maukka’s test
No, it did not arrive yet, must be stuck somewhere. Usually packages show up again at some point, I’m waiting for that 
what would be the differences with the 6500 9080 and say a 5700k 9050 ?
I have a build but wondering on these two choices.
pro’s con’s?
thanks.
I see some D320 CRI70 E21 in your shop…
can we expect it to have 43% higher efficacy than D220 in your tests?
If yes, a quad would do 1000 lumens with over 170 lm/W in XM footprint….
We (me and Jensen567) tested the singles and quadtrix using those E21A sm505 D320 R7000 so we can related the 144AM sm653 E1200 R7000 results from earlier Djozz’s tests
Using this graph we can estimate lumen output, by interpolation using data sheet. The results is 2173 lumens at 11,2A (35,6 watt) and it was still climbing albeit at shallow rate. The peak estimated at 12,4A. The test stopped because my PSU couldn’t feed it no more.
That means a single E21A D320 output peak was 2173/4 = 543 lumens. My recommended max current is 80% of the maximum ~12A which is 9,6 A for 4p setup or 2,4 A for 1s (single). Single E21A D320 should outputs ~519 lumens at 2,4 A.
1000 lumens output for 4p E21As is an easy task and very safe for the E21A quadtrix. You should get 1000 lumens at 3,25A in 4p config / 1,625A in 2s2p / 812,5mA in 4s configuration.
Each LED will only get 812,5 mA which is way under the OEM rated current (1400 mA)
I hope you’ll get them soon. Jensen’s earlier order also took 2 months delivery time.
- Clemence
This is not an easy question unless you tell us your application first.
- Clemence
Clemence, I am considering modding a coast hx5. This light has a tiny board mounted on a tower. I wonder if I could cut just the very center of your single board and then scratch the white insulation in the right place to expose copper for electrical contact?
Counting pixels I redid your calculations and arrived at 1000 lm@3.02A with Vf of 2.91V. 114 lm/W. A far cry from my previous calculations…do you have any idea why is it so?
EDIT: I made a mistake. It’s over 3.2 A with slightly higher Vf. Still over 100 lm/W, but even farther from 170…
[quote=Agro]
[quote=clemence]
Use it as rough estimation only.
The real output must be much higher though. I used wall bounce method only to check the maximum current capability. Let’s wait Djozz test result if he has the time
- Clemence
OK 
In my Armytek Wizard Pro testing, at low current (firefly modes) even the R9080 is more efficient than XHP50 CRI70. Then as the current went up, the efficiency got lower.
- Clemence
I did my first mod with the E21A sm603 R9080, and it makes me like cool white tints much much more
The Optisolis…
- Clemence
Very curious about these
Have you tried mixing them with 5k?
No, I just have the 6500K ones.
And mixing leds is a bit of a believing thing, not real. For me it would only be useful to obtain a colour temperature that is not available, so out of shortage of supply. If you mix two colour temperatures, the spectra simply add up and you have no different tint than if you use just one led of the in-between colour temperature, but not as good because the light will not be perfectly mixed.
Correct! Mixing several different CCTs will also (in most cases) results in lower CRI than individual CCT. You can check Maukka’s test to see it. It’s more like a trick.
- Clemence
Of course you are right. My goal is to get the high noon tint of 5700K by mixing the 5K and 6.5K. Re the imperfect blend: I want just a mule and I am sure that for this specific quad board with the LEDs positioned so closely to each other, the two tints would blend perfectly when the light is more than a few centimeters away from a target.
That sounds a good reason and a good way to mix the leds
Sorry guys, I’ll try to stock more CCT later. The VR16SP4 was designed for CCT switching (Warm/Cool option in a room), not mixing in the first place, to compete in general lighting applications. But later found that it’s a perfect little board for tint/CCT mixing too. The ideal solution is of course to use fixed CCT.
- Clemence