Cheap, DIY LED fixtures

[blasterman]

I can't build enough of these things.

After getting tired of the erratic reliability of commercial LED heads I came up with a DIY approach that's been working really, really good. Might not look as 'high-tech' as a state of the art, triple LED commercial Cree head, but my solution is cheaper, and I get to choose all the components. Not to mention I can stick it anywhere there's a metal surface.

I use 3/4 x 3" steel pipe you can get any any hardware store for about $1.25, and then glue the Star to the side of the pipe with thermal paste. Wires from the soldered star contacts get threaded through the pipe - I suppose you could stuff a small current regulator in there if you wanted. I prefer 12volt external power supplies and a resistor (usually 5-ohm)matched to a series of 3 heads.

I then glue the optic of my choosing with a dab of epoxy to the top of the star. My preference is for reflectors vs acrylic optics because there's no side 'splash' of light with reflectors, but both work. Acrylic optics are certainly available in more flexible beam patterns, and the wider angle acrylic lenses usually have enough diffusion to negate any artifacts associated with emitters; rings, color bands, etc.

Tape off the the top of the optic, and spray paint the rig the color of your choosing. Next, I use a quarter size neodymium magnet to affix the head to a ceiling...usually ceiling grid. These type of magnets 'grip' the steel pipe strong enough that it ain't coming down on it's own, but it's still lose enough to swivel and rotate and be pointed where ever you wish. There you have it - a LED head that can be stuck about anywhere for way less than $20 that's has an excellent, albeit 'ghetto' heat sink. This has worked with Cree, Luxeon, K2's, bare emitters....although I haven't tried Rebels yet.

First, here's a shot of one of my first prototypes. Note it's pretty beat up rolling around in my trunk, but it still works fine. Use a couple coats of glossy paint, and they darn near vanish against ceiling grid.

Next is a shot of three of the heads (blue 3-Watt Pro Lights) mounted on a 14' ceiling of a club I installed them in. The beam shots are exagerated by a nearby Hazer, but you get the idea.

 

[R33E8]

that second picture looks sexy :naughty:

 

[SafetyBob]

Oh, I can't believe I am the first one to say excellent job. That is a tremedous job of using cheap, commonly available material.

I love the way the fog in the club makes those blue lights look.

Now I wish I had a loft appartment in an old warehouse somewhere, your idea would be killer in a situation like that!!

Great job

Bob E.

 

[blasterman]

A lot of people have drop ceilings in their basements, etc., and these things work great for snapping on the metal grid between tiles.

I can't see why they woulnd't work outside either - as long as you used acrylic optics, which protect the emitter. Who cares if it eventually rusts. It's 12volt - big deal.

 

[I'mIn]

Excellent my man! What currents have you been successful at running with this type of setup? It kind of looks like the roundness of the pipe would seriously hurt it's heatsinking capabilities, but it must be good enough for at least 350mA. Have you experimented with different sized pipes, what were some of your conclusions if so?
Thank you for posting your work.

 

[brickbat]

Very clever design. I especially l like your idea of the magnetic gimbal - simple, cheap, but effective.

 

[blasterman]

It kind of looks like the roundness of the pipe would seriously hurt it's heatsinking capabilities

I thought that originally, but even when I drive them really hard the bottom of the star never gets hotter than the pipe. I drove a K2 at over an amp for about an hour, and the pipe got TOO hot to hold, but nothing failed. Anything less and the pipe/star doesn't seem to get beyond barely warm. At 350ma it stays at room temp.

This is the reason I stick with 3/4 and not 1/2 though. The bigger diameter pipe makes better contact. Who knows how much metal contact the star actually needs, but so far 3/4 works just fine. Obviously the surface area of the pipe, inside and out, is more than efficient enough. Funny how that K2 was able to heat it up though :twothumbs

Yeah, we can't wait to install some more. I'm thinking of using amber for contrast. Shoot them down walls, etc. The club is at capacity as I type this, and it looks frikken cool with people walking through the beams

 

[I'mIn]

Orange is blues complementary color so amber would be two double D's!

 

[Nitroz]

Excellent work!

I might add some blue lighting to my kids gaming room.

 

[I'mIn]

So I ran 3 LEDs on 1"X8" steel pipe($4 @ HD.) The results are great. I'm pushing them at 800mA and the pipe got warm after an hour use. Definately not hot!
It's kind of funny so I'll tell you guys how I got them on there. The LEDs were wired before attaching them to the pipe and I came to find that it was impossible to put them all on there flat and aligned with each other long enough for the paste to harden. The solution?, superglue :naughty:. Worked great too, no problem with heat transfer, LEDs have got no hotter than the pipes.
Lovin it

 

[blasterman]

High Five dude!

LEDs have got no hotter than the pipes.

Yep - ain't that cool? I'm using thermal paste, but the more I think about it the more I don't think it's really necessary. It's having a close, metal to metal bond that counts for efficient heat xfer - not fancy epoxy with a couple milligrams of silver in it.

Super glue forms a very tight, close, strong metal to metal bond. Plus, it sets fast so you don't have to deal with stars sliding around for an hour until epoxy hardens. Can't wait to try super glue with bare emitters.

 

[I'mIn]

You were spot on about those pipes. I spent over an hour looking around at different heatsinking options, but nothing seemed to be worth it in comparison. Really they even had 10' pipe for $25, that would be equivalent to like 60cents per 1"X3" pipe. So for $25 for the pipe and $1 for a four pack of superglue I saw at Wally World today, one could allow heat dissapation for about 40 LED's. Not bad at all boys.
One more thing I should note is that these were luxeon rebels, I reflowed them onto stars and some square blanks. The square blanks were extremely small and difficult to maneuver, but after installing them to the 1" pipe they apparantley still transfered heat just as well as the stars:thinking:. This is even though they had a fraction of material to bond with than the stars did. Hence it may not be as important as I thought to have lots of contact between the slug and the heatsink. This I think helps to support Blastermans theory that even with a curved surface, a 3/4" pipe will provide the necessary contact area needed to quickly transfer the heat.

 

[bretti_kivi]

my wife would flip if I did this, but I like the idea; i hadn't thought of using reflectors (I don't know why, either ?!) and some simple metal heatsinking. Keep it simple and it can't break!

Thought about using RGBs on the ceiling?

thanks

Bret

 

[I'mIn]

Not really, but here's a good example of one that's already designed for you to develop some ideas:
http://www.instructables.com/id/RGB-Color-Controllable-High-Power-LED-Room-+-Spot-/
What I like about it, is it not only tells you the specs, but even has the part numbers listed for each component.

 

[sandos]

Very nice. With this being so cheap I might just be able to convince my better half that I actually should be allowed to "waste" money on LED lighting

We're moving in two months' time and Ill have my own computer/hobby-room, so Ill just experiment in there and see what she thinks of it.

 

[purduephotog]

So I ran 3 LEDs on 1"X8" steel pipe($4 @ HD.) The results are great. I'm pushing them at 800mA and the pipe got warm after an hour use. Definately not hot!
It's kind of funny so I'll tell you guys how I got them on there. The LEDs were wired before attaching them to the pipe and I came to find that it was impossible to put them all on there flat and aligned with each other long enough for the paste to harden. The solution?, superglue :naughty:. Worked great too, no problem with heat transfer, LEDs have got no hotter than the pipes.
Lovin it

Steel doesn't get hot. It is a very poor conductor of heat- so it would only feel warm when your LED die might be cooking.

Try the same experiment with a copper pipe. If the copper pipe feels hot- that's a good thing- that means that the die is radiating the heat away- otherwise your LED is keeping all that bad heat for itself.

Think of the heat as a fluid- if the pipe isn't letting it drip into space then the LED is simply making more and more.

 

[blasterman]

Steel doesn't get hot.

Second I think you've said this, and it isn't true. Steel (depending on blend) has no more than 15-20% the conductive efficiency of Alu or copper. This means if absolute heat spreading in a small packade is desired, steel is a poor choice of material because thermal transfer is too slow. If you have a big heat sink made of steel and aren't driving it with a MC-E or P7 then it probably won't matter. A 3" length of 3/4 inch steel pipe is perfectly sufficient for these things - if not over-kill.

I recently made a dozen more of the same fixtures using 1" copper pipe I cut myself to save cost, and while the copper warms up faster when the emitter is driven hard, the steel mounted emitters eventually catch up and assume the same temp as the copper. The heat has to go somewhere.

Among the more unusual things I've tested for heat-sinks is when I mounted K2 emitters to the ends of 1/4 x 3" inch steel bolts. Driven at an amp the bolts became too hot to hold within 10 seconds.

Somewhere in my old computer lab I have passive heatsinks designed by Intel for their P4 Xeons, and the bases are *stainless steel*.

 

[Mr_Light]

One simple test I do for effective heatsinking is to touch the STAR base of the LED to see if it is close to the same temperature as the surrounding heatsink. If so, then I figure it has to be doing it's job. If the star base is too hot to touch, you have problems beyond your burned finger...

 

[purduephotog]

Second I think you've said this, and it isn't true. Steel (depending on blend) has no more than 15-20% the conductive efficiency of Alu or copper.
(snip)
Among the more unusual things I've tested for heat-sinks is when I mounted K2 emitters to the ends of 1/4 x 3" inch steel bolts. Driven at an amp the bolts became too hot to hold within 10 seconds.

Somewhere in my old computer lab I have passive heatsinks designed by Intel for their P4 Xeons, and the bases are *stainless steel*.

You've made my point for me. The poor heat conduction of steel makes it a bad choice. I never said it wouldn't reach a steady state- but in order for it to do so the LED STAR must be at a higher temperature for a longer duration before that steel gets to the same ambient temp. Steel also does not radiate energy into air as fast as other materials.

If the STAR produces 1 watt waste heat then the heat sink must radiate that. Heat sinks that are 10C hotter than the surrounding ambient will radiate heat at a much faster rate than sinks that are not- in fact, designing heat sinks to be at minimum 10C hotter than the surround will ensure effective heat dissipation.

If I may point you to this article, slightly dated, for P4 heat sinks:
http://www.overclockers.com/oldsite/articles373/p4sum.asp

If what you say is true than any steal chunk would sit on the die and be perfectly acceptable as a heat sink. Unfortunately what happens is that the die temp continues to rise until, in the P4 Xenon state, they'd go into throttling. Those sinks are also rather large and Intel was known for their high speed fans- which generate more air than the passive systems and effectively 'increase' the heat transfer coefficients.

If you want to test take a K-type thermocouple and epoxy it (black only- other colors will radiate IR energy and effectively leak heat) to the side of the star nearest the die- you'll need bare aluminum (resist will insulate). You can also try a non-contact IR probe but I've found them difficult to aim at the distances involved.

Many new LED products are coming with maximum rated die temps (annoying) while manufacturers are stating maximum board temps (pleasure). It's easy to measure a board temp as that is just a small thermistor epoxied in place and hooked up to a fluke (or a temperature probe). They're relatively cheap in quantity and have fast responses.

Measuring die temp is almost impossible unless you drill the star and bury it in the material (let me know if that works- it was hard enough for heat sinks). Thus you can only approximate how hot that die is getting.

Anyway- I've gone on too long (eating lunch at work). My background is chem eng and I've done work with heat exchangers. Remember that it's always easier to model a person as a sphere- it makes the math easier.