DIY LED Aquarium Lights.

[bitterman]

Great discussion guys! Very interesting.

I'd be interested in putting together a 6-12 bulb fixure for a nano tank. Anyone else on board?

I'm interested in doing something..... Even if it is only a few parts as part of an order so I can do a proof of concept. My 12 gallon FW has it lights burt out..... so.... Maybe 2 of the below LEDs and a Dimable puck? so about 100 US.... I have some 12V power supplies here that will under drive things a bit.

For a small tank these put out 4x the lumens with the same 350 mA driving them... but they are alot more expensive....... http://www.ledsupply.com/creemce-w430.php 4 would be about the same lumens as 12 white of the other ones, just not sure about distrubution of light. But on a 10 Might be ok..... too bad they are $34.49  ea (USD).

[Canoe]

Thanks Hookup, great stuff
< I may not have to read the article link you supplied>  ;-)

... PAR meters are measuring available, useable light... flickering is not going to be an issue, because it's measured in meters^2/sec, you can kind of think of that as a VOLUME measurement over time... and flickering is taken into account... so get  your PAR meter reading high-enough to create saturated levels of photosynthesis and you win... flickering or not, no difference...

... The initial setup as described in the original thread (first post) is NOT typical.... It's so far off the "normal" that it is suspect.  So few bulbs producing such high PAR is completely a unique event.. not saying it didn't happen, but this is NOT the norm...

This is great. It appears that PAR readings represent the sum of the energy below the power curve of the light. It also adds credibility to the PAR readings obtained in the original article: LEDs driven continuously by DC have a flat line continuous power output - for the same perceived brightness, more total coral-usable energy.

4.  They create a STRONG spotlight effect, this can be problematic for coral placement...  very small shifts in your corals can put them into and out of saturation levels... initially this can cause corals to "burn" with too much light, or grow-slowly with too little PAR.

... the other unique thing about that build is the builder did NOT use focusing elements....  As discussed in that thread, at length, focusing elements are used to push the light into sharp-focused cones to get deeper penetration into the system.. this means coverage area goes down, so more LED's are added to fix the problem of coverage...

That's what impressed me when I read the specs on the LEDs he used. From their 100% brightness on axis to a LED, they're still putting out 50%, smoothly towards the sides, at a little under a 90 degree spread for the coloured LEDs and a little over 90 degrees for the white. Given enough LEDs to cover the area, should be no problem. Focusing/concentrating optics should not be required (and not desired) except possibly for particularly deep tanks?

One of the things I love about tank lighting, is the light shimmer one gets from metal halide bulbs. Being so small (compared to a fluorescent tube) MH bulbs are a specular light source. Compare a tank wide tube to a cloudy day and a MH bulb to the sun on a sunny day. Cloudy days give you very soft shadows and soft, even hidden, detail. Sunny days give you sharply defined shadows that emphasize small visual details. Hence, the MH specularity is what gives the light shimmer from surface water ripples. Other lightning and MH reflector proprieties determine evenness of light coverage and the contrast of the shadows. LEDs are also a very specular light source. Provided multiple LEDs don't completely fill in the shadows of each other's light, there is potential for lots of shimmer and lots of great visual detail in the tank. It is credible that such wide spread LEDs provide both specular key lighting and each other's fill (-the-shadows)/ambient lighting while still producing shimmer. The shimmer from 12 light sources won't be the same as from one or two NH bulbs, but there should be shimmer.

5.  Their performance over 50,000 hrs is not documented in terms of PAR output, to my knowledge.... What happens 2 years from now when they Look as bright, but are not... All of the documentation I've read says they "last" 50,000 hrs, but that is an On/Off thing... time till burnt out... there is alot of gray area there... Might be well published, i just haven't seen any, and I've only just started to look...

Say we agree, that LED PAR is 100% constant over the 50,000 hr lifespan... woo-hoo, saturation levels achieved, long lifespan... win/win... wait, what about color temp of the lights... what if after 5,000hrs they shift into the yellow/green/red area... i.e. lose some of their blue-ness... well the corals do not care, but we do.. the colors are gone... now what?  Start replacing perfectly good LEDs??  Again, that's entirely spectulation... but that's the point.. no one is publishing this type of study data yet....

Fortunately, the last thing a white LED can do is become less blue (it's driven by an underlaying blue LED) and a blue LED can't shift its wavelength, so no matter what happens, the bulk of the light energy, even after 50,000 hours, will still be in the major coral benefit colour of blue. There may be a visually unpleasant perceived difference in colour temperature of the white LEDs over time, if say, they output less yellow, or less orange, etc., if the coatings that re-emit blue light to other wavelengths degrade over time (not likely? no data yet) or use (likely tied to % of maximum current, if the failure of the first generation white LEDs are an indication).

As I understand it, the mean-time-between failure is well over 50,000 hours driven at maximum current, which the original article does for the white. Less current or lower duty cycle (pulsed current) results in more service time. The manufacturer also states we'll have at least 70% output at 50,000 hours.

So, we can drive the LEDs below their maximum current and get more life and brightness for longer. Dimmable drive circuits achieve this (for those not making their own drive circuits, dimmable drive pucks are available). Use enough LEDs to ensure we can get our full target PAR after the LEDs have degraded to 70% output, and without driving the LEDs near their maximum current. This should ensure enough PAR to keep the tank healthy for over 50,000 hours (11+ years). The only risk would be early failure of an LED or other component (it can happen), or an unpleasant colour if the coatings in the white LEDs degrade and cause a drop off in wavelengths other than blue.

To know how many to combine, it will take some experimentation measuring PAR values with dimmed LED strings.

Turning off the white strings while dimming the blue (be it Royal Blue or Blue you prefer the look of), gives you built in moon lighting. With the blue dimmed way down, it shouldn't affect LED life too much. Or, have too blue strings: turn one off, leave the 2nd dimmed way down for moon lights.

(a long one, hope I haven't made too many mistakes)

[bitterman]

FYI the white LED's were driven at a higher current as the LED specs say 5000 K to 10000K and it is obvious the builder wanted the light frequency closer to the 10000K for his tank mixed with the blue to simulate antinic. Driving them at lower levels will result in a less white light and go more twards the yellow.

I am worried driving them this hard will effect the life expectancy of the LED's and they may not hit the actual 50,000 hours. I guess that is why they used the big heat sync and the fans to keep the temp down and hopefully make it live. I wonder how long it will actually last. Some of the LED manufactures had had issue with lights burning out in short time-frames.

Bruce

[Hookup]

Agreed completely Bruce.  And Canoe, I also agree with your points above, for whatever that's worth.. haha...

LED systems are a great theory at this point that seems to have significant benefit... Xenon is using them on his system, and has been playing with them to get the results he's looking for.  The nice thing about Xenon's setup, is that he is running high-tech and I'd put his system paramaters, waterquatlity, stability up against anyone out there... so if he cannot get his corals to grow and color-up, the major factor, in my mind, would be his LED lights...   Not a definitative test by any means, but it's a starting point.

The original article, the DIY guy keeps changing and tearing down his system.. so coral growth cannot be effectively measured and contrasted with the new LED lights he put on... he's changed his lights, torn-down his aquascaping, etc.. all in a period of 3-4 months... so again, other than having a great build that's easy to follow, we're not going to get any reliable results where it matters most, in the coral health and color...

Right now my focus is on skimmer testing and design, so I'm not going to thow money at an LED system for testing... I'd like to, but I cannot... I might have access to a PAR meter if anyone does go this route...

[bitterman]

For a proper test someone here needs to a comparison test using 2 smaller tanks like 10's or 20's:

Build a fixture for say a 10 or 20 gallon to keep costs down. Setup another 20 Gallon beside it with conventual lighting.
I'd lean towards the 20 since you could DIY a 24" T5HO setup. Pick bulbs to keep lighting spectrum as close as possible to the LED.

-Put all frags in the tank same place etc.
-Use the same amount, quality, sized pieces of LR (next to imposible to get an exact match).
-Measure the par reading for the tanks were the corals are paced and try to ensure placement in equal flow/light.
-Measure electricity usage between the two tanks.
-Keep same # of hours of light.
-Monitor usage of core aquarium water chemistry and compare.
-Follow same water change schedule.
-Target same Mg, Alk and Ca levels and dose daily keeping track of how much of what is used.
-Run test for 1 year and see what the tanks look like side by side every month.
-Don't even think you would need a fish in the tank.
-Keep the same C-U-C

Maybe one of the forum sponcers would sponcer this, people here could donate different parts? They would get all the corals and LR back at the end bigger and better we hope a the end of the test... but someone would have to shell out for the tanks and lights, powerheads, DIY time etc.

Bruce

[Hookup]

Tis a good testing setup... I would only suggest that getting a 75gal tank would be a better way to go, and light 1/2 of it (24") with LED and the other 1/2 with whatever else was your baseline...  then use coral-frags of same size, polyup count, etc under each lighting setup... it's not a "show tank" so you can simply build some egg-crate and place them at the same height/flow...

The main advantage is 100% of the water paramaters now effect both systems/setups equally... You just need to get balanced and equal flow on both sides, and again, 2 powerheads of the same type, could be positioned at either end of the system... etc...

Twould be a good test for sure...

[bitterman]

Tis a good testing setup... I would only suggest that getting a 75gal tank would be a better way to go, and light 1/2 of it (24") with LED and the other 1/2 with whatever else was your baseline...  then use coral-frags of same size, polyup count, etc under each lighting setup... it's not a "show tank" so you can simply build some egg-crate and place them at the same height/flow...

The main advantage is 100% of the water paramaters now effect both systems/setups equally... You just need to get balanced and equal flow on both sides, and again, 2 powerheads of the same type, could be positioned at either end of the system... etc...

Twould be a good test for sure...

I was orignally thinging this, but was worried about light spillage from one light to the other.  I have 2 20 gallon tanks that are drilled... could plumb them together..... with a sump (got one of those laying around too...) Just no spare pumps. to keep water chemistry the same and not effect light spilling from one to the other.

Got extra eggcrate also.

Bruce

[Canoe]

In regard to drive currents in the original article, I could be mistaken but I believe he was driving the Cool White at 1000ma and the Royal Blue at 700ma, while both are rated to 1000ma continuous maximum. With appropriate heat sinks - required for all high power LEDs (which means any we use for our main lighting) - they should exceed the 50,000 hours service. I would recommend against driving continuously near the maximum rating of 1000ma, but continuous 700ma on each would be easier on them and a dimmable drive circuit that is high-frequency pulsing the current to a 700ma peak would be even easier, but would have a reduced PAR. PAR can be increased by adding more LEDs. More cost, but greater reliability in expected/anticipated performance over time. Given the light spread of these LEDs, adding more should not cause hot spots or burning. REMINDER: the neutral and warm white XRE LEDs have a maximum continuous rating of 700ma, not 1000ma.

That type of testing would certainly be comprehensive, and presumably definitive, but is it necessary?

If we can trust, as in truly rely upon, the following:

• PAR meters work the way they're claimed, and
• corals only need enough light to get to saturation for optimum growth (not health?) and the wavelength that gets them there is irrelevant,

then we should be able to rely upon the use of PAR meters to ensure that the light we provide is sufficient, healthy and if to near saturation, even optimum. There should be no surprise unsuitability of LED lighting for aquarium use. The lurking surprise is in the decreasing output over time, which could really bite if you didn't know about it, but it can be designed for by initially having more light output than you need and your PAR adjusted over time by the dimming circuit.

The testing proposed, assuming that all other factors could be kept perfectly stable, would just identify that there is a problem, but would it be able to identify what that problem is? Was it a defective PAR meter? PAR meters can't reliably or accurately measure LEDs? The corals used in the test thrive/die differently depending on wavelength of provided light? .... I don't see how we could get any answers, only questions that we don't have the equipment or expertise with which to determine answers.

The LED components in the original article have published values for minimum life (50,000 hours with appropriate heat sinks; all high power LEDs should have heat sinks) and minimum output (minimum 70% at 50,000 hours). What needs testing is the LED components for life, output drop and colour shift over time at various levels of service. I doubt anyone here has the equipment for accelerated service testing. I sure don't (but I wish..., no I don't - I don't have enough time). I have a friend who is expert in this and we're overdue to catch up. He's currently doing thermal management vs. service life with UHF-RF chips and IR sensors, but I'll see what he knows about the high power LEDs and how reliable such published data is and how it relates to real-life, when I catch up with him.

However, what we can do is measure PAR at given depths when a given LED is at a given height above the water, driven by a given drive method at a measurable current. As each does their own testing, the results can be published here. If there are multiple LEDs in an array, then publish what the array spacing is as well. Over time, we'll gather a pool of useful data that increases the reliability of our own builds/designs.