Attempt to reduce OTF light loss with Carclo optics

Inspired by Djozz experiment, I decided to find a way to reduce the output loss in the super popular Carclo optics. In this case I’m using the triple 10507

LED used were the well known Cree XPG to represent small die LED. Next I will repeat the test with Nichia 319A to represent bigger die LEDs. I carefully enlarge the three positioning hole in Noctigon MCPCB to 2,7mm. This way the optic can sit directly on the LED while still maintain the correct position. All three optic legs are snugly fit in the holes.
The idea is to capture as much light as possible from the LED. Carclo optics designed for multiple brands of LED with various heights. So the LED input aperture height adjusted to accommodate the tallest LED this optic targeted, plus a little bit for safety clearance. This is OK with tall LED but inefficient with thin and(or) large LES size LEDs such as E21A, 219B, XPL-HI, XML2, Samsung LH351D, etc…

XPG = 0,73mm thick; 319A = 0,52mm thick; Carclo 10507 clearance = 1mm

This also means the focal point vary between LEDs. I found the best focus attained with the focal point set to slightly below the phosphor layer. In this experiment the goal is not the best focus or the prettiest beam.

Cree XPG

Nichia 319A

319A has thinner ceramic substrate and larger LES

RESULT

All test done in 26°C - 27°C air conditioned room. 285mA current

So, the option are clear: higher intensity or brighter output? It’s your choice. I prefer the latter because by making use all of the emitted light we should get nicer mixed tint, especially optics with micro lenses (peebled).

Happy modding…

  • Clemence

Very cool test Clemence. Thanks for sharing.

Very nice! I did not know that Carclo optics were designed for imperfect focus of many leds. 3% more light extraction is huge IMO because it is completely free.

The enlargement of the holes of a Noctigon triple board comes with a risk of shorting the led-minus trace with the core of the board (copper bits crossing the dielectric layer). Did that happen in the experiment? I assume you used a drill to enlarge the holes, and a sharp drill bit will help.

I’d be interested in seeing how this mod affects various different LEDs, including XPL HI, LH351D and SST20.

In the picture his solder blob on the negative trace does look offly close to the drilled whole. It appears to be touching the edge.

I think this mod should be safe to do, but it might be wise to take extra care to insure the solder blob on that trace does not go towards the hole.

One option to avoid that risk is to only drill 3 of the holes (in a quad) or 2 of the holes (in a triple). Do not drill the hole near the negative lead at all…. instead use some cutters and snip off the bottom of that post on the optic.

Fortunately I didn’t short the holes :slight_smile:

Noctigon, Sinkpad, and Kerui have ~ 0,1mm thick dielectric layer, and less likely to short as long you use sharp drill (I used standard Nachi HSS center drill) followed with sharp single flute countersunk drill bit (0,1mm chamfer is enough in most cases). Altering the shape of Virence board is another thing. Very hard dielectric and extreme dielectric thinness (~0,01mm) requires tungsten carbide and fine grit sandpaper finishing to prevent shorting.

You can see in the picture below, the dielectric layer is clean from any copper bits. Dull drill bits and high pressure usually results in shorting. Copper is very soft and easily fused together with too much pressure.

- Clemence

Instead of drill pcb, is it also ok to shave a bit of the optic legs to make them a bit shorter?

That probably won’t be as accurate. A counterbore with the correct sized pilot, in a drill press or mill (where you can adjust the stop on the quill to ensure perpendicularity and accuracy of depth) would be the best way.

What I was thinking of, instead, is to cut the bosses off the legs, then holes could be drilled into the legs for insertion of rods for the alignment holes on the PCB. This would make it easier to adjust the height of the legs all at the same time. It would require however accurate drilling of the holes.

Great test Clemence! I wonder if the same can be done with other lenses (Emisar D4S? :wink: ).
And I also wonder how much throw can be gained but focusing them properly…

You actually gained 4.56%, extra 45.6 OTF lm on 1000 OTF lm light. :slight_smile:

Each TIR size is too small for most 3535 leds. We defenetly need other triple boards to work with osram small die leds. They will probably need lots of fine adjustment work but who knows.
I think such test could be more usefull if you could add lux test results. I guess you have lost some throw after this mod.

OP updated with more information.

Enjoy…

  • Clemence

Nice work clemence. The hot spot looks very clean. :beer:

Yep but slightly floodier. TBH, this is my first time to see 319A under Carclo 10507. Never thought the hex die still visible. I thought it would be rounder just like in single optics. Perhaps 319A helps to lessen petal artifacts in very tight optics or reflectors.

- Clemence

Great work Clemence! :wink:

Oh my…sorry guys. :person_facepalming:

:laughing:
OK, I changed my avatar!

If you could only find a 1.9mm center hole annular cutter, that make trimming the carclo studs down a breeze. Possibly just turning it by hand would work since were dealing with plastic. I couldn’t find one though.

Trimming down PMMA optics are so much easier than most people would’ve thought. I’ll post a video how I do it later.

- Clemence