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

Flashlight Modding and DIY Parts
99

It is the king, the previous title holder was my X6 with SBT-70 at just over 17A. But that is a single emitter on a single cell, so it still holds a title of it’s own, different league. :wink:

The LG HE-4, freshly charged, will hit 21.3A in this one, with 4123 lumens.

Oddly enough, it’s little brother the HE-2 only allows 19.54A but has the highest lumens number at 4278. I always find it interesting how the cells act and react. Ya just never know which one will be the top dog in a given light.

A pic of my daughter’s old room, brightly painted with enough colored objects around to get a good idea how this one works…

Here’s the base shot, set as white balance before taking the above pic…

100

I had one 219C leftover (more coming..) and because I want them dedomed as well for a build, I went into the slicing business. Never really been there before, I'm a hot dedomer and quite disturbed that my good old method does not work on these.

I had one slightly succesful slicing expereince on a Luxeon Q before, and remembered from that one that I needed a slicing guide. The Nicha dome being so stiff I reckoned that it needs slicing thin slices at a time so here's how I solved that (anyone done that before?):

I have this singfire SS C3 waiting for a build sometime (it is actually too heavy for my taste), with a nice brass pill with flat ledboard base. Sanded that base a bit flatter even, and the flashlight head is a pretty convenient adjustable slicing device. It needs a pretty stiff slicing blade, so I used a spare blade for my little Davids planer. One of those break-blades for utility knives should work as well, as well as a blade for the traditional Stanley knife.

With pill upside down, by slowly screwing the pill out (one 20th turn at a time) the dome could be removed slice by slice, making sure the blade was at about the same angle every time and making long smooth slices with a minimum of pressure. The ledboard was pushed down onto the pill pressing my nail right next to the led. This only works well btw if the led is reflowed nice and flat on the ledboard, or the slicing will not be plane with the die.

I ended just above the die, dangerously close in fact, a fraction of the mm, you can see that at the right side of the die the silicone already crumbled a bit and makes a dip.

It helped for controlling the slicing that it was done under a stereo microscope.

At the right at 0.5mA.

From how that looks, you could sell that :-) . I have not proved the result by making a thrower with it that rivals the latest dedomed XP-G2 S4, but it should work alright I guess, at least the tint seems a nice 4000K to me (but I want to see that of course ;-) )..

101

^ Damn djozz. Beautiful craftsmanship there. Congrats and thank for all the tips you just gave. Looking forward to hearing more about how this dedomed emitter works for you.

102

Not only the skilled surgery, but check out the macro photography!

103

Anybody know if it will be offered in 4500 or 4000k? I suppose I could just shave the dome until I get the tint I want… Does shaving affect the CRI at all?

104

welight (=Cutter) said that they will have the 4000K 83CRI soon

105

in stock now

106

Are they on the website?

107

A bit disappointing result. I built the sliced Nichia 219C above into a UF-1504, with the prototype LD2 driver (that still acts funny when mode changing with the clicky, but high setting works). The current through the Nichia was 5.85A (long live the low Vf), about what the driver is supposed to deliver when in regulation, and close to the ideal current for the led for maximum output. According to the graph of the OP, the output of the Nichia 219C at 5.85A should be about the same as the output of the XP-G2 S4 2B at 4.4A. n10sivern measured this XP-G2 S4 2B led in a UF-1504 at 458kcd where the current was 3.8A. So if this Nichia would perform as well as a dedomed XP-G2, the throw should be above 458kcd, say 480. But what I just measured was just 296kcd, that is 40% less, way less than what any of the normal variations, like lens quality, bin, focus can explain . So this sliced led does not behave like a dedomed Cree led for whatever reason. (or something else is wrong, I ruined the led somehow, I don't know).

The beam is nice enough, a perfect evenly illuminated square (with some 'corona' around it caused by diffused light through the leftover silicone at the side of the die) of very nice tint, but what counts in the end is sheer throw of course. A quick beamshot (not completely sharp focussed):

108

Has anybody tested or have theories on the affect of shaving or dedoming on CRI?

I still have a couple 5000k sitting on my desk, but I prefer 4500k. I’d be awesome if shaving can get the tint there, but pointless if I lose the High CRI in the process.

109

I’m guessing it has to do with the surface quality of the sliced face scattering the light coming from the LED. Thoughts?

110

Essentially like a dirty dome? I know I have seen threads of people polishing sliced domes.

111

Of the (many, various) 80+ CRI leds that I dedomed thusfar, dedoming makes the tint warmer but the tint always stayed nice, and colour reproduction stayed good. I can not measure CRI, so it is a guess, but I think CRI is not much ruined by dedoming.

This shaved Nichia 219C led I guess at 4500K, so it had a limited tint shift (may or may not have something to do with the quality of the dedoming effect)

112

My gut tells me that CRI is the product of the phosphors used, therefore unaffected by dome slicing. But obviously the dome has an affect on tint, so…… I wonder if MEM has tools for measuring CRI….

113

A lot of high-CRI led have the magic high-CRI producing fairy dust mixed into the dome material. I saw some microscope shots of some Nichia LEDs where red granules were embedded in the dome.

114

Yes they are

115

We do a lot of spectral testing here for customers, as a comparative point. When you test with and without a secondary optic you can get up to 300Kelvin shift, the logic would say a dome on a led is a primary optic so the same shifts should be expected

116

Djozz, they are the same dome style as a 219B, correct? Very similar? I have dedomed a few 219Bs, they can fool you. Nichia has a clear-coat layer ON the phosphor. It’s perfect after the de-dome, to explain what I mean, the die looks like one of these CREEs after a de-dome (picture below), making you believe there’s extra material stuck on because it’s glossy—I found it to be filled with phosphor. You do not want to remove the glossy layer after the dome, then.

I use a 2-step de-dome procedure now that is a chemical de-dome in about 4 hours, I did a time-lapse video showing the dome detach right around 4-5 hours if done with a fast solvent followed by a slow solvent because of varying plastic types in the dome. I first attack the base glue, then swell the dome until it pops by itself. No slitting, I do not touch the dome.

117

MEM, in your opinion should the CRI be unaffected by dedoming or slicing as long as we don’t remove that layer you described?

118

Sorry, I had completely missed where you had asked about CRI measurement before I last posted. I have a tendency to catch the start/end of threads when I’m in a hurry to read. I need BLF-subscribed posts on a 24/7 audio translated PA stream that I can listen to as they come in over horn speakers throughout my home and shop. :bigsmile:

OK, to explain this, you must first understand how difficult it can be for one to fully explain. Even the simple answer of a yes or no, it’s always dependent on a few different refractive properties, phosphor thickness, bonding, sealing layer contour, etc etc. IMO, I think a good de-dome is still going to change CCT noticeably enough. However if you read on…

We deal with two primary rating systems: CCT & CRI. We expect CCT to directly impact CRI-rating, in some reasonably predictable fashion. People go as far to assume this is an automatic change and the two are linked. We know CCT will change with a de-dome, and that means a CRI change. Well, technically that is not true. CRI is independent.

About the CRI, first of all. The latest “news flash” of high-lux LEDs that also have a “high CRI” rating is something we mostly have not been used to seeing or hearing about—especially when our focus gets affixed on CREE LEDs and their numbers. These trends tend to form opinions about CRI with some mistruths involved about de-doming. I would highly recommend that you read this webinar from OSRAM Opto, about the entire nature of CRI, and newer CRI rating types which exist (Ra, R9), that take things like color saturation into consideration, as well. This PDF will help many better understand how “CRI” is developed, what “CRI” actually is telling us, and also how effective the CRI rating system can and cannot be across the large scope of all available LED choices out there.

The PDF webinar is for download here: CRI Sources Explained

So even though intensity increases, and CCT will change, the same colors are technically still there in the LED. Those individual wavelength regions are further influenced by lens design, such as if the lens collimates all available colors at the same image projection without throwing some color away, due to aberration, for example.

I have a method I feel might possibly work. I have a 50”x50” fresnel lens in a frame, with a white wall near the back. The effect is interesting, because the fresnel seems to take on the full tint of the projection, and it is this displayed, like a TV channel. So I’ve wondered about the Adobe CIE color system combined with “RAW” photos. Let’s say I aim the camera so that the entire image is the fresnel lens lit by an LED projection, the photo should measure the tint, to some degree better than a human eye I would expect. This data can then be seen in Adobe. So I would setup, take a few baseline images, then de-dome, redo the test, and see what Adobe says the new CIE data is composed of. Seem or sound logical? I haven’t tried this yet, until I realized how well the entire surface took on the tint, even if hit by a 3x3” square of light, the entire lens still glows evenly. I could do some tests, but I haven’t thought this all the way through, so I don’t know if it truly is a good idea to start doing. I suppose it’s better than catching the pure hotspot alone in a photo and trying to analyze that—then other tints are involved (from small objects, wall reflections, etc) and would create erroneous (though possibly minute) extra color data.

I will try it, but must get some bezels turned out right now. Cerakote runs in the morning. Won’t be tonight.