Using osram color calculator and integrated measurements at 700mA
3x6500+1x2000 = 5092K –0.0080 , CCT too high
2x6500+1x5700+1x2000 = 4891K –0.0080
1x6500+2x5700+1x2000 = 4697K –0.0079
1x6500+1x5700+ 1x5000+2000= 4548K-0.0075
4500K = 4467 –0.0038, so with the last mix you can get a lower duv at ~4500K, but not quite like sw45k, which at 2.8A must be under –0.01
Clemence had a nice chart showing how the tint results with each mix.
https://www.flickr.com/photos/140459745@N04/46543309595/in/dateposted-friend/
I find the 4500K E21A already nice and rosy.
If you mix 2x 2000K & 2x6500 you will get VERY rosy. In this extreme case, the tint mixxing is very important, so you need to use the right optic.
VERY true,
duv of sw45k is lower than the duv of E21a and Optisolis
if you want sw45k tint, why not use sw45k? 
I tried E21a, and also Optisolis, to give myself a chance to see them for myself.
then I went back to sw45k, and I am happier with the more pink tint (lower duv)
it is also easier to buy and reflow mcpcb for sw45k, than for E21a and Optisolis.
Mixing E21A is advantageous for lower CCTs because there are no available LEDs with rosy tint at low CCTs and we can achieve <–0.01 duv by mixing 2000K+ 4500K or above.
good point, maybe… depends on what CCT is the desired objective
your post with mixing calculations were not lower CCT than sw45k, so I dont see a good reason to do that,
sw45k already works without the extra hassle of mixing different CCT, plus mixing LEDs adds an additonal requirement to use optics to blend the separate LED beams
Yeah, drgentr asked for 4500K so I tried to get close to it, I agree it’s not worth it.
For lower CCTs, here are the tints I get in a tint ramping flashlight for example :
Using 6500K-2000K would have lowered the duv even more, as shown in maukka’s test, and for a larger portion of the CCT range. (also in this flashlight the leds run at half current at the middle of the tint ramp, so the duv is not as low as it could be )
thanks for taking the time to share info
I really appreciate the picture you posted, makes it easy to see what is happening, when using a 2000k LED as the bottom CCT of the mix
even in stock form, the IF25 with built in tint mixing that uses a 2700k as the bottom CCT, looks very interesting. But I dont know the duv values
You’re welcome.
Yeah 2000K really bring down the mix into the rosy region compared to 2700K thanks to that BBL curvature into the red.
I made some measurements of the stock IF25 here , of course the sst-20 6500K are pretty bad. A few posts below I put sliced LH351D in it. But the CCT range is limited compared to E21A which allows to mix very spaced CCTs
I didn’t get a reply in the Hank thread so I’m going to try asking here in case anyone knows. How much does the CCT on the E21A get pushed up when it’s run at 1.2A versus the 0.7A in your tests? Like say 5000K would be more like 5500K and 4500K becomes more like 5000K?
There is a spreadsheet with all the measurements : https://docs.google.com/spreadsheets/d/1CERSuF2JvYtjsoyPh-wEZ0dG6fSI-FDCjcTEVMSppiw/edit?usp=sharing
Thanks I didn’t know about that. CCT in those tests are fairly steady but how is it measured?
In Maukka’s review of the KR4 here: Review: Noctigon KR4 (18650, 4xE21A 5000K, CRI96) the 5000K model shifted quite a bit from lowest power to highest power setting.
I don’t know what to make of it.
The MCPCB in the KR4 is not nearly as good as the Virence ones, thus the LEDs heat more and there is more tint shift.
These are DTP MCPCBs, E21A/E17A don’t have a thermal pad so you can’t use those (unless you have a grounded cathode driver and DTP the cathode but that’s a special case and even then only one side will have good thermals). They have a poor thermal conductivity dielectric layer and will perform badly. Virence’s MCPCBs have a very thin (relatively) high conductivity dielectric layer.
Edit : new ones even have 3oz copper pours ! https://www.eurekatronix.com/product-page/er10s1-21
I’m pretty sure they can make same or better MCPCB.
Yes they can do similar non DTP MCPCBs, up to 8W/m.K dielectric (they don’t detail the thickness) and apparently up to 3oz copper pours. Must be very expensive though.
Does anybody have a copy of Maukka’s table with all the measurements ? it’s not accessible anymore, I’m pretty sure I had it saved for offline use and yet I can’t find it in my Gdrive : This one
He hasn’t been connected for 10 months, I hope he’s well and just has other interests. Has anybody got in contact with him since then ?
No I’m talking about the Google spreadsheet which has more data.
agree the spreadsheet is gone
this post has all the individual data though
maybe that can help you find the data youre looking for. (what data are you seeking?)
