PDA

View Full Version : nstiesi's garage engineered complete DIY fixtures


nstiesi
08/17/2012, 06:51 AM
I have been posting all this in my local forum, so here is a copy and paste so far:

Sorry, not done yet, but I've read so many of these threads here lately that I wanted to throw my hat in. Now, I realized rather late in the game that I didn't bother documenting the build that much until now.

It has been a slow, over engineered garage build. The mechanical engineer in me came out, and though I work for an LED light manufacturer with a full machine shop, I wanted to challenge myself to make something without using my professional resources (mainly because it is hard to get machine time in between production), and I hate asking for favors. In the end, I want to remove my canopy all together, so I wanted something that would look attractive hanging as a pendant.

As for the LEDs, after MUCH consideration, I choose against the new trend of the multichips. The industry I work in uses multichips exclusively for high power lighting, and after my first few months here, I was very excited about the idea of using multichips, and was (and still am, to a point), convinced that they are the future of aquarium lighting. We use the most powerful and efficient multichips in the world, but they proved hard to come by, lacking optic choices, and arent currently available in correct spectra.

So, why did I choose against the "aquarium" multichips:

1) China......nuff said. I am skeptical as to their quality. The lack of useful detail on their spec sheet alone was enough to give me pause. I found the driving current data, thermal limits, and lack of spectral definition worrysome.

2) Lack of optic choices

3) Concentration of heat over small area. Wanting to get heat away from the tank, build something attractive, and build something WITHOUT FANS (that was a big driving factor in my build, my tank is kinda loud). 3w arrays spread over and area will make more efficient use of the heat sink, which I will get to in a moment

4) Spectral control: with the 3w arrays, I was able to choose 6 different colors, and control them all individually. This is give me almost infinite, analog control over my color distribution

5.) Efficiency: I ran the numbers, and those particular multichips just cant touch cree yet. (our multichips at work are BETTER than cree, but again, too warm of color). Less efficiency means more energy goes to heat than light, and heat is the enemy of LEDs, and tanks w/o chillers.

So, what did I come up with? I have two fixtures, with 24 emitters each. The colors I chose were Cyan, Violet, Blue, Royal Blue, Cool White, and Neutral White. The ratio is 3:4:3:6:4:4. The panels are linkable, which means I can use either 1 or both connected in series with eachother, no alterations to the power supply or wiring will be required.

The heat sink is a flat, 1/8" aluminum plate, no fins. There are no fins because there are no fans, and without forced air convection, the increase in surface area of the heat sink due to the fins is largely negated by the stagnant air they trap. Air is an EXCELLENT thermal insulator. Fans are loud, require more power, and fail often. The whole unit is painted, which takes the emissivity of aluminum from 5% in its raw form, to 95%, which means this unit will rely heavily on radiation off the back surface for cooling.

For optics I chose 80 degree reflectors, eschewing the rapid led lens. Since I will have a splash shield, I'm already losing light once by traveling through an optic media. A reflector should catch just as much peripheral light and focus it down as a lens, but without the extra loss. Plus, the reflectors are much smaller than the rapid lenses, which allowed me to keep the overall height of the fixture to 2"

Well, sorry for the novel, its alot of words for just a few photos, but I will be sure to take more from here onward, especially once they are actually shining. I'm praying that I like the results. For years with halides, I have never been truly happy with their color.

https://lh4.googleusercontent.com/-MvlvGU6ZgXo/T9cusPi8MRI/AAAAAAAABM8/vOh4Lp_mxgE/s640/2012-06-11%252020.01.27.jpg

https://lh6.googleusercontent.com/-ZItCP7J-22s/T9cuuczdF-I/AAAAAAAABNE/COCQN_nusZI/s512/2012-06-11%252020.01.15.jpg

https://lh5.googleusercontent.com/-Kshp0Uz1fC8/T9cuxbnEDDI/AAAAAAAABNM/toBmbhmyTig/s512/2012-06-11%252020.01.02.jpg

https://lh4.googleusercontent.com/-_jwNkDNRXK4/UBFBMvFU8nI/AAAAAAAABOQ/sSc-grUe9Ek/s640/2012-07-23%252020.20.29.jpg

https://lh3.googleusercontent.com/-aTHC6D46E1o/UBFBZSwUo8I/AAAAAAAABOY/13t3yN7qeIo/s640/2012-07-23%252020.20.18.jpg

nstiesi
08/17/2012, 06:52 AM
Thanks to the help of my lovely wife, I got all the leds glued down, and the optics glued on tonight. This weekend its on to wiring.

https://lh3.googleusercontent.com/-6xDtIJa5Kyk/UBIP80ahM5I/AAAAAAAABOs/_jwEbCN4dcY/s640/2012-07-26%252023.42.06.jpg

Close up of the reflector (these things are TINY):
https://lh3.googleusercontent.com/-edezPMP1--E/UBIP1xCVrEI/AAAAAAAABOk/6Zj5KO_opxM/s640/2012-07-26%252023.42.31.jpg

nstiesi
08/17/2012, 06:53 AM
Some progress this weekend, but nothing exciting, just alot of wire stripping, soldering, and stabbing my thumb with a xacto knife. Looks like a rats nest, and it is, but I have a closeout plate that will cover it all. Still need to do short/ground testing, but otherwise the fixture wiring is done. Couldn't get eye bolts shorter than 2-1/2 inches, so I have to cut 8 of those down. Then the next step is to wire the plugs, and create a power/control box, and I can FINALLY see some light, after 2+ months.

https://lh4.googleusercontent.com/-D3cli_OioZ0/UBYKCIMJRnI/AAAAAAAABO8/6SxE3bdrHgo/s576/2012-07-29%252013.12.39.jpg

nstiesi
08/17/2012, 06:53 AM
Finished the mechanical on one panel, tested it for shorts, lit up each string, and zipped it up. Just need to find some more stainless eyebolts and do the same to panel #2; then its on to control.

https://lh6.googleusercontent.com/-TIKuO0RnZ0w/UBif_uyPE1I/AAAAAAAABPM/-1GKKsOka3E/s640/2012-07-31%252023.16.30.jpg

https://lh4.googleusercontent.com/-jPxRqw1eCUw/UBigAckKWCI/AAAAAAAABPU/BCyErNtbYG0/s640/2012-07-31%252023.17.02.jpg

nstiesi
08/17/2012, 06:54 AM
Made some progress this weekend, and tonight.....this 2 month odyssey is almost complete.

Friday I built a small control box to house 6 potentiometers that will handle manual dimming (which is the plan for the time being....)

https://lh6.googleusercontent.com/-GFeTSrgStv4/UCB9OxLuaSI/AAAAAAAABP8/Xc_koS2QCJ0/s640/2012-08-03%252021.40.47.jpg

Next I began laying out the main power and driver box:

https://lh3.googleusercontent.com/-UIzvYmutnFs/UCB4_-j_amI/AAAAAAAABPs/swp_-zxclHk/s640/2012-08-06%252022.07.24.jpg

This week I will be building an outer housing box to mount these plates in.

nstiesi
08/17/2012, 06:55 AM
So, I am about 99% done with this project, but I hit a nice snag last night.......

First, I got the box together to house the electronics earlier in the week:

https://lh6.googleusercontent.com/-OcxT-gbsa98/UCT4-rsT_wI/AAAAAAAABQU/NbuvFYi_pCw/s640/2012-08-07%252019.24.39.jpg

Got it drilled and painted, and bent up a cover for it last night. This is ALMOST officially the last part to fabricate, but that turns out not to be the case:

https://lh6.googleusercontent.com/-ubQylnUarnQ/UCT4kFXVg5I/AAAAAAAABQM/xNKdE0pGk6Y/s640/2012-08-09%252017.56.09.jpg

On to the snag....when I finally got everything plugged in, the unit wouldnt dim. Come to find out at least one of the pots blew up, and I was too tired last night to really figure out what the hell happened.

But, I was super anxious to see what it looks like over the tank, even though I could't dial in color, so I plugged her in and just held it up. Those of you prognosticating "disco ball" were right, and I figured that might be the case, so I was prepared. The solution to disco ball can be really simple: I got a sheet of that clear-ish, faceted diffuser material that you see in 2'x4' fluorescent lights that you see in drop ceilings. It worked almost TOO well, as it took most of the shimmer away completely.

And yes, I know some will be quick to point out the loss of par through a diffuser, but many of these fixtures tend to be overbuilt, and run dimmed all the time. The consequence could be a simple as driving it a little harder. Only time will tell.

I'm really hoping to get my crap sorted this weekend and finally put this one to bed.

nstiesi
08/17/2012, 06:55 AM
Final layout of the power unit with all wiring complete and strain reliefs installed:

https://lh3.googleusercontent.com/-gRfybgScvI8/UCb-oI5kOdI/AAAAAAAABQs/RLrJQpi7s78/s576/2012-08-11%252018.57.08.jpg

Cover painted and secured with sheetmetal screws. She is ready to mount to a wall:

https://lh6.googleusercontent.com/-1rYeLpb5UJU/UCb-soMdOVI/AAAAAAAABQ0/UNIJdU-NHOY/s576/2012-08-11%252019.03.57.jpg

And for the first time, both panels blazing:

https://lh6.googleusercontent.com/-81Y6lli4Z1g/UCb-eLMDI9I/AAAAAAAABQk/4XqjE8fkFJ4/s640/2012-08-11%252020.20.16.jpg

I wanted to get some pictures of the individual colors, but the camera was not cooperating. I had one unit over my brothers tank last night, and I must say it is fun to cycle through all the colors and be able to tweak it for the exact color you want. Can't wait until it is permanently installed. Some more observations:

1.) Of course, diffusion will be necessary to blend the colors, but thats pretty standard these days. My next mission is to find just the right amount of diffusion

2.) I haven't had it on for a lengthy amount of time, but they seem to run extremely cool so far. All along I have been planning a proper thermal test of a sample of emitters, so stay tuned for that. Also need to make sure the power supplies stay manageably cool. Again no fans.....fans are the enemy.

3.) Even at my low ratio of white to other colors 1:3, they can STILL overpower the rest of the unit, and will probably be turned down under normal operation.

4.) The phillips rebel cyan leds are greener than I expected based on their published wavelength. I need to further evaluate just what effect they have on coral coloration, but I kinda see them turned down most of the time as well....possibly even replaced, if I get brave enough to try and remove them (they are epoxied!)

Anyway, all that is left is to replace the one pot that I burnt out, and hang these suckers!

nstiesi
08/17/2012, 06:56 AM
and this is todays post:

First piece of equipment permanently installed in its new home:

https://lh6.googleusercontent.com/-vEusiKcvTFw/UC48NdTCK_I/AAAAAAAABRE/d5FKfEYoW7Q/s512/2012-08-16%252021.27.09.jpg

Went out last night and bought more eye bolts, chain, and quick disconnects for hanging. For now, I am just going to suspend it with short chains in my canopy for the time being. My end goal was to build a new, completely floating canopy with a worm-gear winch and pulleys to raise and lower it at will. Just wasn't in the budget right now, and I just want to get this puppy burning since my halide lamps are about 18 months old at this point, haha!

nstiesi
08/17/2012, 06:57 AM
Also, earlier this week I took the unit to work and did a full thermal test on a few leds, this is the corresponding post:

"Well, at the risk of being completely anal, I decided to take the light in to work and do some thermal testing. I monitored a sample emitter what I considered to be the 3 worst case colors based both on the thermal resistance rating of the chip, and the expected dim level under normal operation. My XP-E Blues have the worst thermal resistance, at 10C/W, and should experience the most use. I also monitored an XT-E royal blue, and a XP-G neitral white. thermal saturation occured after about an hour, and I let the test run for an additional 2.

XP-E Blue = 77.5degC of an allowable 150degC MAX
XT-E Royal Blue = 63degC of an allowable 150degC MAX
XP-G Neutral White = 60.6degC of an allowable 150decC MAX
Average heat sink temperature = 45-50degC

I'm very pleased with this result, as I am below 50% of the max rating. I also was able to plug the unit into a kill-a-watt meter, and discovered that at 100% power, she draws 170w....less than ONE of my current halides. At 0% power (remote dim to off), the psu's and arduino eat 7w, so I will probably still have the unit on a timer."

Also ran a quick test on the power supplies, and found the air temp in the box to be 56C, just at the start of the power supplies' derating curve. I will keep and eye on it, but it should be an issue since I'm only loading the power supplies to 75% of their rating under worst case conditions (no dimming), and based on what I have seen so far, some colors will be run below 100%.

Mike31154
08/17/2012, 09:45 AM
Interested in more details about how the LEDs are powered. You mention Arduino & I can see two standard DC power supplies in your box. What voltage are the power supplies & how are you controlling current to the LEDs, other than with the dimmers? That is, are you employing some sort of constant current device?

nstiesi
08/17/2012, 10:05 AM
I have 6 DC-DC constant current drivers in the box below the power supplies. Each drives a single series string of a given color.