting more confident about my test results I just tried this:

It was a fully loaded Panasonic CGR18650CH (IMR) battery, and it works well indeed, I did not do a current reading but the initial output was estimated 369 lumen (OTF of the reflector+glass lens), which corresponds to a current of a bit over 2A. It declined steadily in 10 minutes to 313 lumens, nicely following the discharge at 2A according to HKJ's test of this battery.

With the fully loaded battery under the load of this Nichia219 I measured 4.08V with my (good quality, but it is not a calibrated Fluke) DMM. The above graph however, reads 4.2V at the 369 lumens that was measured with this battery, this is why I guess that the voltage readings from my power supply are about 0.15V higher than the real Vf of the led (this of course is another wild guess ).

Conclusion: So what to do to get the most out of the Nichia 219 on copper? A suggestion to Old Lumens would be to power it with four NiMh batteries (max 5.4 volts?) and use a Nanjg 501C 3A driver to burn off the more than 1 excess Volts (but I don't know if this driver will be happy with this amount of heat?). I myself like the idea of just direct drive it off one 18650 li-ion

Thanks for reading,


I ran a Nichia 219 5000K on copper direct drive at 1.6A and it handled it fine. It’s now in my Photon Fanatic custom Ti neck light and performing flawlessing on 700mA. Nichia rates it at 475 lumens at 1.5A. This is NOT the high CRI emitter everyone brags about, but a nicer white emitter that handles more current and produces more output.

Perhaps you’re using the same emitter?

The Nichia 219 Hi CRI in my MBI HF running direct drive off a High Discharge rate 10250 Nuke puts out something like 200 lumens (according to Guy at MBI). I just tested it on a cell with a rested 4.15V and it pulled 2.04A in the flashlight, a solid copper light.

Edit: I see that the copper star is on aluminum, wonder how much better yet the numbers would be on all copper? The emitter mentioned above is on a copper SinkPAD attached to a copper bar.

Some nice info here. I’m hoping someone makes a triple copper board. This way you can get the benefits of more lumens (3 emitters + copper bond) but still keep Vf low since you are running each emitter at 1/3 the amp.

EDC+ makes the Triple Nichia 219 92CRI driven at a low 2A and making 465 lumens, measured. Nice drop-in, all copper construction with silver plating to deter oxidation. Have that as well as the Triple XP-G2 at 720 measured lumens in Solarforce L2P hosts and really like em. Of course, they’re floody with triple TIR optics.

Nice work djozz, thanks!

I too would love to see parallel, XP-G/Nichia, 3up, copper MCPCBs. Only problem if that happened is I’d have to rebuild all my triple lights.

Those VF are high compared to XP-G . If we subtract 0.15V they are still high.

At 1460mA is the 4V an error in the graph?

I used the 92 CRI 4300K Nichia, obtained at Illumination Supply.

If that 2.04 A from the MBI HF is also going through the led, the initial lumen output must even be a bit higher than 200 lumens, closer to 300, but the upbuilding heat in such a small light (even though it is copper, the total surface area is just not very high) is costing lumens when it runs hot.

Although there is some debate on this (someone test this to settle the discussion?), I personally don't think it matters a lot if the heatsink is aluminium or copper, the main bottleneck for the heat path to the outside is the small surface area between led and board (it helps a lot if the led board is copper and even more important that there is no dielectric layer there), once the heat arrives at the heat sink the area the heat must go through is so much higher that any well conducting metal will do.

It was what the powersupply measured, the Vf probably being a bit lower. That reading was from the first run before the led had ever been exposed to more than 2A current. After I did the 4A current test I went back to lower currents and by then the Vf appeared a bit lower. This could be the reason for the inconsistent readings (or it was just my wacky test set-up ).

The small copper light heats up in pulling heat from the emitter but indeed, there’s just not enough mass there to make a huge difference. Also the 10250 Nuke runs out of power very fast at those levels. You can hook up a fresh hot cell for the amperage test and watch it plunge.

I also suspect you’re right about spreading the heat out to anything that conducts it, but wouldn’t a copper sink keep the copper star cooler such that the emitter stays that little bit cooler, and brighter? Only really useful I suppose when an emitter is being pushed hard, way beyond it’s limits. Like taking a 1.5A rated XP-G2 to 4A+. Or taking the Nichia 219 also a 1.5A rated emitter, to 5A and more, all copper would really make a difference there and I’m betting the numbers would prove that out.

Encase it in Titanium and watch it fry. (Ti nut, getting sold on copper for heat)

great work djozz! It’s neat to see how well the sinkpad allows the Nichia to cope with increased current, but it still shows how lumens/ watt drops off a cliff at higher currents (much the same as Match’s and the tests on CPF showed). Still, a 1.5A buck-boost driver would be perfect for this LED.

If you wanted a triple parallel on copper, why not Stinkpad-ify one of the IS 20mm triple stars? It’d be a bit of effort and a drill press would be very useful, but something neat to do. You’d still be stuck with the Carclo triple optics though, so it’ll still be more of a flood light. I’d be half tempted if I didn’t have a list of things to do as long as my arm and no particular need for one :slight_smile:

I did a 219 mod with this light that I had waiting for it, a nice and cheap small thrower, I did a small review on it earlier:

I used the Nichia219 on the Sinkpad that I have done the above measurements with. I cleared the driver from all components and wired the battery + and minus directly through to the led. Assembled it all, the business end now looks like this:

It looks like that black spacer around the led is blocking light, but although it did that before, I sanded that ring down so much that it does not block the light anymore. Also I blackened the inside of the bezel with heater paint to remove the annoying rings in the beam profile (see the review), it looks now like this, nice:

The numbers coming out of this light, although it is completely direct drive now, are still less than I hoped for. On a freshly charged cgr18650ch (IMR) cell tailcap reading is 1.9A, and ceiling bounce lumen estimate is 300 lumen, where I hoped for 2.1A/350lumen. Apparently there is still a bit of voltage loss somewhere in this host. The throw by the way is 13klux@1meter, not at the least impressive, with the original xpe-led it was 25klux@1meter.

Oh well, it has still become a nice flashlight, the only Nichia219 'thrower-ish' light I have, I am looking forward to using it on nature strolls at night. And this is apparently about as far as the Nichia 219 goes in a single li-ion cell flashlight without boost driver.

This is cool stuff. I gotta do something with a H1 B10 using 2S 18650 and a buck driver. I know the output will be much lower, but I want high cri more. Based in your experience so far, what is the max current you would recommend if properly heat sinked? I know I would be taking a risk, but I would like to try the max current that would most likely not blow the LED.

As seems to be the case with most emitters on copper (or even on aluminium), the maximum output is reached long before the led blows (at least with good heatsinking). As can be seen in the graph the maximum output on copper is obtained at about 3A and although the led survives at over 4A, there is no point going over 3A.

Support the cause here, the more demand Intl Outdoor sees the more likely they are to make it. It should happen as they said but it takes them quite some time to make it.

Lets spawn their thread and emails with requests of XP-G/Nichia 219 Noctigon triple. :stuck_out_tongue:

And I’m not going 2A triple or something, direct drive triple 0:)
Whether it’s going to be on alu or IO manages to make a copper triple board.

I admit this all goes a little above my head.

But if direct drive isn’t doing it for you, would 2x18350 and a buck driver not be an option?

Thanks for sharing djozz.
Very valuable info!

Hope you will do more stuff in the future if new emitters appear. :slight_smile:

That would do to get the led at 3A, but since the OTF output of a single 219 light will never exceed 450 lumen, I want the light to be small, 2x18650 is just too big. Thinking of that I can just throw in the 2 Efest 16340 IMR batteries that I have in this light, direct drive will be no problem because the batteries can not deliver more than 3A anyway at this voltage (I doubt even if it reaches 2A with them, thinking of the high Vf of the Nichia) (to be continued...)

So in the end the waiting is indeed on the triple copper board...


Just encase you missed it, I did say 18350, not 18650. So 2x18350 will fit most 1x18650 tubes. (I have some AW IMR 18350 that work brilliantly in my Crelant lights).

Also I may have misunderstood, but are you saying you can direct drive off of 2x16340’s? i.e. direct drive from 8.4v?

And yes there is no proper driver for using Nichia 219 in general or XM-L2.
No good single or multiple cell drivers able to do boost/buck at 1.5A or 3.0A. (Efficient and programmable with a flat stabilized output curve.)

If the batteries suck enough then yes. But you will overload them, the batteries may not be able to give you the necessary current so they drop voltage a lot. If it’s some tiny and low capacity batteries, cheap brand batteries, that is quite common.

Sure you can’t do that with 2x good 26650 as they won’t drop and will burn it.

Also overloading the batteries means loosing usable capacity. You get more runtime when they are not overloaded.