High CRI freaks - check this out.

LEDBE
B

blasterman

Flashlight Enthusiast
Joined
Jul 17, 2008
Messages
1,802
When I first saw this product brief from Bridgelux I had my usual cynical skepticism you all, well, know and love me for. :wave:

Then I looked at the actual spectral graph....then I read it again....then another double take. Yeah, 97 CRI. Look at the red plot and how deep it gets.

As for efficiency, I think Bridgelux has set a record for power hogging LEDs (the numbers are that bad), but I bet it sure looks pretty.

If jtr1962 is lurking around he's got to check the this wild spectral graph out.

http://www.bridgelux.com/products/decorarray.html
 
I resent being called a freak :naughty: :banned: but that spectrum is pretty 🙂

Now who wants to build me one into a Mag? 😀

EDIT: Just spotted the Vf. That just makes it more of a challenge for a modder 🙂
 
Last edited:
Would like the colour temp to be a little higher, but WOW! That's very impressive. I find it interesting that for the cooler option (ok, so only 2700K vs 3000K) it's got a higher output and a higher efficiency (as before, only by a bit). I can see some very nice fixed lighting or flood applications for this, pending a suitable way of powering and cooling it. A Hyperboost would work but only goes up to 1.4A, which is fine for the smaller version but not the larger one.
 
My main PC is down. I'll be mostly MIA for a while, until I can fix it.

Yeah, that spectral graph is wild. First thing popping out at me is why have any output at all past about 700 nm? It makes efficiency tank, but really adds nothing at all to CRI. Second thing is efficacy isn't all that bad considering-they're over 45 lm/W in warm white at 70°C with nearly perfect CRI. That's 3 to 5 times better than the incandescents this is designed to replace. If they dropped the >700 nm output, I bet they could get past 60 lm/W. And I really wish this were available in 3500K to 5000K, even if the CRI numbers were only low 90s.
 
I suspect that there's no getting around the output above 700nM if you want deep red in the spectrum. It won't matter which phosphor type they are using in their blend to get that big red hump it will have a wide spectrum including deep red / near IR. Unfortunately that's where all the efficiency goes down is in generating the red.

Still, nice product. 97 CRI, wow. Thanks for posting the link blasterman.
 
And I really wish this were available in 3500K to 5000K
Click to expand...

Yeah.....the lack of high CCT / high CRI LEDs is increasingly becoming a glaring omission rather than a curiousity. While it's a specialty market, we both know it's not THAT much of a specialty market and high CRI applications typically have a deeper wallet than residential applications. Plus, LEDs would have an even bigger efficiency advantage over tubes in the daylight / high CRI market.

Looking at the graphs the only thing thing that would need to change in order to bump from 3000k to 4500-5000k is the blue and green correct? Amber-orange-red could stay the same. Blue is easy to fix independant of other colors, but green and cyan intensity in high CRI / CCT spectra needs to lie pretty flat. Wondering if it's just not possible with current technology to get cyan-green to hit the proper ratio at high CRI / CCT so nobody tries.
 
Very impressive color figures :thumbsup:

I decided to digitalize the SPD on the datasheet and check them. No surprises, I got almost same results for the typical spectra. Graph correspond to a R4 bin, but fall into a 3 step McAdam ellipse of 3000K (blackbody). CCT of 3040K (typical of 3045K according to datasheet). I got a slightly better CRI yet, of 98 (98,3 exactly), but as well slightly lower R9 (97,46). All figures below 1 point of CRI from the datasheet values, which can be accounted in the margin of error of working from a graph.

The R10 to R16 figures don't correspond to CRI samples, but samples of the R96 metric. With official CIE samples I got:

TCS1

[TD="width: 152"]TCS2[/TD]
[TD="width: 152"]TCS3[/TD]
[TD="width: 107"]TCS4[/TD]
[TD="width: 107"]TCS5[/TD]
[TD="width: 107"]TCS6[/TD]
[TD="width: 107"]TCS7[/TD]
[TD="width: 107"]TCS8[/TD]
[TD="width: 107"]TCS9[/TD]

98,11

[TD="width: 152"]99,69[/TD]
[TD="width: 152"]96,05[/TD]
[TD="width: 107"]96,78[/TD]
[TD="width: 107"]98,74[/TD]
[TD="width: 107"]98,22[/TD]
[TD="width: 107"]99,53[/TD]
[TD="width: 107"]99,55[/TD]
[TD="width: 107"]97,46[/TD]



TCS10

[TD="width: 107"]TCS11[/TD]
[TD="width: 107"]TCS12[/TD]
[TD="width: 107"]TCS13[/TD]
[TD="width: 107"]TCS14[/TD]

93,62

[TD="width: 107"]86,71[/TD]
[TD="width: 107"]98,48[/TD]
[TD="width: 107"]96,89[/TD]
[TD="width: 107"]98,89[/TD]


I have calculated the Color Quality Scale as well, of 96,22. Anything over 95 is excellent quality, so color rendering is almost perfect. Color appearance of the light itself would be excellent too, coordinates correspond almost to a blackbody (Duv=0,00081).

The drawback is the efficacy, obviously. For the 26W model, its 46,47lm/W (1220lm, 26.25W (700mA, 37.5V), Tcase=70ºC, Tj about 85ºC), and the 2700K version has lower lm output yet. With a luminous efficacy of Radiation of 261 lm/W(emited), it means an efficiency below 18%. At 350mA (12,42W) efficacy raises to 54,32lm/W (675lm output). But this device seems a good replacement for 35-50W halogens. Similar lm output at half the power, same color quality, 30x lifetime, reduced heat load to AA system and no IR or likely, UV.

Similar products (design and power) from Bridgelux are priced below 30$, so if they sells this one at similar price, it's cost effective, although still on the long run. But for DIYers, it could be right now a faster cost effective alternative to halogens.

Interested on the comment about the emission over 700nm, I have eliminated it on the datasheet. Result: same CRI, R9, CQS, LER enhanced to 287lm/We (that would result on 7,2lm/W more). Eliminating all over 690nm, only R9 drops (to 96), while LER raises to 296lm/We. Eliminating over 680nm, the same, CRI and CQS unaltered, R9 drops to 93, LER goes to 308lm/We, so still excellent color but that would achieve way better lm output.

So phosphors patterns cost a lot. Two red chips (635 and 660nm peaks) would result on similar metrics of color but way better lm output if the phosphor emision were peaked way shorter but keeping emission along yellow and orage. Unfortunatelly, I dont know any phosphor that does it.

Continuing the experiment, cutting emission over 670nm drops R9 to 88 and LER goes to 322. Cutting at 660nm, CRI lowers to 97, R9 to 76 and CQS to 94, still very good figures with LER=340lm/W. Cutting up to 650nm, similar to obtained by using an standard red LED, Ra=93, R9=51, CQS=92. Not as excellent as before, but still very good color rendering (except for deep reds) and LER=361lm/W. Cutting more drops fast all color metrics.

After this analysis, I want to try the experiment of virtually add the emission of orange and/or red LEDs to a blue one using a remote phosphor, as the 4000 o 5000K of Intematix Chromalit and see what happen with color metrics. I would like to calculate the Gamut Area Index aswell, but implementation on Excel is more complicated.
 
Last edited:
We're looking at the voltage, so the V is capitalised given that it's a unit named after someone (like A, F, and so on). f is lowercase (or should actually be subscripted?) since it specifies what particular voltage we're looking at, the forward voltage in this case.

It's not in the link blasterman provided to but actually in the datasheet linked from that page, 37.5 V at 700 mA for the smaller one and 24.6 V at 2100 mA for the larger one.
 
Last edited:
This is a neat product, but I agree with the others who would like to see higher color temps available. I know it is consumer mentality, but I don't see why it is necessary to keep trying to mimic the yellow light from incandescent. Even at 100 CRI, it doesn't do blue colors any favors. Most people who I persuaded to give 3500K lighting a try liked the whiter light that still had a touch of the warmth.
 
I've noticed that Bridgelux tends to be cautious with their product offerings and they don't typically introduce something without an existing and established market ready to integrate it. Certainly less daring than the folks at Cree who occasionally throw something out there with the intent of 'we think this is really frikken cool, find someway to utilize it'. Between the two companies there ain't nuthin you can't build. Really curious if Cree decides to introduce a competiting product, or if it's not worth their time. Maybe this will increase pressure for a true, 4000-5000k uber_high CRI emitter you folks are asking for.

In any respect, I'd like to be a fly on the wall at Solux corp. :huh:

I have a couple designs I'm working on this will work perfect for. Ya know, looking at the graphs I'm wondering if adding 475-480nm blue in a mixing fashion might be enough to kick the CCT up 1000k or so and keep CRI linear. Heck, add the blue/cyan on a dimmer to you can pivot color temp to taste. A Mean Well 48D will easily drive two of of these and dimming is a piece of cake on those.
 
blasterman said:
I've noticed that Bridgelux tends to be cautious with their product offerings and they don't typically introduce something without an existing and established market ready to integrate it. Certainly less daring than the folks at Cree who occasionally throw something out there with the intent of 'we think this is really frikken cool, find someway to utilize it'. Between the two companies there ain't nuthin you can't build. Really curious if Cree decides to introduce a competiting product, or if it's not worth their time. Maybe this will increase pressure for a true, 4000-5000k uber_high CRI emitter you folks are asking for.
Click to expand...

I see Cree is aiming right at Bridgelux with this product:
http://www.cree.com/products/xlamp_cxa2011.asp
Although there is no Cree super CRI version.
The Bridgelux high CRI array may be their first response to Cree.
 
Personally I like 3000k. I think a ~2000 lumen mule-head mag mod with that hyperboost driver is smelling pretty good right now 🙂
 
wquiles said:
That should be pretty easily handed by the TaskLED Hyperboost:
http://www.taskled.com/techhboost.html
Click to expand...

That's what I said previously, but really only for the smaller one if you're driving them at spec since the Hyperboost maxes out at 1.4A. If we want to drive them to their maximum ratings, the small one takes up to 1.4A (that's actually perfect for the Hyperboost), but the large one takes up to 4A!

Then again it might be a moot point for flashlight builders, with thermal constraints I can't really see anyone using these in a flashlight and powering them past 350 mA or so.

Edit: Now that I've said that I'll sit back and wait for it to happen.
 
Last edited:
Kinnza said:
After this analysis, I want to try the experiment of virtually add the emission of orange and/or red LEDs to a blue one using a remote phosphor, as the 4000 o 5000K of Intematix Chromalit and see what happen with color metrics. I would like to calculate the Gamut Area Index aswell, but implementation on Excel is more complicated.
Click to expand...

I see last Thursday Intematix posted a press release discussing their new phosphor blend that combines three separate material families for a CRI of 98. Hopefully more details with be forthcoming shortly.
http://www.intematix.com/news/34/39...omplishes-Near-Perfect-Light-Quality-for-LEDs


Stephen Lebans
 
Well, I did the simulation of adding virtually red to Chromalit phosphors,with promising results:

SPD of750+R0,2
103,1105,57107,66100,4498,8292,9896,3697,0698,3100,52
3497,54036,84531,92993,12729,42738,22692,72807,53028,83478,1
-0,01589-0,00969-0,00367-0,02120-0,02369-0,02375-0,01791-0,01673-0,01430-0,00892
0,3900,3730,3580,4090,4210,4210,4320,4260,4160,397
0,3490,3520,3540,3460,3440,3440,3600,3610,3630,367
0,2430,2300,2190,2580,2680,2680,2680,2630,2550,240
0,3270,3260,3250,3280,3280,3280,3340,3340,3340,333
305,45316,4325,93293,97287,08270,89291,5294,96301,21312,73
89878186828191929589
71,5789,143,8344,433,7323,2172,3879,4494,2467,24
84,0681,4977,9583,883,1683,2385,3785,6985,4683,88

[TD="width: 107"]750+R0,12[/TD]
[TD="width: 107"]750+R0,06[/TD]
[TD="width: 107"]750+R0,3[/TD]
[TD="width: 107"]750+R0,37[/TD]
[TD="width: 107"]750+R,32+DR,12[/TD]
[TD="width: 107"]840+R,23[/TD]
[TD="width: 107"]840+R,2[/TD]
[TD="width: 107"]840+R,15[/TD]
[TD="width: 107"]840+R,08[/TD]

[TD="bgcolor: #00FF00"]Luminous Efficacy (lm/Wi)[/TD]

[TD="bgcolor: #00FF00"]CCT (K)[/TD]

[TD="bgcolor: #00FF00"]Duv[/TD]

[TD="bgcolor: #00FF00"]CIExyz 1931 x[/TD]

[TD="bgcolor: #00FF00"]CIExyz 1931 y[/TD]

[TD="bgcolor: #00FF00"]CIEUVW 1964 u[/TD]

[TD="bgcolor: #00FF00"]CIEUVW 1964 v[/TD]

[TD="bgcolor: #00FF00"]LER (lm/We)[/TD]

[TD="bgcolor: #00FF00"]CRI (Ra)[/TD]

[TD="bgcolor: #00FF00"]R9[/TD]

[TD="bgcolor: #00FF00"]CQS[/TD]



The title SPD means the chromalit phosphor used (750 or 840), the number after R mean the amount of power of the red LED (1W of white power, or the blue chip inside is assumed). In order to get an idea of the lm/W obtained, I've used the conversion efficacy stated on Intematix datasheet (220lm/Wrad for 750, 200lm/Wrad for 840) rounded to low, and a 50% efficient royal blue chip (Rebel and XT get it currently). Red chip with an eff of 40% (an N2 Cree, for example).

I aimed the simulation to obtain around 2700-2800K, 3000K, 3500K, 4000K and 4500K. I plan to build some lamps for home with this system, I was looking specifically to the 3500-4000K tones I prefer with good color quality. Except when going below 3000K, distance to Blackbody locus is small,in some cases very small (<4 step McAdam ellipse)

One thing I like of the results is all have Duv negative, meaning chromaticity is below planckian locus, toward purplish tones and not green-yellowish ones, as I value specially this characteristic.

AS conclusions, the 70CRI 5000K chromalit can be enhanced adding red (200mW) to 89CRI and 3500K, at 103lm/W. Excellent,IMO.

Using deep red rarely worth, and when it does, on very small amounts (about 50mW, a 20mA LED).

The CL840 can be enhanced aswell to 89CRI by adding just 80mW of red, but obtaining an smaller Duv at slighly lower efficiency that doing it with the 750 (100,5 vs 103,1). Using this system is possible easily to obtain a CRI 95 at 3000K, adding 150mW of red. And as you can see, at 2700-2800K, CRI is still over 90 at very good efficacy (97lm/W).

CQS stands for the NIST proposed Color Quality Scale.

Get 95 CRI in 5000k and sign me up for a couple
Click to expand...

Love the challenges! :devil:

Im going to play with it, but I need to use other phosphors. It is required >5000K? Source reference changes from blackbody radiator to D illuminant when calculating CRI at exactly 5000K, so it is different a 95CRI spectrum of 4999K to one of 5001K.
 
Last edited:
You must log in or register to reply here.

Similar threads

J
Nichia vs Luminus SST-20 high CRI
Replies
5
Views
2K
JoakimFlorence
J
RCantor
UV Emitters - high power - advantages and disadvatages of different varieties
Replies
2
Views
453
PhotonWrangler
PhotonWrangler
L
How to change color temp on LED UFO High Bay
Replies
2
Views
816
JoakimFlorence
J
PhotonWrangler
CRI and bruised skin color rendering
Replies
11
Views
2K
PhotonWrangler
PhotonWrangler
B
Glass sealed LED bulbs turning dim from outgassing
Replies
0
Views
194
Bright+
B

Latest posts

Back
Top