Attempting a DIY LED fixture - Part 2. New pics 9/27 (56k)

[malaybiswas]

For housing the lights I am thinking of aluminium cases like DVD player or amplifier enclosures. But I am not able to locate any seller who sells them in cheap. Any pointers?

[malaybiswas]

How would these be for heatsink?

http://www.radioshack.com/product/in...ductId=2102857

[evilc66]

Barely adequate, especially if you are only running at 350mA. When I talk heatsink, I mean HEATSINK. I have used these guys on a number of projects with great success. Sure they are overkill, but it will support whatever size array you could want. Using the heatsink as the base of the fixture, you could skin the outside with acrylic or wood. If you are going to run everything at ~350mA, you could run a heatsink of this size passively.

After looking at your layout pictures (which was difficult), I think you have used the wrong angle in your lens approximation. The Fraen lenses are 25 degrees total output, not 25 degrees half angle. It looks like you have a 50 degree cone in your plans.

It does seem like 15 LEDs might do it. 18 would get a slightly better overlap, but might not be needed. One reason I was saying to stay away from the 25 degree lenses is the very high PAR levels you will hit, even while only running at 350mA. A wider angle lens will lessen it and get you back to a level that is still sensible for a high light setup.

From what I have seen, there is a slightly better selection of wide angle lenses for Cree LEDs over Luxeon. Don't know if you are sold on Luxeon or not, but I know a good place to get Crees for a good price and optics too.

[gmccreedy]

Rather then using a dimmer, is there a way to have this wired so you have independent strings of LED's?

This way, for example purposes, your have a 10x4 setup. The outside 10's would be on their own tied circuit then the inside 10's.

You won't get the "ramping" of light that you are going for, but can certainly get a burst.

[evilc66]

Sure, but it means more drivers and more cost unfortunately.

[malaybiswas]

Originally Posted by evilc66

Sure, but it means more drivers and more cost unfortunately.

Exactly...too many drivers, timers and wiring....don't want to go there. Dawn/dusk effect is nice to have but not necessary. If I can make a dimmer circuit well enough or else not.

[malaybiswas]

Originally Posted by evilc66

Barely adequate, especially if you are only running at 350mA. When I talk heatsink, I mean HEATSINK. I have used these guys on a number of projects with great success. Sure they are overkill, but it will support whatever size array you could want. Using the heatsink as the base of the fixture, you could skin the outside with acrylic or wood. If you are going to run everything at ~350mA, you could run a heatsink of this size passively.

After looking at your layout pictures (which was difficult), I think you have used the wrong angle in your lens approximation. The Fraen lenses are 25 degrees total output, not 25 degrees half angle. It looks like you have a 50 degree cone in your plans.

It does seem like 15 LEDs might do it. 18 would get a slightly better overlap, but might not be needed. One reason I was saying to stay away from the 25 degree lenses is the very high PAR levels you will hit, even while only running at 350mA. A wider angle lens will lessen it and get you back to a level that is still sensible for a high light setup.

From what I have seen, there is a slightly better selection of wide angle lenses for Cree LEDs over Luxeon. Don't know if you are sold on Luxeon or not, but I know a good place to get Crees for a good price and optics too.

Ok. I was considering the radioshack ones cos they are light weight

As far as the enclosure is concerned, I am planning all aluminium box that I am going to build off sheet aluminium (found what I need). Even for the base of the LED it will be a 22 gauge alu sheet so that will also provide some heat sinking. Let me think over those.

The pictures in my plan are 25 degree. All dimensions (radius of 22" high cone base are calculated off 12.5 (1/2 25) degree angle. But I will double check anyways.

Is there any formula to calculate PAR level at a certain distance based on bulb specs (lumen output, wavelenght, voltage, current etc.)? I have a formula but not sure if that is correct.

I am inclined towards luxeon just becos I managed to find out all the specs and accessories to plan. Did not manage to find enough about cree yet. But if you and send me some pointers I would certainly like to compare those before buying anything.

[malaybiswas]

I have no idea why the last 2 images keep hiding in the first post. All 3 images are from the same album in picasa. Anyways, just in case if it happens again, I am putting the links separately as well.

Sorry about that.

[kid creole]

Originally Posted by malaybiswas

I have no idea why the last 2 images keep hiding in the first post. All 3 images are from the same album in picasa. Anyways, just in case if it happens again, I am putting the links separately as well.

Sorry about that.

404 on the links. I had something like this happen. I think you have to click share and the image size before the link is a public one.

[malaybiswas]

Originally Posted by kid creole

404 on the links. I had something like this happen. I think you have to click share and the image size before the link is a public one.

The album is public. In fact the first image is from the same album. God knows what's going on. Anyways, I will upload them here instead.

[Orlando]

I just came across this post on apc... http://www.aquaticplantcentral.com/f...tml#post451612

It looks like he has grasped the LED concept pretty good

[malaybiswas]

Originally Posted by Orlando

I just came across this post on apc... http://www.aquaticplantcentral.com/f...tml#post451612

It looks like he has grasped the LED concept pretty good

Yeah, I have already seen this thread. Besides I found some more useful threads from nan-reef.com (and evil66 has some good insight there as well)

The latest link I found is this one. Have to analyze it and hopefully I would get an idea to calculate PUR & PAR

[evilc66]

I was probably the guy Corey mentioned hitting 600+ PAR. I had an array over an 8" cube tank with 64W of Luxeon Rebel LEDs. Way too powerful for that setup, but was an interesting experiment. I almost had a 24" Solaris I4 beat at all distances. I had better color though.

There is no way to truely approximate PAR. PAR is derived from the spectral output of the lamp and can't be approximated by using lumens/lux. Two different bulbs can have the same color temperature rating, and have the same lumen output, but can have very different PAR output, based on the spectral output. Only good way to get PAR numbers is to measure it.

Pretty crazy spreadsheet that guy has though.

On the enclosure, if you are going to stick to just a 22awg aluminum sheet to mount the LEDs to, adding a few inexpensive heatsinks to the other side will provide a little added security. Running at 350mA, you might get away with it. I'm used to running at 700mA+ and the heat builds up quickly.

[malaybiswas]

Originally Posted by evilc66

I was probably the guy Corey mentioned hitting 600+ PAR. I had an array over an 8" cube tank with 64W of Luxeon Rebel LEDs. Way too powerful for that setup, but was an interesting experiment. I almost had a 24" Solaris I4 beat at all distances. I had better color though.

There is no way to truely approximate PAR. PAR is derived from the spectral output of the lamp and can't be approximated by using lumens/lux. Two different bulbs can have the same color temperature rating, and have the same lumen output, but can have very different PAR output, based on the spectral output. Only good way to get PAR numbers is to measure it.

Pretty crazy spreadsheet that guy has though.

On the enclosure, if you are going to stick to just a 22awg aluminum sheet to mount the LEDs to, adding a few inexpensive heatsinks to the other side will provide a little added security. Running at 350mA, you might get away with it. I'm used to running at 700mA+ and the heat builds up quickly.

I don't want to approximate PAR or PUR. I would like to know the method to calculate it. I am pretty sure there is a formula (or a set of them probably) to calculate it based on light energy properties (wavelength, intensity etc.). Since PUR and PAR are both derived properties (dependent on wavelength at least) without specific mathematical calculation(s) there is no way a PAR meter electrical device can be made.

I know that PAR is dependent on wavelength. However in plain english, PAR is the amount of photons emitted. Apart from the wavelength the light density in a given area should also govern how many photons are hitting the surface. I don't have the right amount of physics knowledge on light energy to establish the fact, but it just seems logical to me. Check out this link. I have not completed reading it, but it seems Irradience has some effect on PAR.

Anyways, the previous link I sent is good study link, but here is the summary link of the same thread.

This may be a over kill for a normal planted tank enthusiast, but hey, there is information out there, so if you are interested...

[malaybiswas]

Originally Posted by evilc66

On the enclosure, if you are going to stick to just a 22awg aluminum sheet to mount the LEDs to, adding a few inexpensive heatsinks to the other side will provide a little added security. Running at 350mA, you might get away with it. I'm used to running at 700mA+ and the heat builds up quickly.

Yes, I think I should be good with that. Heat dissipation depends on the "exposed" metal surface. This is one more link which gives a good approximation and calculation logic to decide on heatsink selection. I rely on it since I saw the same thumb rule in electrical engineering text book in my local library.

Going by it, each of my LEDs forward voltage is 3.99 (max) and forward current 350mA. So total power of 15 LEDs = 15 * 0.35 * 3.99 = 21 watts max.

Total square inch of open metal surface required = 21 * 10 = 210 sq inch.

The mounting surface (alu sheet 22g) is 24" X 15" = 360 sq inch on the top surface (opposite to LED mounting surface). Assuming 50% of the LED mounting surface is exposed, that would be 180 sq inch. So I have 540 sq inch of metal surface. This is already more than enough for the LEDs I have. Any additional heatsink (however small) should work.

Don't get me wrong here, I checked out heatsinkusa.com. But my goal is to keep it simple, cheap and light weight. And its experimental, so I want to try out what scientific calculations suggest. If things don't work, I want to find out why it did not work out before going to next stage to fix it.

So right now, my plan for heatsink is 24" X 15" 22AWG aluminium sheet on which LEDs will be mounted. In the test run, if they still get hot and burn out, I will add heatsinks on the surface but not one big piece, but multiple smaller pieces starting from the center and spreading out towards the periphery. That way I can estimate what is the correct exposed surface required for this configuration (hopefully I will never need to go there ).