Minimalistic multichip DIY LED build

[Lassef]

Ok, after reading like 20+ pages of this thread, I've decided to ask a few questions. I am planning on building a DIY LED fixture for my 15 gallon (24L x 12H x 12W) tank out of 3w LEDs and now i'm interested in these 10w LEDs. How many LEDs and what type/colour would I need for growing hard and soft corals? If possible could someone PM me what drivers, LEDs and whatever else I would need?

I was looking at maybe 2 of these
http://www.ebay.com/itm/New-40mil-1...ltDomain_0&hash=item3f17422c3f#ht_2746wt_1026

and three of these
http://www.ebay.com/itm/10W-10000K-...ultDomain_0&hash=item3f10351676#ht_1636wt_906

I don't know what i'll need for a driver but I want to be able to dim (either the driver itself or a pot?) the 10k and actinic separately (For the amperage do I need to add all the ma's to get a total of 5amps? and get a 3amp and 2amp driver or do I just need to get a 1amp driver for each set?). Also what power adapter would I need to plug it in the wall? or can I just use a simple 12v computer power cord? Also what lenses would I need?

Instead of the hybrid´s I would use two of these instead. And I prefer these white.

Put the two blue in a serie and you can use this driver ELN-30-24

Choise D or P type (D if your controll unit use 1-10, P type if it use PWM)
Adjust the driver to leave 1000 mA

Put the three white chip together in a serie (daisy chain) and use this driver ELN-60-48

Choise D or P type (D if your controll unit use 1-10, P type if it use PWM)
Adjust the driver to leave 1000 mA.

Sincerely Lasse

 

[Lassef]

Some shots of the engineering sample they sent me. Unpowered:
https://plus.google.com/photos/1030...29709073018401/5772929705352496930?banner=pwa

Powered up. Close up:
https://plus.google.com/photos/1030...29709073018401/5772929726772956626?banner=pwa

Actual spectrum analysis of my engineering sample from ac-rc:
https://plus.google.com/photos/1030...29709073018401/5772931484969085826?banner=pwa

Explaination of data provided by ac-rc:
https://plus.google.com/photos/1030...29709073018401/5772932106219953890?banner=pwa

Thank you a lot

Will he have this in his shop later on?

Sincerely Lasse

 

[Danielism]

Alright, thanks. So five total 10w LEDs are enough? to light a 15 gallon?

 

[sfsuphysics]

Some shots of the engineering sample they sent me. Unpowered:
https://plus.google.com/photos/1030...29709073018401/5772929705352496930?banner=pwa

I notice that the 5th row down has a different junction structure to the chips, these are the 420nm ones perhaps? (The picture with them on makes them all look identical.

Also the spectrograph doesn't seem to show that 445nm peak, only the 455 peak (unless my eyes are just too tired to see straight ).


As to the multichip, if you're going to go full out with distinctive channels my thoughts on what there should be

1) 20 - whites ("10000k"), 5 - red
2) 20 - 420nm
3) 25 "royal blue" 455nm
4) 25 "blue" 445nm

Heavy blue to be certain, but the idea is you can tune the chip as you see fit, and most people have bluer ones on most of the time, with the whites being the "peak sun" time period, so you could always tune them as you like, my experience with Crees is that even at 2:1 ratio of RB to CW the white really dominates from a visual stand point.

You mentioned 100w, but have 100 chip examples.. so would this be true 100W without running it to the max?

 

[ronreef]

I notice that the 5th row down has a different junction structure to the chips, these are the 420nm ones perhaps? (The picture with them on makes them all look identical.

Also the spectrograph doesn't seem to show that 445nm peak, only the 455 peak (unless my eyes are just too tired to see straight ).


As to the multichip, if you're going to go full out with distinctive channels my thoughts on what there should be

1) 20 - whites ("10000k"), 5 - red
2) 20 - 420nm
3) 25 "royal blue" 455nm
4) 25 "blue" 445nm

Heavy blue to be certain, but the idea is you can tune the chip as you see fit, and most people have bluer ones on most of the time, with the whites being the "peak sun" time period, so you could always tune them as you like, my experience with Crees is that even at 2:1 ratio of RB to CW the white really dominates from a visual stand point.

You mentioned 100w, but have 100 chip examples.. so would this be true 100W without running it to the max?

Your color input doesn't add up to 100

Based on experience, the spectrum analysis shows an average of the two peaks. The resolution of the sweep is not set to show each peak separately. In short, it's likely a spectrum analyzer setup issue more than anything else.

Keep in mind the 10000K and 15000K contain blue peaks in their spectral output. I personally, like more neutral look, maybe slightly tinted blue than any other look.

Yes, this would be true 100W running each 10 led series leg with 350ma @ ~36V.

 

[tomservo]

sfsu: I would hope that it would be a 100w built using 100 45 mil chips; in essence a 252w chip if you drive it to the max.

I'd be in for one, the first version sounds appealing. but either one would be nice. I could stand the red left out or perhaps a lower K subbed in.

 

[marspeed]

are all those legs connected to two common buss bars that's what it looks like in the picture ?

 

[ronreef]

are all those legs connected to two common buss bars that's what it looks like in the picture ?

One rail is power. The other rail is ground.

 

[gtfcniles]

100W (5 leg)
1) 20 x 445nm
2) 20 x 10000K
3) 20 x 420nm
4) 20 x 455nm
5) 20 x 15000K


I would prefer this, would also need 6 of them, but would want one to test before shelling out to find there no good.

 

[ronreef]

Another option:

100W (5 leg)
1) 20 x 445nm
2) 20 x 455nm
3) 20 x 420nm
4) 20 x 10000K
5) 20 x 15000K

Changed the order. Put the blues and whites adjacent to each other. This allows both blues or both whites to be powered by only one driver by externally shorting the rails.

Each of the five rails can be independently controlled, but will be able to use less drivers by externally shorting rails.

 

[gtfcniles]

Ron, i would like to see a UV range in there maybe 5 or 10? Saying that it would be possible to run a 10w UV LED eitherside of the 100w, is he pricing these at the same price point as the other hybrids? If these are successful he would would get a lot more orders anyway I imagin he would continue production of them after he has paid the 1500 fee for setting up a new template. So just put me down for one depending on the price? I'll email him directly as I have dealt with him quite a bit over the past year, no point sending it to the US for it to come to the UK.

Wouldn't it be better to space them out blue white violet white blue?

 

[BeanAnimal]

I would do away with 10 of the CW and add a row of 660, total 6 channels.

100W (6 leg)
1) 20 x 445nm
2) 20 x 455nm
3) 20 x 420nm
4) 20 x 10000K
5) 10 x 15000K
6) 10 x 660nm

 

[tronyony]

Ordered my LED's, drivers and some misc stuff

LED's:

16000k x 1
14000k x 2
10000k x 2
455nm x 6
445nm x 4
420nm x 2

Drivers:

HLG-80H-48B x 4

Fans:

50MM x 6



Thanks again,
Tony

 

[tomservo]

I don't like the idea of grouping the colors together; I think it's important to mix the blue/white/etc as much as possible in the multichips, or we'll get the striping effect as seen in lassef's pics of the 100w hybrid. Only you'd have white on one side and blue on the other.

Tying non-adjacent channels together is still trivial; a small piece of wire would attend to the job quickly. My concern would be, with different bins and colors, there will be different forward voltages preventing you from linking them in the first place unless the voltage binning is superb. I expect tying the whites and one of the blues onto one channel will be no problem as they will all be the same blue chips, but the others?

An order more like:

1) 445nm
2) 10000k
3) 420nm
4) 15000k
5) 455nm

Or with the whites on the outside, I'm not certain which would be best but my gut feeling is mixing them like this would provide more even tones across the output. If the center were blue/violet/blue then I think you'd have white stripes on the outside.

I'm not entirely against the idea of including red, but I'm not sure if it's possible to add it and actually have it be pleasing. Perhaps a better solution would be a different mix of whites, maybe a lower kelvin? It's probably unreasonable to hope for a custom phosphor blend with a bit more emphasis on red. It's a hopeless dream, but someday I'd like to see something like the sharp zenigata phosphor blend on one of these - they have substantial output past 700nm!

 

[Lassef]

Another option:

100W (5 leg)
1) 20 x 445nm
2) 20 x 455nm
3) 20 x 420nm
4) 20 x 10000K
5) 20 x 15000K

Changed the order. Put the blues and whites adjacent to each other. This allows both blues or both whites to be powered by only one driver by externally shorting the rails.

Each of the five rails can be independently controlled, but will be able to use less drivers by externally shorting rails.

This seems to me as a major breakthrough. Could this been done - I´m sure we will have the best out of two worlds.

In my experiences - we do not need either red´s or lower K if our aim is to light up a reeftank in a both biological and esthetic way. For a FW-tank I had take away one of the blues and put i a red leg.

In reality - for the eyes - you mix blue and white 50/50 because your eye will hardly not se the 420.

Its only needed 700 mA (FV ~36 V) to drive one of the five legs - this means that low cost drivers like Meanwell LPF-25D-36 works well and there is others that works as well.

Personally, I prefer to mix the blue and white legs evenly, but I can live with either solution.

Well done Ron!

Sincerely Lasse

 

[A Reef Scene]

The idea of a custom chip just for a reef is a great idea.

 

[Lassef]

Ordered my LED's, drivers and some misc stuff

LED's:

16000k x 1
14000k x 2
10000k x 2
455nm x 6
445nm x 4
420nm x 2

Drivers:

HLG-80H-48B x 4

Fans:

50MM x 6



Thanks again,
Tony

I´m sorry that I did not was able to show this link for you before you order. Look at the pictures in this thread - is in Swedish but pictures are international - you can get some ideas from this.

Sincerely Lasse

 

[gtfcniles]

Don't use red, it promotes algae growth where you don't want it, in your display! The most you would want is one or two maximum in my view, but not worth including.

1) 445nm
2) 10000k
3) 420nm
4) 15000k
5) 455nm


This would get my vote and should blend nicely.

 

[Lassef]

Thanks, Lasse.

I agree, I'm not convinced red is needed. The practical reason for not mixing blues and whites is control for the person that wants control of blue and white separately.

You missunderstood me - I meant mix white and blue legs like your first suggestion. To be able to control blue and white separately is important and the very meaning and genius of your design.

Sincerely Lasse

 

[tomservo]

Someone mentioned wanting UV, I'm assuming that is meant the 395-405nm range.. Just wanted to comment that it is not effective at producing fluorescence, I have a couple of 395s and you have to get it right by a coral (in this case neon green austaliensis monti) to produce much effect, where the 420nm has a MUCH more profound effect, IE the same as the UV produces at 2" but from 12"!

ronreef: Did you look at lassef's photos of his 100w hybrid testing, with lenses? (http://www.reefcentral.com/forums/showpost.php?p=20234869&postcount=676) If you look you can see the color seperation at the edges. While in this example it is not too prominent, consider if you have the white on one side and the blue on the other, rather than white between two blue, the seperation will be at least twice what it is in that photo. I maintain that bridging two non-adjacent bars is trivial, and we are assuming it will be possible (IE the voltage bins must match perfectly for different color bin chips)

Either way I'll be interested in several of these, I'd commit to 5. I'm assuming the price will be around $150 each given the level of 420nm we are wanting.