How to make a square LED into a circle (round spot beam!)

Flashlight Modding and DIY Parts
1

How to make a square LED into a circle (round spot beam!)
This is an idea I’ve had for a while but just never got around to trying it out.

The concept is simple, mask the square LED die with a thin material with a hole in it to make a clean circular spot instead of a square (for aspheric throwers)

Step 1: get some aluminum foil and cut out a square. It’s ok if the square is big, we can trim it down later.

Step 2: find some hard material to sandwich on either side of the foil. I used cardboard but something like thin aluminum or wood or acrylic will work better. We want it to not be squishy or flexible.

Step 3: drill through the materials with a 1mm diameter drill bit (or other size, I chose 1mm because the LED I used is a white flat with a 1mm x 1mm die. This way I get a perfect circle but with minimal lumen loss.

Step 4: remove the foil, if there are burrs then press it down against a flat surface to flatten it and then manually spin the drill bit through the hole to shave any excess burrs off.

Step 5: Put the foil on the LED die and center it such that you can only see the yellow of the die. A small flashlight and a magnifying glass can help with this.

Step 6: Glue it on with some epoxy or other temperature resistant glue (some LEDs can get up to 100+ C)

Now you have an LED that produces a round spot!
.
.
.
Notes / observations:
-foil is very flimsy and hard to work with because it is so thin and because it is aluminum. I would recommend using a steel shim between 0.001” and 0.005” instead for added rigidity and makes it easier to drill a perfect circle with no burrs.

(I have ordered various steel shims so I will test them and see which thickness I would recommend)

-since the foil is reflective there is slightly more artifacts around the spot in the beam. I will try painting with black spray paint or possibly blackening the steel shim with black oxide (similar to anodizing)

-the foil is so thin that this would still work in combination with a RLT collar, however the outer reflective surface will definitely need to be painted black in that case to avoid misalignment reflections.

-the intensity of the spot was measured before and after the modification. In fact, it did not reduce performance, it INCREASED it.
.

While the lumens were obviously reduced (35% less light emitting area) the intensity increased by 15%.

I was not expecting this at all, but it seems like covering the LED with foil worked similar to a light recycling collar.

A 1.03mm square LED has 35% more area than a 1mm circle, but the covered LED had 15% higher relative lux reading.

This means that the lumens from the covered areas were recycled at nearly 50% efficiency back into the beam!

I should also mention that this test was done at low LED currents only, so thermals may change depending on how overdriven the LED is and how much of the die is covered.

At least now anyone can turn their own LED into an SBT-70 without needing to spend $70 :slight_smile:

2

Excellent. Very well done sir. Sounds like you are soon to have a new business on your hands. Better order a bunch of boards, emitters and foil. The request will be filling up your inbox soon. :smiley:

3

Whoa… we might be onto something here!

I can only imagine if the foil is implemented precisely and thoughtfully, we could actually provide an improved thermal path for the phosphor. Not only is there light recycling and increased luminance due to that, but peak current could increase as well!

Imagine a hybridization of a centering ring and a collar! It’d be made of aluminum and create a high conductivity thermal path from the phosphor to the aluminum reflector or other top-side clamp in an aspheric!

Of course this would really only be applicable to emitters without bond wires or those with them safely buried under silicone.

4

Hahaha thanks :slight_smile: it would be a pretty popular mod I think!

5

Ah that is true, I didn’t think about the possible thermal benefits.
The aluminum is far too thin to carry any heat away, it is 0.01mm thick, but maybe something thicker that tapers down towards the hole would work :slight_smile:

6

I got my steel shims (link https://www.amazon.ca/Precision-Carbon-Assortment-Temper-Chemistry/dp/B00065UX42) and here’s what I found:

0.001” (0.025mm) is too thin, not much different from aluminum foil
0.002” (0.051mm) worked well, very little distortion around the hole
0.003” (0.076mm) was best, the hole was almost perfectly round and no defects around the edges, even when drilled without using the sandwich method.
0.005” (0.127mm) I didn’t try because that’s starting to get too thick and the vertical height will interfere with the LED output.

Here is 0.003”:

My recommendation (if you want something better and more permanent/durable than aluminum foil) is 0.002” drilled using the sandwich method between two aluminum plates.
0.003” is the next best alternative for free-hand drilling without sandwiching, only down side is it is slightly thicker.
Hope this helps :slight_smile:

In the future I may try oxidizing the steel to make it black (for less reflections/aberrations) or cold bluing/patina brush-on fluids.
The black oxide coating on steel is less than 0.0001”, much better than paint.
Painting (one coat adds ~0.001”) seems like a much less durable finish and also adds thickness.

7

I am wondering one thing; why?

Does it improve the beam pattern, or output in any way?
Logically, you will loose output, since you block the edges.

Could you explain what you are trying to achieve?

8

A round beam profile instead of a square one. Simple as that.
Having a square beam is something some obviously don’t like. Doesn’t bother me, nor do Aspheric lights.
That aside - excellent idea and implementation Enderman :+1: if square beams bother people this is the solution.

9

So is this like the opposite of cutting the dome off—it would have the same effect as adding a dome to a flat square emitter?

10

No, the die is still square, he wants a round die

11

Aspheric lens flashlights project a square die image.
You don’t use domed LEDs for these lights because the dome increases the apparent die size, which means lower intensity (aka less throw)
Even domed LEDs still give a mostly square die projection, because it’s still a square LED.

This method of cropping the die into a circle produces a perfectly round light source, like an SBT-70 LED does:


Yes you get a bit less lumens, but apparently it increases the intesity too, so more throw.
The main objective is a round spot, like shown above

12

Thanks :slight_smile:
I don’t mind square beams but I know people that do.
I may start doing this to my throwers now that I know it offers intensity improvement, I’m always looking for more throw!

13

Have you tested this at high amps? 15% throw improvement on a CULNM1 is no joke…

14

I’m going to soon :slight_smile:
We will see how high it goes…

15

This was on my long list of things to try- glad someone has given it a go! Thanks, looking forward to the results of further tests!

16

Thanks :smiley:
My 0.003” drilled shim is being glued to a white flat right now, maybe tomorrow night I can do a test to figure out the max intensity cd/mm^2

17

Also, at these sizes/thicknesses, a punch might make cleaner holes.

18

You’re right, A punch would be ideal.
I couldn’t find a 1mm diameter metal punch though.
I’m sure a 3mm diameter one shouldn’t be hard to find if someone wanted to do this to a SBT90/CFT90 which is 3x3mm square

19

All measurements done at 1m using TASI 632A luxmeter

First test:
228 lux = 290 cd/mm^2

Second test using liquid metal:
232 lux = 295 cd/mm^2

Both peaks were at ~6.3A

Seems like a slight decrease from the 300+ I’ve gotten from other LEDs, but I didn’t test this exact one before modifying it so I can’t know for sure.
I also can’t remember if I used this same luxmeter for my last CULNM1 test a few years ago, but this 632A is the most accurate I have.
.
Even though I don’t have an exact relative comparison with before and after, the raw 295cd/mm^2 by itself is already an excellent number.
Assuming an ideal lambertian distribution, we can find the total output lumens is 729 which would theoretically be 928 if the full 1mm x 1mm die area was exposed.
(cd * pi = lumens, more info here https://depts.washington.edu/mictech/optics/me557/Radiometry.pdf)

20

Interesting…
You need to paint the metal shim black correct?