Improved Color LED thread

Discussion in 'DIY' started by rygh, Oct 4, 2010.

  1. rygh

    rygh Supporting Member

    I am gradually grinding through the tasks for my new 250 gallon tank.
    Next step : Lighting, so time for an LED thread.
    Note: This thread is not meant as a basic how-to on LEDs.
    I have experience on that, and will happily help others as needed, but there are a
    lot of other good basic threads, here and in RC.
    This thread is about taking LEDs to the next level. And maybe failing - who knows.


    I want to start with a short story:
    I built a small 2 x XPG cool White + 1 x XPE Blue + 1 x XPE Royal Blue LED cluster on a heat sink.
    I pushed my current tank lights over a bit, and looked at it.
    Results were GREAT!
    A few minutes later, I realized my mistake.
    So I turned OFF my other tank lights, and tried it again.
    Bummer - did not look nearly as good. A bit washed out. Pretty good, but far from great.
    So the combo was great, but alone, not so much.

    SO THE PREMISE : CURRENT LED SYSTEMS ARE ONLY "PRETTY GOOD"
    As I have personally experienced, and as I see on the net,
    the current standard build of Cool White + Royal Blue is really not bad at all,
    but very few approach my ideal of a high quality Metal Halide + T5 Actinics.
    I am hoping that can be fixed.

    ----------------------------------------------------------------------------------------------------

    # THE PROBLEM WITH COOL WHITE + ROYAL BLUE:
    This is the most common setup these days.
    And like mentioned above, the results are actually "Pretty Good".
    Plus you can pretty easily tweak the amount of blue to get the color temp you want.
    The thing is : It looks nice given a fairly uniform white target.
    So on bare live rock, can look great.
    The problem is : It is your eye fooling you. The spectrum is actually pretty bad.
    You have HUGE peak of blue at 450 nm.
    You have a bump of light from roughly 525 nm to 625 nm, green through orange.
    But you have almost zero reds, especially deep red.
    Plus, you miss Cyan, and the deeper violets.
    So while you can trick the eye into thinking it is white, it is actually far from it spectrally.
    The problem is if you have a red coral for example, and no red light, it will look almost black.
    Now there is red, just not that much, plus there is fluorescence, so you will see some reds.
    But in general, it will look washed out and greenish-blue.
    Also, chlorophyll need that missing red.


    # THE PROBLEM WITH EVEN SPACING:
    Again, this is the most common setup these days.
    A lot of systems, even professional, evenly space the LEDs, often alternating White / Blue.
    That is wrong. You really want to keep LEDs of different colors very tightly clustered.
    If you don't, you can get terrible color banding effects, especially in shadows or with a rough water surface.
    What happens is that you will have some regions where blue dominates, and some where white dominates.
    For example, directly under a blue, next to a high rock that blocks the nearby white.
    By itself, normally the color is really not that different, since light bounces around a lot.
    But your eye is really good at detecting subtle differences in a pattern.
    As a result, you get a mottled effect, and really notice the odd bluish areas or odd yellowish areas.
    Note that optics make clustering hard, and make the problem worse.


    # THE PROBLEM WITH NEUTRAL/WARM WHITE:
    So a common "fix" you hear about is to add a few neutral and warm whites.
    This is done usually after looking at the data sheet, and seeing how it looks shifted a bit toward the red.
    The thing is, the data sheet curve is deceptive. Oddly, the best thing to do is to look through a cheap spectroscope.
    What really happens on NW/WW is that the orange and yellow end is raised considerably, and the blue is dropped.
    Yes, there is a bit more red, but not enough to fix the root of the problem.
    So you move the energy just to where you don't want it - yellow.
    Then you add a lot more blue, so you end up with largely blue + yellow, which is even more washed out looking.
    To make it worse, people add a few NW/WW here and there, and they end up widely separated,
    which in turn cause color banding problems.


    ----------------------------------------------------------------------------------------------------

    ### BACK TO THE BASICS ####

    The big problem in my mind is having multiple issues and targets as follows:

    1) Match coral spectrum. (PAR)
    Clearly, chlorophyll need blue and some red. But it is more complex
    since there are many more cartinoids, and coral seems to be adapted more toward the blue.
    Best seems to be : Lots of blue, a bit extra red, and moderately low flat values in between.

    2) Match the human eye. (Lumen / CRI)
    It needs to look good of course. But again, it is not that simple, since the
    eye is more sensitive to certain wavelengths. And to make it worse, it does not match PAR.
    Best is a fairly flat spectrum, perhaps with a bit more blue for color temp personal taste.

    3) Reduce power.
    A big point of going with LEDs is to reduce power.
    But that means more than just replacing it. You need to pick efficient LEDs, and
    you need to not waste a ton of power in unnecessary wavelengths.

    4) Reduce the yellow look.
    Fish tank water tends to get a yellow tint. Dead algae, and other bio.
    Plus, most of the pleasing colors you want to "pop" are red + blue + green.
    As such, you want to reduce the yellow/orange part of the spectrum.

    5) Fluorescence.
    A lot of what we talk about when making colors "pop", is actually fluorescence.
    Hence actinics.
    It is key to note the science here. : A shorter wavelength is emitted as a lower energy longer wavelength.
    Because of that, we generally want as short/high energy as possible. Royal blue works, but UV works better.

    6) Reasonable cost/build.
    We are limited by the manufacturer and what we can buy.
    Plus we are limited by space, drivers, soldering, and more.
    A perfect system may not be that practical.


    ----------------------------------------------------------------------------------------------------

    # BACKGROUND:
    My current tank is a mixture of LED/T5, and my turf scrubbers are 100% LED, all DIY.
    Plus I have done a few DIY LED lights around the home. So I have a fair bit of experience.
    But I have never done a 100% LED main tank before.
    New Tank is about 250 Gallons. 88"L x 30"W x 24"D.
    I plan mostly softies / LPS, but center raised section will be SPS.

    # MY PLAN ???
    Still working on that.
    I have a lot of LEDs to play with.
    I have a Lux meter, and a cheap spectroscope.
    Lots of soldering and tinkering to do. This is where the thread comes in.
    Any and all advice is very welcome!!

    The current theory:
    Have a cluster that is partly the normal Cool White + Royal blue.
    But also add a set of smaller 5 mm LEDs, that "tweak" the spectrum, particularly
    in the deep red and deep violet.
    The cluster will be mounted with LEDs clumped as tight as possible, on a single large heat sink

    # LED CLUSTER
    - 1 x XP-G Cool White @ 1200 mA
    - 1 x XP-E Blue @ 700 mA
    - 1 x XP-E Royal Blue @ 700 mA
    - 1 x 5 mm Deep Red 660nm @ 25 mA
    - 1 x 5 mm Red 630nm @ 25 mA
    - 1 x 5 mm Cyan 505nm @ 25 mA
    - 1 x 5 mm Violet 420nm @ 25 mA
    - 1 x 5 mm Semi-UV 400nm @ 25 mA
    CW = Cree XP-G, 6500K+, buy special bin toward cool
    RB = Cree XP-E, Royal blue, roughly 460 nm
    BL = Cree XP-E, Standard blue, roughly 475 nm
    Probably need 40 of these cells.
    CW/RB/Tweak will all be independently controllable, so I can play with the color a bit.
    But not dimmable on the fly.
     
  2. screebo

    screebo Guest

    My entrance into LED's is very cautious. I've simply added a strip of Reef Brite blues to my existing lighting which is abundant. I love the way the all blues look when all other lights are out. I'm also aware that the suplement of blue (actinic) light is very usable to coral. The coral glow that is afforded by the LED's is fabulous. I really, really, really want those of you who are experimenting with LED only tanks to be very, very,very successful. I'd LOVE to go strictly LED's at somepoint but I just have yet to hear of long term success using LED's only in a mixed reef. I wait and watch carefully and thank you "pioneers" who are brave enough to stop using traditional lighting and move on to LED only. Keep up the good work, I'm not brave enough to make a transition at this point. :beer:
     
  3. rygh

    rygh Supporting Member

    Being cautious is a very good idea. It is a pretty big up front expense, even DIY.
    This is why I do not rush around advising people to switch, even though I am all for LED.

    Note that there are quite a few long term successes with LED on RC. But certainly outweighed by the failures.
    The major problems seem to be when people grossly under or over estimate the amount of light needed, then shock/starve
    and kill the coral with the different intensity.
    The next major failure is when people simply confuse the new look with the health of the coral.
     
  4. sfsuphysics

    sfsuphysics Supporting Member

    Grrrr I fricking hate this message board, I had a huge post and did one of my type too quickly and hit some "special" code that makes the browser go back and it was all gone, other boards if you go forward again the typed info is still there.

    Anyways to summarize.

    Nice read :)

    Halides + Actinics vs LEDs : Problem is LEDs typically are used to replace halides, not halides plus actinics. So you'd need dedicated blue strips in place as well as your clustered LED setup, and even then most LEDs don't hit that violet terribly well.

    Reds are overlooked but most halides are a bit lacking in that department as well, not sure it really makes that much of a difference, coral chlorophyll will grow with blue light, however the algae we don't want not terribly well at blue.

    I've put my 12 LED setup (7-5 ratio of RB to CW) over my tiny RBTA tank, with a couple zoanthids, and some leathers just to see how well it does, it looks ok for now. And yes the anemones are quite red.
     
  5. rygh

    rygh Supporting Member

    Agree with the problem.
    But tend to disagree with wanting blue strips. I would argue it is better to add more blue to the cluster.
    With any different colors that are not clustered, you will get color banding effects. Likely minor with a lot of strips, but will be there.
    Yes, finding a good violet is a pain, unless you use 5 mm ones, but then you need a lot more.


    I can definitely attest to algae not liking pure blue. I tested my RB/CW light on my algae turf scrubber for fun. Definite growth loss.
    Are you sure some of the red in the anemones is not from florescence?
    And is it more orange-red, or deep red?
    When I look at it, I sure see deep reds missing.
     
  6. treylane

    treylane Guest

    I've been running a "pink" halide bulb for the last year or so. In the SPS and softies arena (less so with LPS it seems) there are tons of corals with pinks and reds - purple validas, pink birdsnests, pink millis, pink shrooms and palys, and they stand out and look awesome on the pink side of the tank, while they just look kinda pale and silly on the 'traditional' blue-halide side of the tank. As I've come to really like the pink look, I notice that reds are conspicuously absent from the LED fixtures arena, and I wouldn't buy a one without a decent amount of red in it.
     
  7. rygh

    rygh Supporting Member

    Some tinkering results:

    Here is an interesting combo of:
    Red + Deep Red + Violet 420 + UV
    The UV is the blue looking one. Seems to be visible blue + invisible UV.
    [table]

    [img width=144 height=112]http://lh6.ggpht.com/_eaxg2H-5-nk/TK1isX0TYXI/AAAAAAAAAHs/dbccQ460ZaE/s144/5led_combo.jpg[/img]



    From Aquarium_Release



    The tinkering bench.
    [table]

    [img width=144 height=108]http://lh4.ggpht.com/_eaxg2H-5-nk/TK1irYlFJTI/AAAAAAAAAHY/UJMU6pSA1yk/s144/5led1.jpg[/img]



    From Aquarium_Release



    UV Causing security strip to fluoresce
    [table]

    [img width=144 height=124]http://lh4.ggpht.com/_eaxg2H-5-nk/TK1isMKiL2I/AAAAAAAAAHo/mZboM64_iSg/s144/5led_uv.jpg[/img]



    From Aquarium_Release

    [/table]

    Cyan : Very little green. Basically blue, so pointless.


    [img width=144 height=93]http://lh5.ggpht.com/_eaxg2H-5-nk/TK1irvQ98JI/AAAAAAAAAHg/GngOOWt2gXY/s144/5led_cyan.jpg[/img]



    From Aquarium_Release

    [/table]

    Violet (420 nm and 440 nm were completely identical.)


    [img width=144 height=79]http://lh4.ggpht.com/_eaxg2H-5-nk/TK1ir84jfaI/AAAAAAAAAHk/kpYRNWd-CQ0/s144/5led_purple.jpg[/img]



    From Aquarium_Release

    [/table]
    Not surprisingly, camera having troubles.
     
  8. GreshamH

    GreshamH Guest

    Have you read Dana's AAOL. article on corals and the red spectrum?
     
  9. rygh

    rygh Supporting Member

    No, I had not, very interesting reading!! Thanks.
    Also, it had a pointer to a similar older article with some good info as well.

    For others interested:
    http://www.advancedaquarist.com/2008/12/aafeature1
    http://advancedaquarist.com/issues/june2004/feature.htm

    Seems the theory is that corals do not possess a pigment to properly regulate photo-reactive red light energy.

    BUT : While useful, there is not enough data, and the experiment is not correct for a true red light conclusion.
    The experiment needed, that nobody seems to have done:
    Have sets of identical setups, same "good" lighting, same corals, etc. Where you ONLY add red LED lighting to one.
    And of course, you need to test both deep water and shallow water corals. Results almost have to be different.


    -----

    I still want to add some red, because if nothing else, having it look good is more important to me than optimal
    coral growth. And quite a few commonly used MH/T5 bulbs have a lot of red, so it can't be all that bad.
    But certainly don't want to over-do it. - Although using 5 mm LEDs, pretty hard to over-do it anyway.
    I also feel that the combo of red + deep red, instead of just one type, might be ideal as well, since it lowers the
    chance of overloading one frequency.
     
  10. rygh

    rygh Supporting Member

    Well, I am making fantastic progress now.

    ... Wait for it .....

    I now have ...... The WIRE.
    Oooh. Aaahh.

    Yeah, with all the different components, I need a fair number of colors.
    And I actually wasted the money on real Marine Wire for the mains.
    Pre-tinned copper, with an upgraded coating. A tiny bit more, but no corrosion issues.
    Even pre-tinned for the small wire, but that makes soldering easier, so a side benefit.

    I am still tinkering with the colors, and not really getting much closer.
    Looks like the new Rebel Royal blue is a nice deep Violet, better than Cree.
    Considering Rebel ANSI-4K for the mains (Neutral White-ish) But hard to ignore the 20% efficiency difference.
    Just really not happy with anything.
    Maybe I will give up and go T5. Nah. Keep trying.


    [table]

    [img width=144 height=88]http://lh3.ggpht.com/_eaxg2H-5-nk/TLagOpT-nAI/AAAAAAAAAIE/qGSRQ5EeAFw/s144/wires.jpg[/img]



    From Aquarium_Release

    [/table]
     
  11. yardartist

    yardartist Guest

    The wire sounds great and worth it. I had not soldered much and that was a difficult portion given that excess heat at installation can burn out the lamps. I have wires popping off when bumped. Should have completed the construction of the housing first so the heat sink just set into it in one motion. All the moving around now is brutal on the handwork completed.
     
  12. sfsuphysics

    sfsuphysics Supporting Member

    nice call on the wire, too many people overdo it with wire that's too thick, 22 gauge is perfectly fine.
     
  13. GreshamH

    GreshamH Guest

    "Seems" is the perfect choice as some one has done it. A kid did it for his science project and wrote it up. Been a few years so it'll take some digging on my part to find it. He only did his with zoanthids though.
     
  14. rygh

    rygh Supporting Member

    Yep.
    (FYI for others): Big thick wire, beyond being a hassle, can actually cause problems.
    The thermal expansion of copper is pretty big, so a thick wire can generate a fair amount of force, possibly
    ripping up those thin bonding pads on the LEDs over the years.

    Another tip to the wire : Get 105C temp rated wire. It helps a lot when soldering, since with cheap wire,
    you can easily melt the insulation way back.
     
  15. rygh

    rygh Supporting Member

    I just ordered a set of 10 LED drivers.
    I need quite a few more, but I want to test these first.
    A whole $1.67 each, with free shipping. DealExtreme
    Based on the PT4115E chip. Basically some super-cheap flashlight drivers, but they have a good quality chip in them.
    At 24VDC source, 6 LEDs per string, I expect about 95% efficiency.
    They will require a bit of tweaking though. Remove a rectifier, add a better output cap.
    Surface mount, but still pretty simple stuff.
    As a bonus, it is pretty easy to change the sense resistor, and adjust current, from 500-1200mA.
    The dimming pin might be tricky, but I was not planning on using that.
     
  16. sfsuphysics

    sfsuphysics Supporting Member

    what's the output amperage of those drivers? Sure it's not 6 of the 5mm leds it can power? That seems awfully cheap for a driver.
     
  17. rygh

    rygh Supporting Member

    It can drive up to 1200 mA. Max of 25V, so pretty easily drive 6 x 3 Watt.
    And yes, it is a buck converter, not linear.

    It is super cheap, and the components are cheap Chinese parts.
    But that said, the price of the meanwell and other drivers is outrageously high.
    All of the components are single chip. All you need is that, a couple of caps, inductor, and a resistor,
    and maybe a protection diode.
    They have no business charging $20.

    http://crpowtech.com/products_content.asp?id=213&ClassId=71&SortPath=0,71,
     
  18. rygh

    rygh Supporting Member

    Well, the main power supplies are here.
    The left 2 are 5V and 12V, really for micro-controller, fans, small LEDs, and misc electronics.
    The main one on the right is the 480 Watt, 24V supply for the main LEDs.
    My current target is about 400 Watts total for main LEDs. Sure hope that is enough. Should be.
    Pretty high end, high efficiency supply.
    93% efficiency. 0.99 PFC.
    Considering that cheapo supplies are 80%, that 13% = 50 Watts wasted.

    [table]

    [img width=144 height=98]http://lh5.ggpht.com/_eaxg2H-5-nk/TL5jDLsSN2I/AAAAAAAAAI8/aIzs_xEP_DI/s144/pwr_supply.jpg[/img]



    From Aquarium_Release

    [/table]
     
  19. rygh

    rygh Supporting Member

    After a fair amount of experimentation, I believe I now have a final plan.
    I decided not to do my own drivers. Too much like my day job to be fun. Even though I have a few parts already.
    I ended up going with RECOM drivers, due to higher efficiency and longevity specs.
    For tweaking the color:
    First, I used a combination of Cree and Luxeon Royal Blue. The color seems different enough to be worth it.
    Second, I used a small set of neutral whites, to get a wider spec. Rebels, since they seem to have a flatter spectrum.
    Third, I used a small set of Reds, driven VERY VERY weakly, to get a hint of the reds back. I avoided deep red,
    because it probably helps algae more than your eyes.
    My current white/blue ratio is roughly 1:1 (based on output, not count), but it is very easy to drop down the white independently.
    My personal color temp preference is more in the 12K range, not 20K.

    LEDS:

    36 x Cluster A : Main white
    Cree XP-G WHT-L1-1B0-R5-0-01 : Cool white, specially binned, at 1200 mA
    Cree XP-E Royal Blue, at 700 mA, 450-465 nm
    Luxeon Rebel Royal Blue, at 700 mA, 447 nm
    Vf = 3.4, 3.4, 3.2 = 4.1W, 2.4W, 2.2W = 8.7W
    There will be 36 of these, with higher density toward center of tank = 313W

    8 x Cluster B : Color Tweak
    Luxeon Rebel Neutral White, at 700 mA, ansi 5000K
    Cree XP-E Royal Blue, at 700 mA, 450-465 nm
    Cree XP-E Standard Blue, at 700 mA, 465-485 nm
    Cree XP-E WEAK Standard Red, at < 350 mA, 620-630 nm.
    Vf = 3.0, 3.4, 3.4, 2.1 = 2.1W, 2.4W, 2.4W, 0.7W = 7.6W
    There will be 8 of these, evenly distributed = 31W

    Totals:
    Roughly 44 whites, and 88 blues, but given efficiency and drive strength,
    that is about a 1:1 ratio.
    Total LEDs = 140

    LED DRIVERS:

    6 x Main white driver
    Use RCD-24-1.20/W/X1 (/X3 ok)
    Recom LightLine, 1200 mA, wired, IP67 waterproof, analog dimming
    Drives 6 LEDs per string. 24V -> about 21V
    Efficiency at 1200mA, 6 LED, 24V about 95%

    16 x Main blue driver (also the few neutral whites)
    Use RCD-24-0.70/W/X1 (/X3 ok)
    Recom LightLine, 700 mA, wired, IP67 waterproof, analog dimming
    Drives 6 LEDs per string. 24V -> about 21V
    Efficiency at 700mA, 6 LED, 24V about 94%

    2 x Small red driver
    Use RCD-24-0.35/W/X1 (/X3 ok)
    Recom LightLine, 350 mA, wired, IP67 waterproof, analog dimming
    Drives 4 LEDs per string. 24V -> about 8.4V
    Efficiency at 350mA, 4 LED, 24V about 82%

    1 x Main supply
    Use MeanWell SDR-480-24
    Outputs 24V at 0-20A
    Efficiency = 94%. PFC=0.99. 150% peak load.

    Notes:

    No lenses. There will be a 1/8 acrylic sheet between LEDs and water, as part of light fixture.
    LEDs will be about 2 inches about water.
    Each cluster is mounted on a large 2x4x2 heat sink.
    Each light set will have quiet 12V fans. Fairly low CFM.
    Dimming will be controlled by Arduino microcontroller.
     
  20. Coral reefer

    Coral reefer President

    Yes please! Can't wait to see
     

Share This Page