LED Array for a Pico

[pulse13]

Sorry to butt into the conversation without introducing myself. I've been around the forum for years but I don't have the time to post regularly, but this is a topic that interests me alot, and I've seen lots of posts here on LEDs and lots of misinformation. I've actually been thinking about doing some experiments with LED lighting for several years now, and I'm planning on waiting until the K2s come out in march. Maybe I'll have some time to go back through the thread and give some of my thoughts later, but there is no way 40 watts of LED power is the equivalent to 250 watts of metal halide when the LEDs you're using have at most an efficiency of 20 lumens per watt, compared to a metal halide that is at least four times as efficient at 80-90 lumens/watt. The differences you're seeing in lux levels have to do completely in the way in which the light is focused and spread from the two different types of sources, you are not measuring the total output from the light source. Try taking a flashlight and measuring the lux with a reflector and then take the reflector off and measure at the same distance and compare the change. Lux is not a measure of total light output, its a measure of brightness.

[zachtos]

Lux is not a measure of total light output, its a measure of brightness.

 

wow, you really need to read the thread, atleast part of it. lumens mean nothing with LEDs. they are focused on what we want to have light on. such as corals. thus LUX/PAR are the best measurements. who cares about lumens that are shooting up and out of the tank as wasted energy? wattage does not mean ANYTHING. apples to oranges really. these bulbs produce light in COMPLETELY different manners. That is why they dont waste so much energy as heat. why are you so skeptical of the results? Is it not posted in a scientific manner? science bad!

[neanderthalman]

Thank you pulse, the defintion and comparison of Lux and Lumens was a while ago. I'd suggest reading the whole thread, but I'll give you a synopsis. Lumens gives you a better measure of the TOTAL light output of a bulb. Lux gives you the luminous flux that your corals will see. It takes such things as distance, reflector, absorption by the media, collimation, and other factors into account. When measuring lighting for a tank, lux is the better way to do it, as it doesn't measure the light that doesn't make it to the tank.

 

If you measure 1000 lux from MH and 1000 lux from LED, at the same distance, guess what? The intensity is the same. Lux is the measurement that matters to your corals. They don't care about lumens, as lumens includes the light that doesn't reach them.

 

The focus of the LEDs is a big part of the reason you can get high intensities from LEDs. Lets say you have a cone of output from an LED array, with 1000 lumens of output. That cone has an angle of around fifteen degrees. At a distance of 0.5m (18"), the cone will illuminate a circle with a 13.4cm radius, or an area of 0.056 m^2. If the light output is 1000 lumens, then you will have a luminous flux of 1000lumens/0.056m^2 = 17857 lux.

 

By contrast, a MH (without reflector) will illuminate in a sphere-like pattern. At a distance of 0.5m, the total area illuminated will be 1.57 m^2 - 28 times the area of the cone from an LED source. 1000 lumens spread over a 0.5m radius sphere gives a luminous flux of 636.9 lux. Granted, with a reflector, this is improved dramatically, but only a percentage of the light from MH actually reaches the tank. With LEDs, nearly all of the emitted light reaches the tank. THAT is the reason you can get high luminous flux from LEDs, the light is focused.

 

If you could focus all of the light from a MH to a cone of angle fifteen degrees WITHOUT significant losses, you will get more light in your tank - this is exactly what LEDs are capable of.

 

When you put that Luxeon array together, be sure to post here. We don't have anyone who's built an array from luxeons yet, and I'd like to see one for comparison.

[pulse13]

Thanks neanderthalman, that was a great explaination, and I might have posted something along the same lines if I had the time and if I could even explain it that well. About my first post, I just didn't want to see another person posting on here that LEDs are almostly perfectly efficient, produce zero heat, etc and misinforming people like has happened here before. I'm glad you posted that. My only concern with the experiment is I didn't think the difference in the radiance between LEDs and other sources would result in that much of a diffrence in lux. I figured at most, due to the better spread pattern of LEDs, you could get away with maybe half the LUMENs of a PC. And if you search way back, you will find threads from others putting together luxeon arrays for nanos here, one of which claimed to be ready to start selling them. Zachtos, I do know quite a bit about the properties of light and have been reading about LEDs for years, and I'll get around to reading the whole thread. And I do know about science, its what I do for a living. It would be more scientific if it were more controlled. What about taking all of the lux measurements underwater to eliminate variables such diffraction, reflection of light back into the tank of the glass, etc.? What about adding a diffuser to help even out the beam from the array? Did you take multiple measurments from different positions ( at same depth) and also replicate measurements? What is the variation like?

[zachtos]

all measurements were an average of the peak lux detected at that height from the bulb tip to that distance in air. The water penetration measurement would likely not provide anything useful. The only reason MH penetrates water better then PC is because it provides more intense light at the surface where it enters. So do the LEDs. I do not have a underwater meter or par meter nor do I have a deep tank or many bulbs to test. I am a nano reefer with a shallow tank and budget. other reefers could help by providing MH/PC measurments. I am tired of not getting any answers at reefcentral or googles. I can only find PAR readings, not much on various LUX depths. I too have been following LEDs since I stopped reefing a few years back. They have made much leeway in the realm of mcd offerings. (11,000 mcd to 40,000 mcd in 2 years.)

[pulse13]

Heres the problem I have with the results. So, if we want to light a certain surface area (tank bottom) at a certain lux brightness, it takes the same amount of lumens. Brightness (lux) = lumens (light quantity) / surface area. Unless you change the area of the brightness, the lumens must stay the same.

 

A 250w aquarium MH produces ~ 80 lumens per watt, so thats 250 * 80 = 20,000 lumens.

 

If your using the brightest most efficient 5mm LED, which you're not, thats 80,000 mcd @ 12 degree viewing angle, at ~ 3.7 volts and 25 ma. Thats an estimated 2.75 lumens per LED according to this calculator:

 

http://led.linear1.org/lumen.wiz

 

http://led.linear1.org/how-do-i-convert-be...las-and-lumens/

 

and its 3.7v * .025a = .0925 watts per LED, so you have

 

2.75 / .0925 = 30 lumens / watt.

 

You are saying 40 watts of LEDs cover the same area at the same lux as a 250w MH. Your LEDs are putting out 30 * 40 = 1200 lumens, and the MH 20,000 lumens. So for the MH to be as bright as the LEDs, due to the fact that the MH is inefficient because it doesn't focus the light as well as the LEDs, 1200 / 20000 = .06 * 100 = 6%. So your saying that on your MH setup only 6% of the light is reaching the tank, the rest is scattered outside the tanke and wasted. 94% of the light of a MH is wasted because of the reflector. Thats some logic there guys. Maybe you should be working on building reflectors for MH setups instead of LED arrays since they are only 6 % efficient at reflecting and directing light into the tank.

[sammydee]

If you are going to argue scientifically then you need to post your sources of information. Where did you get the figure of 80 lumens/watt for the 250 watt metal halide.

 

I'm not entirely sure what you are trying to prove here... If Zachtos has been truthful about the results (and there is no reason why he should not have been) then there is no point in trying to use science to prove them wrong just because you think that they must be wrong.

 

Look at neanderthalman's readings from earlier in the thread - I believes he quotes over 12,000 lumens from his ~40 watt LED array.

 

Sam

[pulse13]

12,000 lumens from 40 watts would be 300 lumens per watt, which is closer to 100% efficient than any light source invented so far. The very best LEDs are 60 lm/w, in 5mm format 30lm/w is as good as it gets.

 

A quick google search will show you the efficiency of MH. It should be common knowledge to most people that use MH, the efficiency is one of the reasons people choose MH over PC and other technologies. Its the reason that street lights and warehouse lights are MH.

 

The first hit from my goolge search: (aquarium "metal halide" "250w" lumens bulb)

 

http://www.discount-hydro.com/lightbulbs.asp

[zachtos]

error: neanderthallman got 12,000LUX, not lumens.

 

Well there is no satisfying you pulse.

 

I posted the results in a scientific manner, which means my results are repeatable. Build your own array using the same parts. Take your LUX reading or better yet get a $300 PAR meter and post the PAR readings. Then you will be satisfied. I will continue to post my coral growth results at Zachtos' Array My macro algae is already growing at an increased rate in only 3 days of LED lighting. I only have GSP right now but they are very happy and open up minutes after the daylights come on and close about an hour after they go out.

 

Also, I thought neanderthalman made good points about the lumens. I too DROVE MYSELF CRAZY trying to analyze the lumens vs mcd goodness. There is no answer. LEDs are a different light source and lumens are useless measure for these. Lumens are great for lights that shoot light in every direction but LEDs dont work that way. They are focused, which means LUX/PAR are all you have to work with and this stuff just isn't in the spec sheets. The best we have to work with is milliamp draw and mcd output. But even watts are useless, as 2 LEDs w/ equal power usage can provide totally different LUX readings. Sorry, but you really do have to manually design these to get results. Which is what I and a few others here have done for you to analyze and try out.

 

Here is the math to make you even more confused. Using your online converters, 20deg beam angle, 40,000mcd rated LEDs, 350 LEDs. converted to lumens = 3.8 lumens/LED. Now multiply that by the 350 I used. 1330 lumens. now to get lux you must divide that by the area in meters squared. converting my 55 inches squared (5"x11" matrix) I get 0.03548 m^2. now 1330lum/.03548m^2 = 37,500LUX. That is about 56% of what I actually measured at 0" from the bulbs 21,000lux actual maximum. these were 3.4Vdc leds running at about 25mA each also. so 36.4LM/W is what you get. MH gives 80-100LM/W... so supposedly 220% brighter... now the theoretical math just DOES NOT correspond to real life measurments. There are too many factors to make that so far off. MH lumens are NOT going to ALL be reflected down on a specified area to get high intensity/LUX. The LEDs on the other hand are guranteed to shoot down nearly ALL of their lumens. Now each LED may not be completely the same and there is no way to get a real standard for what they are actually sending us. mcd is the best we got to work with. Well the results are posted and the math is shown to be only a rough guide. Only physical results/meausurments will work now. a quick MH calc, the MH dispurses its 20,000lumens over a greater area on average, that is why it gets less lux then the leds, 20,000lumens/0.3716 (2'x2') gives 53,821lux... obviosly that is incorrect amount of light that actually gets there, maybe 1/3 to 1/2 of that actually gets to the surface. That is why I started with this small prototype tank, to run a cheap small physical test before scaling up.

[pulse13]

Even if it is 1330 lumens, how do you figure that 1330 lumens can cover the same area as 20,000 lumens of a MH at the same brightness? Do you really think that only 6.65 % of the light from the MH relative to the LEDs are getting to the tank? How do you figure that MH reflectors are that inefficient? Lumens are a measure of total output of light, given the spectrum is the same, how is 1 lumen of light from one source not the same number of photons as a lumen of light from another source? So we can't compare lumens between incandescant, fluorescent and metal halides because they're not the same light source? Did you measure the exact surface area illuminated at a given lux with the meter? how? Its not surprising you can get a higher lux from a LED with less total output, its more focused, but what you can't do is illuminate the same area with 6.65 % of the lumens, there is no way LEDs are that much more efficient at focusing light. Can you light a small tank to the same lux level as a MH over a much larger tank with much less light output from LEDs, yes of course, the area is smaller. You'd have to have a pretty bad MH reflector for only that small of a fraction of the light to enter the tank. Sorry to start something but i'm just trying to help. I've been thinking about all this stuff for years. I argued my point now I'm done

[schaadrak]

I think pulse works for Ushio or something.

 

Anyways, I have a question about LEDs. What limits their size? I realize that the light emits from the NP junction, but why not just make the place they join together longer? Is the small size of the housing what focuses the light?

[pulse13]

I think the biggest limit to the size of the die is conducting all the wasted heat away from the chip and keeping the chip cool. The highest output LEDs are from Lamina Ceramics (www.laminaceramics.com) and use multiple dies per unit to help spread the heat out, they can reach ~ 25 watts for white, although they have much bigger prototypes. The highest output for a single dye are the Luxeon K2s (http://www.lumileds.com/products/line.cfm?lineId=18), they can run at 1500ma, or about 7-8 watts I think. Both the Luxeons and the Laminas require a heat sink to help remove the waste heat. The K2s are able to run at higher output because they have better thermal properties than the older Luxeon line and can run at a hotter die temp without damage. They are also much more efficient.

 

On the 5mm LEDs, the die itself sits in a small reflector, the light is then further focused by the clear epoxy housing which is shaped into a lens.

 

And no I don't work for Ushio, I plan on making an experimental LED setup myself, I'm just arguing that more output is required.

[neanderthalman]

Pulse has VERY good points, lets not get all defensive or angry.

 

MH can light a large footprint FAR more efficiently than LED. There's no question in my mind, and I have never argued otherwise. However, my aquarium as a footprint of 0.0078 m^2. Find me a MH reflector that is designed to collimate all of the light from MH into such a small space. It just doesn't exist at this time.

 

In order to get the same intensity as my array on a larger footprint, the number of LEDs, and therefore the overall LUMENS output by the array increases to compensate. At the density I've wired, I'd have just over 34000 LEDs to cover 1 m^2 at the same intensity as 150W MH. See how it gets out of hand really fast?

 

A 10g aquarium has a footprint of 0.125 m^2, and would therefore require 4250 LEDs to light at the same intensity as MH. At 0.1W/LED, that would be over 400W of LED lighting. MH is a more efficient light source, no question. It is advantagous on SMALL tanks and ONLY small tanks because they do not make MH reflectors that can concentrate the light onto such a small area.

 

MH cannot efficiently light a small aquarium, whereas LEDs cannot efficiently light a large aquarium. The tipping point is at a footprint of 0.44 m^2. That's the same as the footprint of a 2.5g aquarium. If you have a standard 2.5g, you would have to use 150W of LEDs to simulate 150W of MH. LEDs are only more efficient for lighting an aquarium smaller than that.

 

If a company releases a 10W MH with a reflector designed to focus the light on a small area, then LEDs are obsolete, and I will buy one. Tiny reef tanks are a very small niche market, so I don't see this theoretical product being released any time soon. In the meantime, LEDs are a better solution for pico reefs, barring any issues with spectrum.

[pulse13]

I totally understand you now neanderthalman. Now if I can just convince you to use K2s Er, better get back to work first, I just happen to be stuck on the PC a lot today.

[Nednai]

How do you figure that MH reflectors are that inefficient?

I’ve never used MH but when I see pictures from people on nano-reef that do use them it looks like it could light tanks 4 times as large as the one it’s being used over. By appearance alone MH reflectors looks to be inefficient for the typical nano-reef application. I was hoping someone with good manufacturing skills would combine a condenser lens, like ones used in photographic enlargers, with the reflector to better focus the light.

[zachtos]

I'm not sure why neanderthallman said you need so many LEDs. My array of 380 LEDs can light 6"x12" w/ 15,000 LUX at 6", very close to a 175W-250W MH. And that easily covers a standard 2.5G aquarium. I came up w/ number of around 5000 LEDs for a typical 90G tank which would be 425W, still 60% of your normal wattage used for lighting a 90G not counting the needed chiller unit, and that should provide 175-250W MH intensity using the same stuff I used.

 

granted I don't have a PAR reading yet (but my LUX meter is more biased towards 550nm light which these LEDs are only 20% intense at the frequency supposedly). I'm hoping someone at reefcentral will loan me a PAR meter.

 

On another note, I emailed besthongkong to see if they can run a spectrograph for those LEDs mentioning the marketability etc. I doubt they will do that.

 

Also Pulse, if your planning on doing an LED setup, keep it small to start, because you may end up ordering different LEDs If your results are weak. I already did that once halfway through my 10mm LED project (they were junk, atleast my LEDs/p.supplies have stayed under $250 overall). I guess for now just check in every few months on our threads to see how the corals are doing for that final verification. No need to waste money/time since we already have... well, hopefully not wasted.

 

As for myself... I'm already thinking about scaling up to a 2'x1.5' footprint custom tank w/in 8 months. That way if I do get a 90 or 120 gallon, my array will be half done already, and I could throw my 250W MH on the other side for real comparisons.

[zachtos]

"Would get you the spec later on. we just get back from chinese new year holiday.

it'll be very busy days

Kung Hei Fat Choi!"

[neanderthalman]

Zachtos, my calculations are as follows:

 

Tank, hexagonal, 11cm per side. Footprint can be divided into six equilateral triangles with 11cm sides. Total area of each triangle is 1/2 bxh. b = 11cm, h = 1.73*5.5cm. ah hah! found an error!

 

Total footprint area is 0.03 m^2

 

267 leds divided by area gives 8503 LEDs / m^2

 

A 10g would need 1063 LEDs to keep the same density that I used. Keep in mind that it is the density of LEDs that determines the intensity of the array.

 

The tipping point for energy efficiency is at approximately 0.176 m^2. This is between a 10g and a 15g. This still makes sense as a MH reflector can illuminate a footprint MUCH larger than a 10g aquarium.

 

Economically, however, it is cheaper to purchase a MH setup for, say, a 10g aquarium. You would, however, have to replace the MH bulb every year, while LEDs should last much longer. Wiring up 1000 LEDs isn't something I'd really cherish doing, however.

[zachtos]

Well I dont think anyone should ever attempt to drill all of those holes by hand either. I HIGHLY HIGHLY recommend the polycarbonate sheet w/ predrilled holes for $25, it is 4'x2' and has the possibility to be up to 12led/in^2, I went w/ 6/in^2 to get my 15,000lux at 6". It took me 30 min to redrill the 350 holes larger and 10 hours to bend/cut/solder 380 LEDs, 3hours for the 100LED fuge, so that is about as fast as you can go roughly. It's all about time/money. If you think it's worth those hours and time go for it. My girlfriend crochets huge picture thingies for months, I dont view this as any different. Don't expect it to be an overnite project obviously. Just like building plastic model airplanes/cars etc. For myself, I think if I scale up, I will order maybe 10 of those 48LED cluster bulbs to ease the manual labor. $15/4-6in^2... hard to say considering I need to test the coverage needed and the spacing to get full coverage. then I would order the other 50 or so bulbs.

 

I agree w/ neanderthallman, i recommend restricting the LED projects to tanks maybe 6gallon-ish and under for cost / time purposes... for now. If you wanted to go bigger, I would suggest looking into those cluster bulbs I mentioned earlier (they contain the exact same LEDs I used and are prewired for a single 12Vdc connection). I think those bulbs would be great for getting extra light in a big tank directly onto a prized coral specimen or clam. maybe you could keep one clam in a big tank lit w/ PC's this way?

[pulse13]

I decided to do a few quick cost per lumen calculations for various LEDs so to see just how much light you can get for your money (considering LEDs/optics only, not heat sink, mouting or driving variables).

 

For the 5mm LEDs:

 

From zachtos' post, 4600 LEDs cost 1600, so thats $ .348 per LED, and at 3.8 lm per LED you get

 

Cost in lm/$: 10.9

Total cost per 1000 lumens: $91.74.

Total number of LEDs: 263

Efficiency (see post above): 36.4 lm/w

Total watts for 1000 lumens : 27.5w

Watts per LED: 0.085

 

For Luxeon IIIs:

Off the top of my head, I remember seeing Lux III LEDs with optics for about $ 10 each on ebay. Lumileds lists stars minus w/o optics as 4.05, for 80 lumens total output, a lot cheaper now than they used to be. They give a value 19.8 lumens per dollar here: http://www.lumileds.com/products/family.cfm?familyId=8. So thats less than half as expensive, but doesn't include the star LEDs or optics. I think the optics are about $2 a piece. So that brings it to about 13 lm/dollar. Of course this doesn't include the cost of heat sinks. At 3.7 watts each, they have an efficiency of ~ 21.6. LuxIII specs:http://www.lumileds.com/pdfs/DS46.PDF

 

Cost in lm/$: 13

Total cost per 1000 lumens: $76.92

Total number of LEDs: 13

Efficiency: 21.6 lm/w

Total watts for 1000 lumens: 46.3

Watts per LED: 3.7

 

For Lamina Ceramics BL-3000 series 5500k white:

http://www.laminaceramics.com/docs/BL_3_White_55.pdf

26 watts, 560 lumens, $ 81.48 each not including optics at this site: http://www.digikey.com/scripts/dksearch/dk...US&Cat=33227618.

Optics at 5.14 here:http://www.digikey.com/scripts/dksearch/dksus.dll?Criteria?Ref=199880&Site=US&Cat=33227290

Giving a total cost per LED of $86.62.

 

Cost in lm/$: 6.47

Total cost per 1000 lumens: $ 154.56

Total number of LEDS: 2

Efficiency: 21.53 lm/w

Total watts for 1000 lumens: 46.4

Watts per LED: 26

 

Luxeon K2 :

 

For the model LXK2-PW12-S00 here:http://www.lumileds.com/products/line.cfm?lineId=18

Assuming $6 including optics (an estimate, price not listed for this model or optics), they produce 60 lumens from just 1.15 watts.

 

Cost in lm/$: 10

Total cost per 1000 lumens: $ 100

Total number of LEDS: 17

Efficiency: 52.17 lm/w

Total watts for 1000 lumens: 19.1

Watts per LED: 1.15

 

As you can see, the K2s cost a little more per watt but are much more efficient than Lux IIIs. K2s are binned into two groups, those tested at 350ma and those tested at 1000ma. They can ALL be run at upto 1500ma, but the color shade and efficiency won't be guaranteed to stay the same and the life will be lower, but it gives tons of room for improvement if you plan on running at 350ma like my calculations were for. They also can be run much hotter so require less heat sinking that other high power LEDs. The lifetime has been greatly increased, at 1000ma they are at 70% inital lumens after 50,000 hours. It would probably be safe to assume much better than 70% after 100,000 hours running at 350ma.

 

All of these calculations of course are estimates, they don't include many variables and prices can vary. I also didn't consider Cree, Nichia and some Chineese manufacturers of high output LEDs that may be cheaper. Luxeons are also sorted by bin, meaning that these are typical numbers and can vary widely. They are binned by color (including many white shades), voltage and efficiency. Better K2 bins will be out, and much better Lux III bins exist than the calculations were for. It gets complicated. But this should give a general idea of what it costs to use many 5mm LEDs compared to just a few high powered ones.

 

Also,

[pulse13]

One more comparison while I'm at it. Lets see how much it costs to run K2s for their life of 100,000 hrs compared to a 250 watt metal halide. Only considering the cost of LEDs, metal halide bulbs, and electricity, and for the same output in LUMENs of the MH, about 20,000 lumens:

 

Number of LEDs: 340

Cost of LEDs: 2040

Life in yrs at 12hrs/day : 22.83

Life in hrs: 100,000

Total watts: 382

Kilowatt hours per 22.83 yrs: 38,200

Cost in electricity for 22.83 yrs at 8c/Kilowatthour: $ 3056

Total cost for 22.83 yrs: 5086

 

250 watt MH:

Cost of 23 MH bulbs ( 1 per year) @ $60 each: 1380

Kilowatt hours: 25,000

Cost in electricity: 2000

Total cost for 22.83 yrs: 3180

 

So it looks like in order for LEDs to compete with MH in terms of total output, the cost needs to come down and the efficiency needs to go up, and you need to run them for their rated life to come close to getting any cost benefit. Maybe in a few years they will improve enough to consider using on large tanks.

[zachtos]

like we said, if you want to experiment. try a small tank for starters. besthongkong.com has an ebay store and I bought my entire array for $75, and the powersupplies from another company w/ parts for about 65$. still cheaper then a MH which I wouldnt be able to cram in that tiny tank w/o heat issues anyways.

 

*note barely a week in and I am seeing corraline growth all over my power head, small speckles of the hard purple stuff.

[Llamaguy]

Does anyone want some 4000mcd 3mm whites? I ordered them on accident , plus I didn't need the ones I wanted anyways, I have 200 of them and 50 3000-4000 5mm blues. Could use them for a fuge light I guess.

 

The 3mm have 40 degree view angles if you wanted to know.

[hromm]

To neanderthalman, zachtos and others: You guys are doing cool stuff! I have been reading your post the whole evening and got so excited that I decided to join the nano-reef.com and make this post.

 

The problem pulse13 has with zachtos' high lux readings of LED array may come not from the poor reflector of MH bulb but from the wavelength range the lux meter is sensitive to.

 

The other day I measured spectra of some LEDs with spectraphotometer in our lab. It's an expensive piece of equipment and I would trust its readings. Here what I've got (see attachment, mcd and viewing angle are from LED's spec., LEDs were from LDdiodes.com.)

To my eye UV LED was about 10 times dimmer than the blue one, and the white LED was much brighter than the blue one (and this correlates with mcd values).

However, in reality UV LED emmits more photons, as you see from the graph.

 

Similar misrepresentation may happen if one compares MH lux output with LED: lux meter may not catch blue/violet part of MH's spectrum and by eye LED array may seem brighter than MH bulb.

 

To zachtos: Can you measure lux of your UV LEDs only? Just to compare with lux from white LEDs. My guess you will get much lower readings of lux per LED. Maybe one should compare UV and blue LEDs (not white LEDs) with MH, especially with 20K MH.

[sammydee]

Bulk buying over 500 or so LEDs drives the price right down. they end up being about $0.20 each for 12,000MCD 5mm ones.

 

Sam