Nichia 219C, testing a 5000K 83CRI emitter, comparing with a XP-G2 S4 2B and other leds

Ah, while I was sleeping a new emitter came out that I almost missed (thanks ImA for the alert)! Thanks djozz for your testing! Might have to order a couple to sit on my workbench for months and months with no use . . . ahem, I mean to mod some lights with.

-Garry

If your hobby time is limited, the choice is: get some emitters and let them sit there or keep the workbench empty and hope for Nichia to come up in the coming months with a 90CRI version after all :-)

(although they claim not to work on that)

Kawiboy emailed some pictures of his shaved XHP70 to me for posting because his computer was not able to post them on BLF, here they are 4 of them :-) :

1:

2:

3:

4:

I know a few people ordered the 4000k from Cutter. If any of you have extras, I’m looking to buy 2-3. If I needed more, I’d just buy from Cutter myself, but for 2 the shipping isn’t quite worth it to me.

Solved by a helpful Aussie

For an X6 triple I think I might do a tint mix with 2 4000k and 1 5000k.

Ola Djozz,

Why do I not see a donuthole? Is that because of the shaved dome or am I missing something here?
Also, is it place in a Fenix TK61?

Schöne Grüße, :wink:
Nico

Attempted acetone dedome of a 4000k 219C, dome became crumbly (not soft at all) after 90 minutes in acetone. Most of the dome came off very easily, leaving a few bits stuck on the die. There is phosphor under the white stuff, which was removed to reveal a gold-plated substrate. The remaining dome was carefully cut away using a razor blade in successive slices, leading to a good looking result.

Reflector and beamshot (underexposed to show tint)


Hmm, that slicing may have been a little overzealeous especially near the edges of the die… I wonder what the CRI is now…

Ha Nico, those pictures were from KawiBoy1428, I posted them for him. So I can not answer your questions :-(

testedandbaked, looks like that one is now making UltraViolet light that could be damaging… the blue/purple is where too much phosphor came off. You can even see it in the beamshot. Be careful with that one!

So maybe the Nichia junctions produce a shorter wave length than the Crees do and that is how they are able to get higher CRI?

I believe the wavelength is roughly (no instruments for quantification sorry!) 420-440nm, perhaps a touch more blue than shown by the photos. Definitely not UV, as it’s more blue than a 405nm bluray laser.

I’ve since switched LEDs, this time dedomed by a high razor shave then sanding down with 360~~600~~>1000->2000 grade sandpaper, turned out the dome sands well with a light touch. Minimal some tint shift with this method, about 4000k to 3500k.

Ah, the lovely Nichia 219C…

1. 0.01A for 00.45 lumens
2. 0.03A for 12.94
3. 0.16A for 56
4. 0.56A for 202.52
5. 2.60A for 814.2
6. 7.25A for 1804.35
7. 17.33A for 3301.65

This is a small Convoy M2, with a 19mm thick copper spacer/heat sink I made this morning and a quad mcpcb with 4 of these 219C’s. :bigsmile: I kept the lid on it by using 22ga wires, good thing too huh? Yeah yeah yeah, not like me to limit one I know, but I’ve seen 4 of these draw over 21 amp and didn’t want that in this smaller light, large Cu sink or no.

Cell used for the above numbers is an LG HE2 fresh off the charger. Don’t really want to know what a couple of the cells with more potential might do here…

I’m going to be ordering some 4000k from Cutter soon. Thinking about getting some extras to qualify for free shipping and reselling them. How much interest is there at $6 per LED?

Wow, I don’t think any single die Cree has survived 17A+ before! Maybe helped by the 219C’s lack of bond wires? Thanks for the data! Did the LED blue-shift by any chance at those insane currents?

EDIT: oops, realised you were talking about a quad setup!

testedandbaked, it was 4 of em…read Quad… I built a Quad with Carclo optics in a Convoy M2.

As such, a Quad with Cree’s won’t draw 17A even with 18ga leads. I knew my Ti X6 did 21.3Amps on the 20ga leads so I tried to limit this one a bit with smaller wire.

Single Cell, only one 18650, I’ll probably test an LG HE4 and Samsung 30Q as well as the Efest 35A, but I used the LG HE-2 for this one initially.

Point being, the lower Vf of these Nichia’s can lead to some crazy high current being drawn, so be careful what you build with them!

Edit: :wink:

Interesting test Dale, but I’m curious as Djozz tested the 219C to produce over 1000 lm at 4A, but you’re only getting 3300 lm? it is because of the loss with the carclo optic?

You easily get this loss in a flashlight. 17A is 4.25A/emitter, in djozz-lumen 1125lm/emitter. Out-the-front Dale measured 825lm/emitter, that is 27% less. This is not bad. The factors involved are optics loss, heat loss, and differences between Dale's lumen and my lumen (the variation between the personal 'lumens' appears significant, we all try the best we can and it can still easily differ by 20%)

Like djozz says, there are variations between testing styles and equipment. I believe djozz bench tests while I test an actual flashlight, so it’ interesting to see what the emitters are capable of vs what we can expect in real time.

Also, on this light, the small TIR in the optics is going to be less efficient than a larger single TIR or reflector, just primarily due to the small size necessary. Remember, there are 4 TIR’s crammed into a 24mm diameter optic here. 4 of these 219C’s in a larger quad with bigger lens/optics would almost certainly yield a very different output.

Maybe I’ll put 4 of these into my SupFire M6 and see what they do with the larger reflectors in a modified soup can light? It’s doing around 5000 lumens now with XM-L2’s.

Edit: By the way, a freshly charged Samsung 30Q showed 17.86A for 3456.90 lumens. :wink:

These Nichias are a straight swap on an XP-G (2) emitter board, right? Just heat, remove XP-G (2), place Nichia, reheat to attach?

I seem to remember some modification that needed done on the XP-G emitter boards for the previous Nichias (219B's?).

-Garry

They are a straight swap. Just note that the board is a tiny bit thinner than the XPG2s, this usually doesn’t affect anything however.