Rebel Tri-star array for aquarium - 75 LED

zachtos

Newly Enlightened
Joined
Nov 15, 2005
Messages
6
Hi, I would like some design suggestions, ideas and links for my new proposed rebel-tri-star array for coral reef aquarium application. (office tank).

I was going to use some of the blank rebel - tri-stars they sell here.
I was thinking of the following on each star to accomplish 20,000K lighting, I would setup a 25 star array consisting of 75 LEDs for starters.

- 1 cool white 145 lumen on each tristar
-1 royal blue 440nm on each tristar
-one blue 470nm rebel on each tristar

I would run them off two PC power supplies connected in series to provide 24VDC.

I would power each strand of 6 LEDs (3.15V Vf each) with LM350's (3V loss) w/ a resistor between Vout and Iadjust to set the current at 800mA each.

It would also put a endor tri lens to increase output and direct the light to the tank.

This is basically a DIY version of the newest solaris but using my own blend of lights and bins. I think this version could easily surpass a 400W Metal halide, but cost is huge. a square foot of lighting will likely be $500 or so in parts.

*My biggest issue is heat dissipation. I would have to mount the rebel starts on a RAM heat sink and actively cool it.

*Mounting heat sinks w/o screw holes is also a concern.


Do you have any good links/suggestions or have seen anything that I could emulate for my new project?

(I've built several aquarium LED arrays in the last 5 years, I'm at revision 4, Just google ' zachtos LED ' and I should pop up right on top.)
 
I have been using triple rebels for a while now, but without lenses and it creates very high PAR. With a 25 degree lens it should be pretty stupid. I recently was able to test 40 degree optics using Cree XR-E LEDs and hit 250W MH PAR levels, even at 24". Dropping to 25 degrees and triple the number of LEDs will crank up the PAR big time. Keep in mind that the royal blue LEDs are producing almost as much PAR per watt as the cool whites (at least with the Crees).

I'd suggest looking at a much larger heatsink than a RAM heatsink. Not nearly enough mass to deal with ~8W of LEDs. The fan would have to be screaming to keep up.

Also, look into a more efficient driver than the LM350. Simple buck regulators can be made for cheap. Chips like the ST Micro STCS1 is simple to use parts wise, but it's surface mount, and will make it a little harder to use. It's a much more efficient driver though.
 
I have yet to find a cheap driver. If I were to scale up the array, I want to keep costs down and keep the buck pucks at bay unless I can do each one for under $5 in materials? I'm not super concerned about a small amount of energy wasted considering how much we waste currently (think 1500kwh per month on my bill already from the tank alone)

I dont know too much about the cree's still. I prefer the phillips brand and have been considering buying a hot air rework station (hot air soldering gun for mounting surface mount chips easily)
 
Look at something like the ST Micro STCS1 chip. You can make drivers for about $4 each, capable of 1.5A at a max of 40v input, with minimal support components. The only thing you have to watch is the delta between the total vf and the source voltage. Bigger delta gets the chip hotter, but it's an SOIC8P package with a thermal pad on the bottom that you can tie to the ground plane for heat desipation.

There are other drivers like the LM3504 and LM3501 which can be made pretty cheap.

Even if you want to stay away from "LED driver" chips, you can look at more advanced linear transistor based current limiters. They produce a lot less heat than forcing a voltage regulator into current mode. There is a good example of this at Instructables. I used this on my first array after finding out how much heat an LM317 current limiter will generate. They work pretty well and can be built for about $2 if you buy for a bunch of drivers.
 
I would use a single 19 volt 180 watt power brick such as this one. No fan. sealed. switches off instantly in case of short circuit.

I would run three boost drivers off of it, one for each string of 25 LEDs, although up to 30 per string would be possible at max 100 volts, and 2 watts per LED. Linear, National, and TI produce controllers that can be used to similar effect.

What sort of control do you need? Does it need to be programmable or would it be enough just to have three pots, one for each color?
 
I'll look into the controllers when I have time, I've gotten very busy lately with other projects and this is on the backburner again.

I would like to be able to adjust the color temperatures by changing pots, but I doubt that will be an easy configuration to solder.
 
It's not too hard. Measure Explorer on ebay has SOIC8 adapters that you can just solder headers to and drop it on a bread board. The only thing that makes the STCS1 a little more difficult to work with is that it will only take a pwn input to dim. This can easily be remedied by making a pwm output from a 555 timer.
 
What is the goal here?
I am concerned that without deep red (long wavelength), this light will be unable to keep plants healthy, at least not without overpowering it with more wattage than it needs. Rebel Red cannot provide deep red wavelengths, AFAIK only LEDEngin makes powerful deep red emitters. There seem to be arrays made of many, many small SMD or T1-3/4 pkgs, at least that's what some of the growmakers who claim to send out "deep red" appear to use (actual presence of DR is something I'm skeptical of), but I've no idea where to find them.

This may be less efficient than metal halide. Well engineered MH is pretty efficient to begin with.

At 800ma these may not have long life (years)- especially not on an Triple Endor Star. IIRC Endor Star had poor heatsinking abilities which is more significant as you attempt higher powers.

Also at higher powers (and temps) the efficiency does take a significant bite, which may compromise the efficiency gain over MH if this was supposed to be a model of long efficiency.
 
Endor Star claims the junction-to-pack of the 3-up is 6.3C/W, but that's for ALL 3 devices which is not the way I'd state it.

Due to superposition of the thermal paths, this makes 18.9C/W per device when loaded 3-up. Now they chose to state that as junction to back, not pad, which is odd but anyhow that means 10C/W (Rebel spec sheet) in the Rebel and 8.9C/W from the board. Plus your sink of course.

When you say 800mA, you don't mean 800mA through each Rebel do you? Efficiency and lumen maintenance drop when overdriven, esp at high temps, most here believe more so than the spec sheet lets on.
 
First post state reef aquarium. Absolutely no need for red here.

LED setups are far more efficient than any other light source for this type of application, when you think about it in slightly different terms. Matching lumen output is not important, as all it will do is damage corals. Where LEDs really shine is in their relatively monochromatic output for colored LEDs, and very peaky output for white LEDs. This allows very high PAR (Photosynthetic Active Radiation) number to be achieved, with realtively low lumen output. An array (any size) of 3W LEDs can produce similar output PAR wise to a 70W MH, with similar penetration characteristics. Reaching 400W MH output levels can easily be reached by the use of fairly narrow optics.

An example I have is a 12" fixture sitting over a 4 gallon tank. It's a 30W array of 4 Q5 cool white, 5 XR-E royal blues, and one 3W UV (not important as to why right now). Whites are driven at 1000mA and RBs at 700mA. PAR levels at 14" were around 135, which is close to the performance of most 70W 14K MH bulbs, for half the wattage. As an additional test, I placed 40 degree Cree optics on the LEDs, and proceeded to test over my 24" tall freshwater tank. 200 PAR on the gravel. This puts PAR output in league with most high end 250W setups. Granted the area illuminated was pretty small in comparison to the size of the tank, but high performance none the less.
 
What formula is determining PAR here? I only recall seeing it described generally, with no method for arriving at a comparative value nor how to work this out for specific plants.

Hmm, I think I see the point on the red, the reds don't penetrate as deeply and, being unavailable to them, the plants have evolved away the ability and/or need to use it?
 
There is no specific formula that I have used to generate these claims. This is all from real world testing. With all the work I have done with LED setups for reef aquarium use, I ended up ponying up for an Apogee PAR meter. I have tested numerous LED and metal halide configurations to come to these conclusions.

There is another reason why we don't use red, beyond it's lack of penetration. There have been studies performed recently that indicate that red wavelengths can actually bleach and harm coral tissue with enough intensity. It's also far less effective as part of the photosynthesis cycle as blue is (430nm) for photosynthetic animals.

And just FYI, incase you didn't know, corals are not plants. :)
 
Neat- did you actually test on coral growth vs MH?
IIRC at least one of the tests on photosynthesis that determined which wavelengths it used was a plant in a bell jar with a sensivitive O2 meter, increasing O2 levels indicating the magnitude of ongoing photosynthesis. Not really applicable to a photosynthetic coral however. Well, you'd need to isolate the zooxanthellae, keep them alive without the coral, seal the container, and use a dissolved O2 meter.
 
Last edited:
My tests on coral are ongoing. So far it looks very promissing as I am seeing good growth and color on all the corals I have in the tank that my latest fixture is over. It is housing a mixture of different types of corals, but the sps (small polyp stoney) corals are the ones of greatest interest. They are notorious for requiring high light, and they seem to be growing nicely. Tank parameters and trace elements in the water are just as important as the amount of light though.

Ideally, I would need to set up two identicle tanks with the different lighting setups to get a true idea as to how well the LEDs will ultimately work. A larger system than my little 4 gallon tank would be preferable.
 
My tests on coral are ongoing. So far it looks very promissing as I am seeing good growth and color on all the corals I have in the tank that my latest fixture is over. It is housing a mixture of different types of corals, but the sps (small polyp stoney) corals are the ones of greatest interest. They are notorious for requiring high light, and they seem to be growing nicely. Tank parameters and trace elements in the water are just as important as the amount of light though.

Ideally, I would need to set up two identicle tanks with the different lighting setups to get a true idea as to how well the LEDs will ultimately work. A larger system than my little 4 gallon tank would be preferable.

Even better might be a divider in one tank, one side getting MH the other LED. That way temp, dissolved gases and nutrients, etc are guaranteed consistent.
 
Thats a definite possibility too. Don't really have the time, money and space to do it right now though. I'll just have to keep an eye on the tanks I have now running LEDs and see how well they do by themselves.
 
Top