My take on the cluster

[bpaston]

Which type of MCPCB you plan to use?

 

I plan on going with copper at this point.

[DNK]

Material of MCPCB does not matter, but material of insulator have HUGE importance. Please familiarise carefully with "Challenge number two. Battling the heat" part of this article: www.advancedaquarist.com/2014/4/lighting

I would like to emphasize - thermal management for tight LED arrays is extremely important. Unfortunately, no one of manufacturer of LED arrays for DIYers not pay enough attention for this problem. Because of this, LEDs on that arrays may overheat and have loss radiation and decreasing of lifetime.

[bpaston]

Well after a couple days at the drawing board I had to remove two luxeon z leds from the original design and slightly increase the footprint. But I believe I am that much closer to moving into producing this, so above is the revision to account for adequate thermal cooling of the leds by using wider traces and spacing in between traces.

[DNK]

Dear bpaston, please be informed - nor wide traces, nor material of MCPCB not give significant increasing thermal cooling. Only direct contact of heatspreader of LED to material of MCPCB can solve this problem. Please familiarise http://www.advancedaquarist.com/2014/4/lighting#h1c

[evilc66]

You do realize that you aren't paying attention to the design? There are only 4 LEDs on that board that are capable of taking advantage of the SinkPAD design. The remaining 22 LEDs are Luxeon Z's which don't have a dedicated thermal pad. As a result, all thermal energy is conducted through the anode and cathode, and must be isolated for proper operation.

 

Don't keep referring to that article every time you want to make a point about thermal management. Linking to that article is basically linking to your commercial product, which is against this sites terms of use, unless you are a sponsor. Either become a sponsor, or stop parading that article around thinking you are better than everyone else. We are all aware of what SinkPAD can do, but you don't need to force it down our throats every time. It's not the only thermal solution (even though it is a good one), and not always the most practical one.

[DNK]

Yes, you right about thermal pad of Luxeon Z's LEDs. But in my opinion - Luxeon Z's should not be use in top-class solutions due to worst efficacy, similar (or even lower!) to cheap Chinese LEDs. For example. Best royal blue LEDs Luxeon T, that available on market, has radiometric power up to 1150mW for current 700mA and Luxeon Z - only up to 823mW. Voltage drop, respectively, 2.8V and 2.95V. So, efficacy will be, respectively: 1150/(700*2.8)*100=57.67% and 823/(700*2,95)*100=39.85%. Difference in efficacy Luxeon Z relative to Luxeon T will be (57,67-39,85)/39,85*100=44.72%. In other words, efficacy of Luxeon Z near to half lower that Luxeon T. It is still not enough? Please recall - all data for Luxeon T is for 85C (hot binning) and for Luxeon Z for 25C! It is, as the French say: "Berezina total" in other words, complete rout!

 

About referring to article. Could you show where in the article or at least in one of my post on this forum has a link to our store? Or may be you can indicate presence of other similar article in Internet? I sure - materials of this article may be very helpful for DYIers for constructing their own fixture. By the way, this article has a lot of know-how, that described absolutely free for all.

 

Excuse me, but your opinion about SinkPAD is carelessly, because good thermal management is primary goal of designing the LED fixture. Neglect of this issue can not lead to good results.

 

If the owner of this resource decides that my presence does not benefit the community, I will abide by his decision.

[bpaston]

Well after staring at this thing for countless hours and feeling like I may regret something once sending the final file in for production, I believe I have reached the end. Here is the final design of the MCPCB, giving the luxeon m's a little more breathing room and adding in two solder pads which I can tack the ez mate pico wire harness from the vero 10 to. With that being done now its time to finalize led choices as far as quantity and channel go. Oh and yes against all strong advice I will not be using a sinkpad for this project and will be using luxeon z's as inefficient as they may be. Maybe once complete with the project I will not like it and deem it a failure, but I would like to have one in hand before i can make that decision.

So as the MCPCB is sitting right now it has a 5 channel molex connector. I have split up the luxeon z's into three separate channels 2 each with 8 leds, and one with 6 leds. Below is a list of how the channels are separated;

* channel 1 / vero 10 (1 led)
* channel 2 / luxeon z (8 leds)
* channel 3 / luxeon z (8 leds)
* channel 4 / luxeon z (6 leds)
* channel 5 / luxeon m (4 leds)

So now the choices to be made are on the luxeon z channels, do I go with the channels as listed in option 1 or option 2;

Option 1:
* channel 2 / 6 luxeon blue and 2 luxeon uv
* channel 3 / 6 luxeon cyan and 2 luxeon uv
* channel 4 / 4 luxeon lime and 2 luxeon uv

Option 2:
* channel 2 / 8 luxeon blue
* channel 3 / 8 luxeon cyan
* channel 4 / 2 luxeon lime and 4 luxeon uv

[Horerczy]

I prefer option one.

[bpaston]

I prefer option one.

Thats what I was leaning towards, but what if once built and fired up i only use the cyan channel at 50%. Then am I taking away the punch from having them on there own channel, in a sense wasting the uv leds?

 

Evil if you are out there would love to hear your thoughts also, along with Ben's.

[jedimasterben]

You will want the Luxeon UV on their own channel. Violets are one that you don't want to piggyback with anything else.

[bpaston]

You will want the Luxeon UV on their own channel. Violets are one that you don't want to piggyback with anything else.

 

So with that said do I piggy back;

 

* 2 limes on the uv channel

* 2 limes on the blue channel

* 2 limes on the cyan channel

* 2 limes on the cyan channel and 2 limes on the blue channel

[jedimasterben]

Lime and cyan make the most sense. You want as much independent control of the white/blue/green/violet channels as possible. You can group white in with lime and cyan, though, if need be (not so much in your case since those will be separate PCB), and blue/royal blue together, but definitely keep all violet separate.

[bpaston]

Well gerber files were just finalized and off to the board house they go. Hopefully it will be received tomorrow, and they give me a timeline on when to expect finished boards in the mail. Since this is my first time designing a PCB for leds, I am hoping I did not make any rookie mistakes...Now back to tracking down the luxeon z limes part number lxz1-px01-0184 which no one seems to carry. If any of you see these out there please let me know! Thanks and happy memorial day out there to all those that serve and have served our country.

[farkwar]

If the owner of this resource decides that my presence does not benefit the community, I will abide by his decision.

 

I agree with Clive.

 

You have a great product. You should become a sponsor.

[bpaston]

So I decided to do some research on sinkpad as DNK suggested, although they will not work with the luxeon z. After speaking to their sales rep, I found that the minimum order for a custom designed MCPCB is $1000. A little out of my price range for something that is likely to be a one off. Total cost for the MCPCB's that I am producing will be $16.10 a board plus shipping, once I am able to get a definite price on the rest I will post up a detailed cost for any of those interested. But at this point it is still looking to be not even half the cost of an ecoxotic canon and will have 5 channels of adjustment and full spectrum.

[DNK]

You should become a sponsor.

Agreed. I will be a sponsor, but little later, when I'll have something really awesome

 

2bpatson Please understand - any types of tight LED assembly without proper cooling (you should read - without SinkPAD) is a naughtiness, because the temperature of the LED crystal will be very high. It will cause dramatically decreasing quantity of radiation.

[bpaston]

I am sure you are correct but for this project and for a product I do not plan on selling, $1000 for 5 MCPCB sinkpads is not a very efficient use of money. I appreciate your input on using proper thermal management, but with the use of luxeon z's this is achieved through the use of wider than normal traces. From now on I would appreciate unless you have some constructive criticism to add please refrain from using the word sinkpad in this thread, I feel as if you are getting paid per use of the word. Please dont take this as disrespect I am just trying to keep the clutter to a minimum in this thread. It has been duely noted that you do not approve of the use of my luxeon z's or choice of MCPCB in this project.

[bpaston]

 

Here is the tank that the lighting will be going over... the tank is 60" long X 20" wide and 20" deep. Plan is to hang from the ceiling which is about 9.5', and the tank is sitting on a counter height stand 36". So roughly 6.5' from the bottom of the tank is the mounting point, but the overall length of the heatsink this will be mounted to will move it closer to the tank. I am figuring I will be about 5' from bottom of sandbed to the actual point of light on the fixture.

[DNK]

From now on I would appreciate unless you have some constructive criticism to add please refrain from using the word sinkpad in this thread

I'm sorry for my persistence. But I have been accustomed to do best or nothing. May be you right, hobby - it is process, not result

 

About using Luxeon Z. Excuse me, but I think use the LEDs with such low efficacy is bad idea. Please look - you will use very big and ugly white LED. You will use it on small current to acheive high efficacy. But you would like to use Luxeon Z with efficacy more than 40% worse than Luxeon K. Why? I don't understand. May be you think in this way you can receive uniform light, i.e. kill disco? I'm sorry, but no. Disco may be killed only with using light diffusing system. Even if more dense LED assembly is need in diffuser to avoid disco.

 

I'm sorry again if my words will looks for you as disrespectful. I really puzzled by this design.

[bpaston]

I'm sorry for my persistence. But I have been accustomed to do best or nothing. May be you right, hobby - it is process, not result :)About using Luxeon Z. Excuse me, but I think use the LEDs with such low efficacy is bad idea. Please look - you will use very big and ugly white LED. You will use it on small current to acheive high efficacy. But you would like to use Luxeon Z with efficacy more than 40% worse than Luxeon K. Why? I don't understand. May be you think in this way you can receive uniform light, i.e. kill disco? I'm sorry, but no. Disco may be killed only with using light diffusing system. Even if more dense LED assembly is need in diffuser to avoid disco.I'm sorry again if my words will looks for you as disrespectful. I really puzzled by this design.

Your posts dont seem disrespectful it just seems as if you are selling sinkpads. I actually like this new post it seems constructive. My reasoning for using luxeon z's was to keep within a certain footprint and the luxeon rebels were to large to stay within the luxeon m border. So that is how that decision was made.

 

Being my first time designing a PCB and a fixture in general I wanted to go with a lighting combination that has received good reviews. So the decision was made to go with the evil cluster idea based on the bridgelux. When you say big and ugly are you refering to how the led package looks, or the actual color rendering of this led?

 

Disco was never a thought as it has never really bothered me much, but we will see with this fixture how it turns out. I may not like the look, wont know till I see it. The reason for mcpcb design was to make a smaller footprint and avoid soldering of individual wires.

 

Now you state luxeon z are 40% less efficient than luxeon k, but luxeon k only come in white and royal blue. Why would you even compare efficiencies of leds that are not comparable for the application. Luxeon z's are being used for color only, white is via bridgelux vero 10, and royal blue via luxeon m. Although I would have rather used luxeon mz, but they do not produce a royal blue version.

[DNK]

My reasoning for using luxeon z's was to keep within a certain footprint and the luxeon rebels were to large to stay within the luxeon m border. So that is how that decision was made.

I have understand. Thank you for explain.

When you say big and ugly are you refering to how the led package looks, or the actual color rendering of this led?

I mean - Bridgelux is big LED. But for what this LED on assembly? May be it has outperformed other type of LEDs in terms of efficacy? Defenitely NO. May be it has special, i.e. unique spectra? Again no. Also I'd like to emphasize - spectra of the separate white LED does not matter. Because in anyway we will use not one type of this LED, but assembly. This assembly is designed for achieving good control of spectrum, isn't it? So, we can control whole spectrum of LED assembly to achieving desirable result - virtually any spectrum, include spectrum with highest possible CRI.

 

Unfortunately, using of such big LED will cause serious problem. We can't use lens (at least for all LEDs), it means we will forced to hang this fixture very low. Or, if we would like to use lens in anyway, we forced to design LED assembly very spaciously, it will cause strong disco.

 

I hope, explanation below will be useful. We have done myriad of tests to understand simple thing.

Visibility of disco is directly depend on wavelenght of LED.

 

I'll try to explain.

 

Human eyes has a very interesting visibility graph:

 

As you can see, green-yellow part of spectrum has a extremely good visibility. Mostly, human eye is seeing the green-yellow light as white if spectrum of this light has at least a small portion of blue

 

This feature of human eye very important to understanding of visibility of light. Let's check as will looks spectrum for human eye comparative to spectrum of distribution of radiation.

I'd like to show one of typical "good reef spectrum":

As you can see this spectrum has a huge peak in 400-460nm and very low part of 500-600nm. Let's count this spectrum with taking into account eye visibility. We will get this result:

As you can see, huge peak transformed to pimples and green-yellow part has rise significantly! It's mean - this spectrum will looks to eye as white with relative small bluish tint.

 

As a consequence. During designing of the LED assembly we should place LEDs with wavelength, that significantly differ from green-yellow as close as possible to white LEDs. But LEDs with very big differences of wavelength we may place not as close. In practice this mean:

1. Green and long-wavelength version of cyan LED may be placed in any place of the LED assembly. Wavelength of this LED is near to maximum of visibility.

2. Cyan, long-wavelength version of blue LED, PC amber, amber and orange may be placed not very close to white.

3. Royal blue, red, deep red LED should be placed close as possible to white LED.

4. Violet, especially short-wavelength (shorter than 420nm) version may be placed in any place due to extremely low visibility.

Now you state luxeon z are 40% less efficient than luxeon k, but luxeon k only come in white and royal blue. Why would you even compare efficiencies of leds that are not comparable for the application. Luxeon z's are being used for color only

OK. Let's check any other LEDs type. For example, cyan.

Luxeon Z for 700mA current has a radiant efficacy 38LPW, Luxeon Rebel has a 56LPW. In other words, Z worse than Rebel on (56-38)/38*100=47%! And it is not all. You can use Rebel on 1000mA also, and it will be dissipate less heat than Z on 700mA! In other words, you have more 30% of light in case of Rebel absolutely free of any obligation, including excessive heat

[jedimasterben]

I have understand. Thank you for explain.I mean - Bridgelux is big LED. But for what this LED on assembly? May be it has outperformed other type of LEDs in terms of efficacy? Defenitely NO.

They outperform the Luxeon K (and therefore the Luxeon M and Luxeon T that both are based on).

 

May be it has special, i.e. unique spectra? Again no.

It sure kicks the shit out of any of the white LEDs that your arrays use.

 

Also I'd like to emphasize - spectra of the separate white LED does not matter. Because in anyway we will use not one type of this LED, but assembly. This assembly is designed for achieving good control of spectrum, isn't it? So, we can control whole spectrum of LED assembly to achieving desirable result - virtually any spectrum, include spectrum with highest possible CRI.

The issue with being able to fine-tune every part of the array is that what makes one coral look fantastic will leave others looking drab. Find the balance between the two... and then you'll have the same approximate spectral signature as using a good white LED from the start.

[DNK]

First of all I would like to emphasize again CRI of white LEDs does not matter. Please be informed - high CRI white LEDs has the aligned spectra, especially in blue part and, for low CCT, in red part. But we will correct this part of spectra in anyway by addition blue and red LED in our assembly. Moreover, we can tune this essential for high CRI parts of spectrum as we wish. Otherwise, if we installed high-CRI white LED, we can't tune this part of spectrum well. In short words, high-CRI white LEDs are not only unnecessary, but even has disadvantage in case of full-spectrum LED assemblies.

 

I'd like to explain about disadvantage of high-CRI white LEDs. As OSRAM showed, high CRI of white LED may be achieved in best way by mixing color LEDs with base cool white LEDs. Please familiarise with OSRAM Brilliant Mix technology. Also, high-CRI white LEDs have worst efficacy relative to cool white, even if if high-CRI LED will be used on low current. Let's check this.

 

High-CRI Brigelux Vero (CRI 97 @ 3000K) on recommended by many hobbyists on this forum current 350mA has only 81LPW! It is extremely bad efficacy. May be high-CRI Vero for higher CRI will better? Let's check BXRC-50G1000-B-04 - it have CRI 90 @ 5000K with efficacy 117LPW. Looks better? Yes. Until we don't check this parameter for cool white Luxeon TX. L1T2-6570 has a 129LPW at higher current 700mA! Also I'd like to emphasize - L1T2-6570 has a usual size of primary optics and usual lens may be use with it. Also please pay attention to this point - one Vero required 26.7V instead of only 2.8V for Luxeon TX. It's mean - for usual 48V driver we can use only one Vero per driver and 16pcs Luxeon TX. Only for joke - 16pcs of Luxeon TX will give 16*369=5904lm. Vero 10 - only 1090lm.

 

In short - I can't see no one reason to use Vero for marine aquaria.

[jedimasterben]

First of all I would like to emphasize again CRI of white LEDs does not matter. Please be informed - high CRI white LEDs has the aligned spectra, especially in blue part and, for low CCT, in red part. But we will correct this part of spectra in anyway by addition blue and red LED in our assembly. Moreover, we can tune this essential for high CRI parts of spectrum as we wish. Otherwise, if we installed high-CRI white LED, we can't tune this part of spectrum well. In short words, high-CRI white LEDs are not only unnecessary, but even has disadvantage in case of full-spectrum LED assemblies.

CRI certainly does matter. Higher CRI LEDs are better at rendering colors, you are aware of this. We keep corals that are not all fluorescent green, it takes green, amber, and red light to keep them colorful. Guess what a high-CRI LED has a better balance of? Can we get the same effect by using dedicated lime and PC-amber diodes, along with royal blue? Sure - but with one LED we can do the same job that is already balanced and mixed in one diode. Cost is lower, complexity is lower, the end result is the same.

 

I'd like to explain about disadvantage of high-CRI white LEDs. As OSRAM showed, high CRI of white LED may be achieved in best way by mixing color LEDs with base cool white LEDs. Please familiarise with OSRAM Brilliant Mix technology. Also, high-CRI white LEDs have worst efficacy relative to cool white, even if if high-CRI LED will be used on low current. Let's check this.

I'm familiar with 'Brilliant mix'. Why go through the hassle of dedicated LEDs when a single one will do the job? Bridgelux and Philips both cream Osram in color.

 

High-CRI Brigelux Vero (CRI 97 @ 3000K) on recommended by many hobbyists on this forum current 350mA has only 81LPW! It is extremely bad efficacy. May be high-CRI Vero for higher CRI will better? Let's check BXRC-50G1000-B-04 - it have CRI 90 @ 5000K with efficacy 117LPW. Looks better? Yes. Until we don't check this parameter for cool white Luxeon TX. L1T2-6570 has a 129LPW at higher current 700mA! Also I'd like to emphasize - L1T2-6570 has a usual size of primary optics and usual lens may be use with it. Also please pay attention to this point - one Vero required 26.7V instead of only 2.8V for Luxeon TX. It's mean - for usual 48V driver we can use only one Vero per driver and 16pcs Luxeon TX. Only for joke - 16pcs of Luxeon TX will give 16*369=5904lm. Vero 10 - only 1090lm.

So Dmitry, why are you comparing a 90 CRI 5000K LED with a 70 CRI 6500K LED? Why not compare a 5000K 70 CRI LED to a 5000K 70 CRI LED?

 

The Vero series enables high lumen density while keeping efficiency high, as well, and color is excellent. But it's not what you use, and it's not Philips or SemiLEDs, and it can't use a Sinkpad PCB, so why bother with it?

 

Cool white LEDs look terrible over a reef tank. Why on earth you would start with them and then add more LEDs to make up for this fact is beyond me.

[DNK]

I'm familiar with 'Brilliant mix'. Why go through the hassle of dedicated LEDs when a single one will do the job? Bridgelux and Philips both cream Osram in color.

Because dedicated LED with high CRI has a worst efficacy.

 

Also I would like to emphasize again. Bridgelux has NO advanced technology. Its LED has a worst efficacy, even for small current, as I showed before.

So Dmitry, why are you comparing a 90 CRI 5000K LED with a 70 CRI 6500K LED? Why not compare a 5000K 70 CRI LED to a 5000K 70 CRI LED?

As I explained, using a dedicated LED will cause a worst efficacy. Please understand - using a LED assembly with different colored LEDs may provide best possible efficacy and CRI, instead of CCT and CRI of white LED in it. Please let me know - it is clear or not?

 

If you don't understand what I mean, we can't move further.

why are you comparing a 90 CRI 5000K LED with a 70 CRI 6500K LED? Why not compare a 5000K 70 CRI LED to a 5000K 70 CRI LED?

Because you advocate a high-CRI Bridgelux Vero. Did you would like to check efficacy of CRI 70 Vero? Ok, as you wish. BXRC-50C1000-B-04 has 140LPW at current 350mA. L1T2-6570 has efficacy 159LPW on current 350mA. Have you any more question why I advocate for Luxeon TX?

Cool white LEDs look terrible over a reef tank.

Agreed. But only if we used it as only one LED. But we use it as part of full-spectrum LED assembly. Did you understand this?