Let's Make Robots!

How to? Glowing IR Ball

I need to make a glowing, IR ball.

So here's the deal, I need to make a ball, somewhere between a baseball and softball, that glows with IR light. It needs to be self-contained and solid IR --no 38khz etc. I think I have seen similar items, one LMR'er gave me a link to a IR ball they use for robot soccer. I like the overall design --the PCB "layers" etc. but it would not work exactly for what I need...

I have brainstormed and it seems the requirements are going to dictate the design here. Here they are in list form...

  • Self contained incl. small 7.2V lipo and any volt/current reg needed
  • Openable to access lights and batts
  • Must "glow" overall
  • Must NOT have individual points of light showing <--wicked important for the wii camera
  • "Spread" of the LED's and the distance from the LED's to the inside surface of the ball is very important
  • Opaque enought to diffuse the individual LED's --Transparent enough that the light will have some "distance"
  • Gently "throwable" --or at least "rollable"
  • A lot of LED's would be OK, fewer would be better.
  • How to mount LED's inside? --Curved PCB strips? Layered round PCB's? Solder Legs together like a LED Cube?
  • We're talkin' a lot of LEDs here --series? parallel? current-draw, volt/current reg, resistors <--all have to fit inside
  • Gotta be able to turn of/off but not inadvertantly
  • Need "on" indicator  --can't see IR for crying out loud!
  • Can't have too much of a IR "shadow" where the battery door is


  Open discussion


************** Update 3.13.11  *************************

Man, this is becoming a sorta hard thing to do.... I finished and tested the "layered" beacon with no-so-great results. I tried different ways to diffuse and spread the light with no real effect on the ability of the wii camera to find it. Oddly enough, crumpled Saran Wrap seemed to work the best... Go figure. Whatever I tried (even the Saran Wrap), the main problem was the fact that the LED's were too far apart from each other and apeared to the camera as single points of light or worse, just not seen.

At any rate, here is a bunch of stuff I learned:

  • The LED's I used were way too narrow-beamed. I went for price (Ebay) over the specific specs of the LED's. Wider the better
  • I used 940nm LED's --again, price. I think the 850's are brighter as they are used for night vision cameras.
  • The wii camera (mounted on a pan/tilt) "tracks" much better with a "pack" of LED's shining to it. The Sensor Bar itself has (4) LED's, packed together, at each end. I have also found that 3-5 LED's packed together is just about right. I have a modified flashlight (torch) with (9) LED's and it "overpowers" the wii camera at close range and you get false numbers coming out.
  • The bottom line is that I underestimated the number of LED's needed (and thus the battery drain), the correct specs for the LED's used, and how close they need to be to each other to produce a "continous light"

I have moved on to version 2 or 3 or so...

This is simple, I can't have what I want and I am going to have to compromise. It seems far, far more pratical to go with a 1/2 sphere. I started with the obvious... IR flashlight and the LED's best friend, a ping-pong ball...


Worked Ok, but didn't do anything to spread the beam. It did solve the problem of "overpowering" the wiicamera when in close range, though.  Still very uni-directional.

--Moved onto polygons...

I have a little hope for the one on the right (the shallow one). With a little bending of the LED legs, I should be able to achieve a pretty good "arch" of light, while still maintaining "packs" of LEDs. If you look in the center of each ear, you will see an extra pad --a little larger than the ones for the LED's. My plan is to run a trace around the outside edge of each panel. This trace will serve as not only the ground plane, but will also allow me to solder one panel to the next. The resistor for each panel will be soldered to that extra pin and continue on the inside to the center. Here, it will meet all the other resistors from all the other panels to then be connected to power.

We'll see how these go, I guess...


************** Update 3.11.11  ************************

Well, I tried them all...

I first tried a wire-frame:

It was OK, I could keep it small but trying to get it to point in all directions was going to be a mess. Actually, it was going to be a big fat short and a lipo on fire. No good.

Moved on to PCB's. I took some very thin, light stuff and started cutting strips. My idea was to bend strips into 1/2-circles, join them into full circles (orbits) and then simply have 6 or 8 "orbits" crossing at the poles. A couple things here. --First, PCB material does not melt or soften with heat. You would think it would, but it is a resin, not a plastic. I knew that, but it didn't sink in. To cut to the chase, the fiberglass that the boards are made from does not soften and flex, instead it does indeed catch fire.

Too big, too clunky and well, I set fire to it while trying to make the bend. No good.

Figured at this point that the robot soccer people had probably done all the engineering needed for their ball --and theirs works, so I stole the design. Layers. That's the way to go. Now I gotta figure out how to power 48 LEDs without a regulator. A bunch of math later, I basically made batches of (4) with each batch having it's own resistor. The batches (12 of them) are then wired in parallel. This was real give/take --I originally was planning on 5 layers of 16 LED's each, plus the ends. After doing all the current-draw math, figured out how bright I could push them vs. how many I could have vs. run time with 900mA pack. At the end of the day, 4 layers with 8 leds each, plus the ends is what I came up with.

I am sorry about the image quality --Kari is off on holiday with the good camera --Had to use my mobile...

Start of the Frame --The tubes are support as well as carry current.


Finished with batts in place. The top layer unscrews from the (3) posts to be able to remove the battery. The lowest layer is soldered solid.

I dunno 'bout that switch. I may do a reed switch or maybe just a little jumper, I dunno. I very great guy/friend/fellow LMR'er, has offered me a clear ball so I think I got that problem licked.

Ta Da!


All that is left is to suspend this guy in the ball, figure out the switch and also figure out how to frost the inside of the ball --and how much. Udda dan dat, I t'ink it's done!! Woo Hoo !

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

I'd go with the solid glop method.
Get one of these guy from craft store or online... you'll need 83 cents though...

Sand the IR LEDs with rough sand paper to diffuse the light.  Sand interior of ball, and possibly exterior to further diffuse light.  
Solder each LED with a resistor - cube style.  Mix up 2 part epoxy, clear silicone, or whatever's handy (Test first to determine it is IR translucent) - and fill a hemisphere full of the crap - Shove the LED matrix into the glop. Push small plug in center for battery space. Bake if necessary (it not you).  Attach battery and use fastening system of globe (looks like a screw)

Mary's your Aunt.

"Need "on" indicator  --can't see IR for crying out loud!" <- Ask Walter silly, he can see IR


That's funny -- I was looking for my stock of casting resin just today, thinking about a solid clear ball! Yeah, I looked into those christmas decorations. Good, cheap and too small... Damn.

I do have to move from the ball for a sec though... I need someone to check my math here. I did some quick calculations and it seems that we are talking about a huge amount of power here...

  • 1.5 to 1.6 FV
  • 60mA draw
  • somewhere in the neighborhood of 60-90 LED's (on constantly)

My math is saying something in the 4 to 5 amp range?!?! Well, that ain't gunna happen. Even if my lipo can do that, we are talking about 18 min of run time.

Looks like I gotta get creative with "banks" of LEDS in series and these "banks" then in parallel. Just gotta get math-creative, I guess in terms of limiting total current draw --at the same time not having to use physically huge resisors, or a ton of resistors, while also trying to make this thing as bright as I can.

Figure in the fact that these are 30 degree LED's (kinda narrow), I can't get past the fact that I am simply going to need a lot to be sure I have 100% coverage of the sphere.

It's the fact that they are all on at the same time that is killing me. And no, I can't PWM them --what if Walter is looking during a low cycle of the PWM?




What if you only power a fraction of the total LEDs at a time so there is always a trackable number of LEDs on but not all of them?

I have a few of those acrylic spheres for a blinky project in the pipeline. I found some that were 4" diameter. That's somewhere between a baseball (just <3") and a softball (bigger lol).

You gotta link for the one's you bought? I'm gettin' nuthin'


100mm acrylic ornament vs. 4" acrylic orname was the key. An eBay example...


If those look appropriate let me know and I'll dig around and send you the company I got mine from. It was the least appaling priced of the craft stores/ebay/amazon. They are cheap but the shipping for all the air inside is a killer.

Our local party store had these globes that come in two halves and screw together. I think they were 5-6 inches in diameter.

Unfortunately one half was clear and the other was colored.

If I pack the led's tighter to the center, they will have a further distance to be able to "spread" and thus, I will need less to have full coverage. Maybe I can get the total LED count down to 48.

Still need help on the math and current limiting options...

Concerning the "Walter may only see one blob at the time" issue:

Buy LED's with known beam angle, say 60°. Arrange them so that only one is visible from any angle, say on the corners of a tetrahedron. Make sure to fill up the flat surfaces to resemble something round.


Not sure about the angle. Forget about the details. Those beam angles are approximats anyway. Usually indicating the angle between half-values of the luminocity curve.