550W LED DIY for 380 gal (34" x 96")

[karimwassef]

ok. I don't know if these images will show up, but I can't include my analysis file on reefcentral cause it's too big.

Here's my LED 550W design:

<a href="http://s1062.beta.photobucket.com/user/karimwassef/library/" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/led1_zps28c22ffe.jpg" border="0" alt="Photobucket"/></a>

<a href="http://s1062.beta.photobucket.com/user/karimwassef/library/" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/led2_zpsc569a330.jpg" border="0" alt="Photobucket"/></a>

<a href="http://s1062.beta.photobucket.com/user/karimwassef/library/" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/led3_zpse4dbceec.jpg" border="0" alt="Photobucket"/></a>

It has 8 strings at 40V and 94 LEDs covering a 34"x96" tank. It's smaller than the tank top but all in a 3.5" x 3.5" array (4 x 24 grid with two missing for simplicity).

Mechanically, the LEDs heatsunk to a 1/4" stainless steel sheet (drilled) with another 1/8" sheet over it (wire and solder joints covered) and u-frames on top.

3 strings are XM-L whites (12 LEDs) with 3A each.
2 strings are XP-E blues and greens (11 LEDs) at 1A each.
2 strings are XT-E royal blues (12 LEDs) at 1.5A each.
1 string UV (12 LEDs) at 0.7A each.

The forward drops add up to 40V each and each string is the same current through all the LEDs.

I've gone through it quickly, but might have made a goof somewhere - so here's to all the LED experts out there for help.

[kcress]

Comments:

Stainless steel is absolute garbage for heat conduction. I would expect you to loose all your strings if you are indeed using the stainless as your heat sink.

Next. You should definitely NOT be pushing your LEDs to the factory max. (if you are) I see many failures when people push to the rating and get some really bad problems with the lenses laminating and the dies just plan cooking.

Take a tip from commercial streetlight builders. They almost never run XPGs at even 700mA. Most designs are based on 350mA.

[Dave Thebrewguy]

+1 on stainless steel. Ever notice that any good piece of cookware has a copper or aluminum disk in the bottom? Its there for a good reason. I'd also suggest you use violet (~420nm) instead of uv (<400nm), violet is important for coral health and growth, UV is usually considered detrimental.

[karimwassef]

I said SS, but meant Al. shouldn't type after midnight...

Aluminum is better and copper is best. Sure.
I was worried about pitting from the spray but I will paint the bottom face reflective white. SS is too heavy too.

They're all dimmable, but the max rating is 550W. I am running the numbers to size the power circuits. Running at 50% to 80% is what I expect to do in actual operation.

Here's the "UV" I was considering: www.rapidled.com/violet-uv-led/

In terms of design:
Enough light for 34x96x27? (27" deep)
Balanced light (cool vs. warm)?
AC to 40V bus?
12 per string look manageable?
3A 10W XM-L scare anyone?
Overall cooling with dual plates and 4u channels look ok?

Some of these are a matter of preference. I'm concerned with anything that would prove too wasteful (time, money) to address later.

[karimwassef]

Also: the string separation was driven by current and voltage, but the intent is to dim/time the white/bluegreen/royal blue/UV to transition through an actinic state for morning and evening... Each coming on gradually over 30mins with a 30min dwell for a 3.5 hour period to full light.

V
V+RB
V+RB+BG
V+RB+BG+W
....
Reverse for off

I'm old school and love the classic actinic reef look at night.

[Dave Thebrewguy]

I agree with tour thoughts on the actinic look and I'm planning something similar. I'm starting from the idea that light has to travel further through water when the sun is low in the sky and that RB is filtered out by water less than other colors. My RB will come on first and go off last.
Violet and blue will come on next followed by whites and reds. These will, of course, turn off in reverse order.
BTW, those LEDs are indeed violet. I have no idea why they are labeled UV.

[karimwassef]

RB before Violet? I was thinking that the highest frequency (shortest wavelength) would appear first and maybe even phasing them into each other for a smoother transition...

But more to the total fixture design... Any feedback there?

[Dave Thebrewguy]

I got frustrated trying to plan my layout in a grid so I gave up and switched to a "clustered" design. As a result, I refrain from commenting on the layouts that others have worked on. I just don't feel qualified to comment.
I like your choice of colors for the most part (I'm not familiar with the cree blue green but its probably close to my Philips cool blues.). You might want to add more violets, my "actinic channel" uses equal numbers of violet (Epiled from Steve's, 417nm), Philips RB (440-450 nm), Cree RB (450-455nm) and Philips CB (470-480nm). I wish I'd used more violets, and maybe a couple cyan, instead of as many CB.

[karimwassef]

Got a link to the violet you use? Got some tank pics?

[Dave Thebrewguy]

My violets are the ones from Steve's, I suspect they are the same ones. I find the link policy here confusing, I'm never sure what is allowed and what is not, so I generally don't post links but it should be easy to find.
No tank shots yet either, still a work in progress that I'll be glad to share as soon as I think the tank is presentable. Hopefully just another couple weeks.

[karimwassef]

Ok. Please pm me?

[karimwassef]

so the UV is smallest in power, but this is due to the lower power per LED (2W UV vs. 10W for the white) and the count (12 UV vs. 36 for the white).

string white 3 strings @ 112W each = 62%
string bluegreen 2 strings @ 36W each = 13%
string royalblue 2 strings @ 56W each = 21%
string ultraviolet 1 string @ 26W = 5%

total 8 strings @ 546W

The UV is really more to bring out the fluorescence associated with actinic lighting and help with the coloration due to lower wavelength exposure. How much would you bump the violet (10%) and at the penalty to which string (BG?)

[kcress]

In terms of design:
Enough light for 34x96x27? (27" deep)
Balanced light (cool vs. warm)?
AC to 40V bus?
12 per string look manageable?
3A 10W XM-L scare anyone?
Overall cooling with dual plates and 4u channels look ok?

Some of these are a matter of preference. I'm concerned with anything that would prove too wasteful (time, money) to address later.

Feedback...

Ummm, I always rather people avoid terms like "3W" or "5W" or "10W" LEDs. No one uses the watts in any design aspect. It is merely a "class" sometimes used by manufacturers. What's informative is the current you intend to run thru what specific LED.

No one can tell you whether your plan is good for a tank because you've gone completely off-road. Off-the-reservation. XMLs are hard to match up with XTEs or XPGs in an aquarium setting. You are on your own. This doesn't mean it won't work, just that we can't say, "it will be fine". You're going to have to see for yourself. I would consul you to put emphasis on changeability of your design so you can easily change your build if you fine you need to.

I will say this. Many builders call XMLs "reef killers" because they put so much light into one area that nothing can resist getting photonically cooked.

Again this doesn't mean someone couldn't use XMLs successfully because they can underdrive them or put them up high or put a shield in front of them or run them for short periods.

Putting 3A to them? Good luck. You'll need it. :p That's in the form of exemplary heat sinking with air flow guarantees.

[karimwassef]

The maximum rating is there to size the power supplies, not under normal operation. The highest I expect to run the XM-L strings is probably 2A. More likely at 1.5A.

I see the XM-L as stand ins for 3 LEDs each, so one XM-L at 1.5A would compare to 3 XP-E at 0.5A each (more focused, sure). The equivalent output should be comparable.

The light setup is 550W max but usage will likely run at 300W. The idea is also to play with the intensity of the strings to get to a preferred balance: maybe run the UV at 80% ~ 560mA while the whites are at 50% ~ 1500mA.

The only coupled strings are blue & green.

The alternative is 2x 400W MH or 3x 250W MH.

[karimwassef]

Thinking more on this, I will split the cool white and warm white strings for more control.

[karimwassef]

http://s1062.photobucket.com/albums/t496/karimwassef/?action=view&current=xm-l_zps0ddb6f54.jpg&evt=user_media_share

[Danny_B]

What color are you going for? You have 2:3 royal blue to white ratio...not to mention the whites are xm-l., the whites are going to need to be dimmed so far down to give a 14-20k look. I have 1:1 rb cw and with both at 1000ma, the tank is white, very white, and the light is flat, no flourence or character. I can't imagine anyone wanting to run anymore white than that. I would up your rb count considerably to at least 1:1 at the bare minimum, id do more considering the xml.

[karimwassef]

I have 62% white and the rest (38%) are royal blue, blue, green, violet.

The white is actually made up of 41% cool white and 21% warm white.

So...

21% warm white (2600-3700K), 41% cool white (5000-8300K) and 38% blues (20K?+)

[karimwassef]

<a href="http://s1062.beta.photobucket.com/user/karimwassef/library/" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/xm-l_zps0ddb6f54.jpg" border="0" alt="Photobucket"/></a>

Here's the image for the cree xm-l lumens vs. current and power

[karimwassef]

double post -

[rrasco]

I'm running 2:1 RB to Whites (4500, 6500, 10K) and my blues are set to run at 50% with whites at 45%. Whites are much, much brighter than blues as far as visible wavelengths go.

I would also recommend you build independent units. I planned a 6' fixture, then backed off that and built three 20" fixtures. Glad I did because my 210 broke and now I have a 4' 105.