does this sound reasonable to build?

[bauerbach]

Im looking to light a fish tank... maybe not entirely, but supplement some metal halide lighting.

anyway, my question is, Im powering the array with a PC PSU. it has a few 18amp rails, figure that should be sufficient at this point.

so I have only played with luxeon I.

Im considering the endor rebel star, the 435lumen variety to be precise. they say they require 9.45volts, the PSU can provide 10v(among other options). Id like to power these things up. 700mA or so, I have heatsinks and active air movement for it. but will that be sufficient? how much heat am I looking at?

and if I do hit them with 10 volts, will they're life be drasticly shortened? would something like a 700mA buckpuck be a wise choice to put in the circuit? if so, how many series can those handle? if this goes according to plan, I could be looking at ALOT of stars, Im looking to get 10-20,000 lumens from this array, so... like 40 series of 3 in parralel. and possibly another 10 in royal blue.

so uhhhh yeah, am I crazy, am I trying to do the impossible? Im not even sure how to go about calculating the current draw of 150 rebels on 50 stars at 700mA each. the PSU might be SOL to try and power all that no matter what I do.

anyone want to smack me across the face and tell me no, or tell me yes, my plan should work?

btw, heres my first test unit, with what I had laying around. unfortunatly I didnt know most of them were cyan instead of blue, and 1 amber got thrown into the mix. just so you dont think Im totally full of it, I am serious about making this possible, its going onto a 240g display salt water reef tank, currently lit with 2 250w MH. it needs more though, another 2 MH unless this LED idea pulls though. this unit is thrown together in an afternoon with aluminum and other bits I had laying around just to get an idea.

 

[R33E8]

Have you thought about using Seoul Semi P7's? You should get almost double the amount of light at less cost..

http://www.dealextreme.com/details.dx/sku.12721

About 20 of these should fulfill lumen your requirements and they are the same size as the endorstars...

BTW, 150 rebels at 700mA would be 105 amps while 20 P7's would add up to 54 amps..

40 rebels in series would be 28 amps

Have you thought of building your own power supply?


You can also get the Cree MC-E's which would be even cheaper and from a more reliable source..
http://cutter.com.au/products.php?cat=47

 

[lazarus2405]

would something like a 700mA buckpuck be a wise choice to put in the circuit?

Yes, I think so! With a 12v input, you might be better using a boost driver than a buck, like the MaxFlex (http://www.taskled.com/maxflex.html). With one of those, you should be able to drive 6 LEDs in series at up to 1200mA. If that isn't enough current, like for a P7, you should be able to put the outputs of two drivers in parallel.

so uhhhh yeah, am I crazy, am I trying to do the impossible?

No sir! Dream the dream and build it... Just don't complain to us about the cost.

 

[LukeA]

Have you thought about using Seoul Semi P7's? You should get almost double the amount of light at less cost..

http://www.dealextreme.com/details.dx/sku.12721

About 20 of these should fulfill lumen your requirements and they are the same size as the endorstars...

BTW, 150 rebels at 700mA would be 105 amps while 20 P7's would add up to 54 amps..

40 rebels in series would be 28 amps

Have you thought of building your own power supply?


You can also get the Cree MC-E's which would be even cheaper and from a more reliable source..
http://cutter.com.au/products.php?cat=47

Those Rebels are lower bin. They won't be as efficient as other LEDs.

No P7 bin would allow 20 P7s to equal the output of 150 RB80s. It's more like 27.

150 RB80s is 336W. 27 P7s running at spec is 265W. 150 RB100s (if you can afford them) would use 268W to match the others in lumens.

However, The P7 is not well designed. Parallel is bad for LEDs.

With this kind of wattage, you may be better served by a bridge rectifier running from 110VAC, if the 120Hz flicker isn't too bothersome.

 

[TigerhawkT3]

Sorry for the raincloud, but it'd be a lot easier, probably cheaper, and just as efficient to put up another 250W MH HID fixture or two.

That said...

You could get a rectifier to turn 110VAC to ~167VDC, filter the output with capacitors, and drive around 45 LEDs in series with an LM138 and a 2Ω sense resistor. If you do choose to work with mains power, be very, very careful, please.

 

[VanIsleDSM]

LM138 only has an output voltage up to 32V. Were you possibly thinking of a different chip, or am I missing something?

 

[bauerbach]

hmmm the other LEDs look pretty cool. but yeah, I find it interesting that it would add up to 268watts... I know LEDs are more expensive, but I thought they were more efficient .

so a 250w halide is still king for efficiency as well as cost? if it was 20, it would be 430$, similair to what MH costs. but what do you think is brighter, 20 LEDs or a 250w HQI MH with a high quality reflector?

I mean, LEDs put off way less heat right? surely that must mean either more light or less energy consumption...

another thing, the mce LEDs, do those run at 350 or 700mA. the site lists it at 350 and the lumens for 350mA. but the PDF shows 700mA also? at 700 its 790 lumens, for like 8$? thatd be awesome if thats true.

 

[R33E8]

Those Rebels are lower bin. They won't be as efficient as other LEDs.

No P7 bin would allow 20 P7s to equal the output of 150 RB80s. It's more like 27.

150 RB80s is 336W. 27 P7s running at spec is 265W. 150 RB100s (if you can afford them) would use 268W to match the others in lumens.

However, The P7 is not well designed. Parallel is bad for LEDs.

With this kind of wattage, you may be better served by a bridge rectifier running from 110VAC, if the 120Hz flicker isn't too bothersome.

I meant the 20 P7's putting out a conservative 700 lumens and total about 14,000 lumen.. That would fall into his requirements of 10,000 to 20,000 lumens..

Also, I do not believe that parallel in LEDs is as bad as people on this forum say it is.. If you heatsink the P7 (or any hbled in parallel) well you "should" have no problems..

hmmm the other LEDs look pretty cool. but yeah, I find it interesting that it would add up to 268watts... I know LEDs are more expensive, but I thought they were more efficient .

so a 250w halide is still king for efficiency as well as cost? if it was 20, it would be 430$, similair to what MH costs. but what do you think is brighter, 20 LEDs or a 250w HQI MH with a high quality reflector?

I mean, LEDs put off way less heat right? surely that must mean either more light or less energy consumption...

another thing, the mce LEDs, do those run at 350 or 700mA. the site lists it at 350 and the lumens for 350mA. but the PDF shows 700mA also? at 700 its 790 lumens, for like 8$? thatd be awesome if thats true.

Well according to wikipedia, MH lamps get 65-115 lm/w. The p7's at 2.7amps will have roughly 70-90 lm/w.. In the long run, the LEDs will last longer, emit less heat (IR), and not have the dangers MH bulbs have:

Eyes

Although an excellent source of lighting for the reef aquarium, there has been concern voiced by some aquarists over the potential ill-effects of close-range contact with metal halide lighting which is demanded by the hobby. Some individuals have noticed temporary blurred vision even after very brief exposure to metal halide lighting. This blurring of vision could be linked to Photokeratitis (snow blindness) - the result of unprotected exposure to Ultraviolet (UV) radiation.

FDA cautions

Broken and unshielded high intensity metal halide bulbs could cause eye and skin injuries, particularly in school gymnasiums. See the following article from the FDA: / Ultraviolet Radiation Burns from High Intensity Metal Halide and Mercury Vapor Lighting Remain a Public Health Concern Also see: / Teachers battle dangerous lighting conditions and / Photokeratitis and UV-Radiation Burns Associated With Damaged Metal Halide Lamps

Explosion Hazard

Metal Halide Lamps of the quartz arc tube variety are susceptible to explosion at the end of their rated life due to the corrosive effects of the halide salts on the quartz. Burning universal position lamps horizontally also increases risk of explosion due to higher arc wall temperature. Mechanical shock such as vibration may also cause explosion, as may leaving the lamp on all day. As such, metal halide fixtures should be turned off at least 15 minutes each week to reduce explosion hazard, and a quartz shroud may also be added for further protection and to reduce halide ions migrating through the arc tube wall (quartz metal halide lamps only).

- http://en.wikipedia.org/wiki/Metal_halide


Do you know how many lumens the HQI MH bulbs put out? I only found that regular 250w MH puts out 23,000 intial lumens.. If the HQI is similar and put with a 99% reflective reflector then the MH bulb will beat the 20 P7's in brightness..

The MC-E led's run at 350mA for a nice trade-off between efficacy and lumens... They use Cree EZ1000 dies which are rated for 700mA for life. They can be run at a higher amperage but the efficacy will start to go down fast. And unfortuantly, they are $25 each, not 8 ...

 

[bauerbach]

man, well guess thats the nail in the coffin.

bottom line, the LEDs are less efficient and cost more. I cant find any major benefits. Im running heaters on the tank, so more heat from light would just reduce my heater use, a good thing. the spectrum from the LEDs isnt optimal anyway, 14-20k is what I want, I was going to have to throw a bunch of blue LEDs into the mix to cover that hole that I wouldnt need with MH.

as for danger, I dont stare at my MHs they are shielded with tempered UV glass as well. I wanted LEDs just to try something different, but I guess not. I understand why you'd use them in flashlights, but Im not concerned with space, I have plenty of it.

I guess last ditch effort, are there any studies that might show me the par value from an LED with a lens at 2 or 3feet from the source?

 

[Oznog]

Vs metal halide... well, it depends on your reflectors actually. Half the metal halide light is directed towards the rear and a poor reflector loses a lot of the light.

Metal halide cannot be dimmed like LEDs.

You cannot voltage drive these LEDs (9.45Vf on a 10V supply). It must must must be current-regulated on a per-series-string basis. Buckpucks or whatever will do this for you. Technically, if you used a higher Vdd and a ballast resistor that would reg the current but the efficienct and heat would suck.

Best idea is the AsianSignals waterproof LED covers
These are pretty cool actually, have an o-ring underneath and the through-holes for the screws are outside the o-ring obviously. They're sized to take a Star (or P7 round) inside there. So you can mount them on an aluminum bar, screw these covers down and have waterproof lighting without a complicated cover.

 

[Gomer]

Vs metal halide... well, it depends on your reflectors actually. Half the metal halide light is directed towards the rear and a poor reflector loses a lot of the light.

Reflector quality in MH has gone up a lot. Lumenarc and lumenbright reflectrs are pretty damn efficient compared to the old spider reflectors.

Metal halide cannot be dimmed like LEDs.

There are actual dimming MH ballasts in the reefing industry right now.

 

[bauerbach]

yep, I use lumenmax reflectors, in this case I would get the elite reflector. claims 99% efficiency. dimmable has some good implications, but most are too advanced for me, things like rolling output levels to simulate a cloud and such, I cant build such a controller, though still even if I could, questionable if itd be worth it

 

[VanIsleDSM]

99% efficiency?

I'd say that's most definitely a lie. They may use 99% reflective material to build it, but there's no way 99% of the light will be delivered into the tank.

That's where LEDs really come into effect.. they're optically so much more efficient in designs that need directional lighting.

Since The heat from LEDs is captured in a heatsink, it could much more easily be transferred into the water to reduce your heater needs. Alot of the MH heat would just be emitted into the room. Cooling the LEDs with the tank water that needs to be heated anyway would be a very synergistic approach that could make the whole operation extremely efficient. Not to mention the fact that the LEDs have a much longer lifespan.

The LEDs are a much more sophisticated approach that, if designed properly, I'm sure would out perform the MH as a whole system.

 

[bauerbach]

you may be right, but it would take a huge investment to ever find out. in the aquarium world we measure everything with par output. since thats what really matters in the end for us. but I cant find what kind of par these LEDs would produce for me after the optics and all, to see if it would net more light into the tank or not.

I kinda assume par and lumens must be related, but not in any easily comparable way...

actually none of its really comparable.

I mean, the MH is around 14k, the only lumen ratings I can find are for 5500k bulbs(which is the 20,000 lumens), which of course is substantially higher than what the 14k will get. so on that hand, its an unfair comparison to try and match 20,000 lumens. but on the flip side, I WANT 14k, so I would end up needing to add a bunch of blue LEDs, which for whatever reason I cant even try to add up since they are not even measured in lumens for some reason, plus the LED alternatives to the rebel dont come in blue, so it would have to be rebel for those.

very complicated lol.

btw, checked sunlight's website, the reflector is 95% reflective they say.

 

[Gomer]

The LEDs are a much more sophisticated approach that, if designed properly, I'm sure would out perform the MH as a whole system.

If an LED setup could do th 380-450region well, then yes. Right now, they really need actinic supplementation. This isn't a dominant wavelenght, but is important for color fluorescing from a visual appeal standpoint. the big 440-470 peak doesn' quite cut it as a standalone.

Also, MH do the deep reds much better.

That there is 1/2 the LED challange. Color rendition. LED can do the intensity part well, just not the color (this coming from someone who has build a few LED setups for FW and SW tanks)

 

[TigerhawkT3]

LM138 only has an output voltage up to 32V. Were you possibly thinking of a different chip, or am I missing something?

It's the LM317 that has a max output of 37V. The LM138 only cares about Vin-Vout.

LM138 PDF

"Since the regulator is "floating" and sees only the input-to-output differential voltage, supplies of several hundred volts can be regulated as long as the maximum input to output differential is not exceeded, i.e., do not short-circuit output to ground."

Regulating, say, 167Vin to 150Vout would be around 90% efficient.

Say, could I borrow one of those 250W MHs? For, you know, testing purposes and such.

 

[bauerbach]

it lives again, new data indicated the high temp MH might be as low as 50lumens/watt unlike their 4500k brethren.

at that point, LEDs might just cut it!

 

[bauerbach]

alright, reworking the design, I think Ive settled on XRE's as the most efficient and cost effective option. the MRE are interesting, but since I have the room available, 4 XREs are cheaper than 1 MRE, available in blue, and should be significantly easier to work with and cool.

so I believe anyway.

Id like to build these in modular 1' sections with 50 XREs running around 750mA.

since efficiency is my goal, what power supply option would be the most efficient? bridge rectifier or stick with my cheap little PC PSU(efficiency at this draw should be around 75-80%... not that great, but it would be extremely clean). however Id like to have dimming ability, which the PSU doesnt offer, could use a pot I guess, but getting less efficient I think.

 

[evilc66]

One flaw in everyones thinking (which I'm suprized Gomer didn't catch) is that you are all shooting to match lumen output. If you were to match LED to MH lumen for lumen, you PAR levels would be so high that you would scorch every coral in the first 6" of water.

Look at the commercial LED fixtures for aquariums. None of them produce the same lumen output, but successfully keep thriving reefs (albeit at a rediculous cost). It's all about PAR.

In the fixture that i have built, I have been able to produce more PAR at decent depths than a 14k MH at double the lumen output, and that was without optics. With optics I'm sure I can smash that.

You will still need a pretty serious array to match a good 250W MH with a good reflector, but it is much more in range than you think. The biggest issue is estimating how many LEDs (and of what color) you will need to acheive the PAR levels that you are shooting for. In my experiments I ended up footing the bill for a PAR meter, but that will end up costing you almost as much as the additional MH

 

[bauerbach]

One flaw in everyones thinking (which I'm suprized Gomer didn't catch) is that you are all shooting to match lumen output. If you were to match LED to MH lumen for lumen, you PAR levels would be so high that you would scorch every coral in the first 6" of water.

Look at the commercial LED fixtures for aquariums. None of them produce the same lumen output, but successfully keep thriving reefs (albeit at a rediculous cost). It's all about PAR.

In the fixture that i have built, I have been able to produce more PAR at decent depths than a 14k MH at double the lumen output, and that was without optics. With optics I'm sure I can smash that.

You will still need a pretty serious array to match a good 250W MH with a good reflector, but it is much more in range than you think. The biggest issue is estimating how many LEDs (and of what color) you will need to acheive the PAR levels that you are shooting for. In my experiments I ended up footing the bill for a PAR meter, but that will end up costing you almost as much as the additional MH

I should be able to borrow a par meter from my local reef club.

my real concern is spectrum(those are a week bit more expensive), and ensuring there is enough LEDs to get the job done, with proper acclimation, extra par should be ok. dimmer can take care of the rest.