Minimalistic multichip DIY LED build

I'm going to look for other options outside of ac-rc as well.

I found a chinese multichip manufacturer that will use either Bridgelux or Epistar. Since my personal preference of the two is Bridgelux, I emailed them, and I'm going to see what they can offer. It would be even better if they can produce a sample to test with. Some info on them here:

http://www.alibaba.com/product-gs/608318305/50w_led_chips.html

If a company is willing to produce single samples, it basically means we don't need to agree on one ratio, and any customized ratio can be ordered. This is how I have ordered chinese led fixtures, and I don't see why the same can't be done for a multichip.
 
The point I try to give you is that I don´t think it is possible to achive the thing you want with adding low Kelvin sources that are multispectral....
That would be fine, except we are already doing it with discrete 3W LEDs. My problem with the 3W discrete emitters is the distance between them ceates the "disco effect".

One reason is that the K temperature in a chip is depending on current.
Correct, but not really relevant. I don't care if it purple turns to green as the current is lowered. I just want the ability to dial in the color I am looking for and save the various settings in my software. This is a trial and error process anyway. We already do this with the newer OEM fixtures that sport 5, 6 or 6 colors of LEDs. Secondly, I have not mentioned the driver topology I will be using in this thread. My prototype will be built around a 16 channel PWM chip from NXP.

To understand the other question about the "blue disease" I have to ask you how old you are? Not to insult you but because it is a fact that older people like myself (62) do not perceive blue light very well. At high K, we see a crispy white light, while younger people may well see the blue wavelengths, and therefore perceive that there is a blue tone in the light.
If a tree falls in the forest with nobody around, does it make a sound? Yes, as the eye ages, the lens yellows and filters shorter wavelengths. I am not waiting 20+ years to enjoy my LED fixture that is too blue today. I hope you understand my impatience :)
 
I guess we are a bit ahead of our time. It looks like a full spectrum multichip wont be happening anytime soon. I didn't like that 100 would need to be made in order to produce one anyway.
/QUOTE]


I do not agree because I believe there are full spectrum multichips. Its pricey, but it looks really nice. Doesn't Kessil use a full spectrum and same with orphek?
 
Last edited:
232323232%7Ffp635%3C3%3Enu%3D3337%3E65%3B%3E%3A36%3EWSNRCG%3D3858398346336nu0mrj
 
Yes - but in this case - you can´t control every different LED-type seperatly and and therefore you do not have the same freedom of choice.

For me personally I do not want to have a chip that has full spectrum. I just want to to have rather little of the wavelengths available in lots in a light source that has the full spectrum. (This for my reeftank - if I want a light for a planted aquarium or for lot of macro algae - then it becomes something else)

Sincerely Lasse
 
Yes - but in this case - you can´t control every different LED-type seperatly and and therefore you do not have the same freedom of choice.
That is simply a matter of how it is wired. It would be trivial to have 10, 20, 40 or more channels. These are not complicated devices. They are hard wired chip emitters on a substrate connected by wire.

For me personally I do not want to have a chip that has full spectrum. I just want to to have rather little of the wavelengths available in lots in a light source that has the full spectrum. (This for my reeftank - if I want a light for a planted aquarium or for lot of macro algae - then it becomes something else)
But many of us DO want this because it is in essence a POINT SOURCE. We can add more point sources to increase intensity.
 
That is simply a matter of how it is wired. It would be trivial to have 10, 20, 40 or more channels. These are not complicated devices. They are hard wired chip emitters on a substrate connected by wire.

But many of us DO want this because it is in essence a POINT SOURCE. We can add more point sources to increase intensity.

Please - read what I am writing and try to understand what I am meaning. I was refering to the chip psionicdragon present in the post direct before my post and to his post before that. Because that I answered direct after him I thought that I do not needed to quote him. I was asume that everybody understand that it was a reaction to his chip that he claimed was the solution on YOUR problems. Obviously I was wrong and therefore I will in the future, clearly addressing which person's post I answer.

Sincerely Lasse
 

Whose chip is that? Ah Orphek.

See this is why I feel the warmer whites are needed. Notice they have just blue/white chips, but this XP line with added colors is for "better florescence"? Why use green and red separately when you can just incorporate them in with warmer whites?
 
Last edited:
I can point you to any number of LED builds that use warm and neutral whites. I can show you several commercial fixtures that incorporate these as well. They all happen to be the ones I've found most pleasing to the eye as well.

And while I dont plan on driving these anywhere near 250 watts it does leave significant room for adjustment IMO. I can also show you three different 5000K LEDs that have very different spectra. I could do the same with MH.

I doubt there are any commercial units that use 1/5 3300k and 1/5 5500k with no visible blue (465-485). I respect that you may indeed like the output, but I'm only trying to help you understand what the finished product will look like.
 
I doubt there are any commercial units that use 1/5 3300k and 1/5 5500k with no visible blue (465-485). I respect that you may indeed like the output, but I'm only trying to help you understand what the finished product will look like.

I think he meant ww and nw along with royal blue. I'm a huge fan of lower K leds, but they must be used with blue.
 
Please - read what I am writing and try to understand what I am meaning. I was refering to the chip psionicdragon present in the post direct before my post and to his post before that. Because that I answered direct after him I thought that I do not needed to quote him. I was asume that everybody understand that it was a reaction to his chip that he claimed was the solution on YOUR problems. Obviously I was wrong and therefore I will in the future, clearly addressing which person's post I answer.

Sincerely Lasse

Lasse,

I understood your context and simply added to the conversation. I also (again) pointed out that MANY OF US are simply looking to replicate the broad spectrum 3W clusters in a multi-chip. I would not classify us as having a "problem". I hope you understand that to many of us it appears that you are going out of your way to convince the rest of us we don't need it or can't achieve it (full color control) with a multi-chip. I would accept language as a barrier, but your mastery of English is superb, to the point that you are clearly able to add OR ignore context to fit your point. We all understand that you don't want or need more color, some of us do. :)

As others (myself, jerpa, etc) have stated: In the end, we may be more than happy with a design like "chip 1" or the ratios that you have found to be pleasing. However, we would like to see how far the envelope can be pushed with "chip 2".

Your opinion (and Wilsons, etc) with regard to the lowK emitters in this context is excellent and certainly something to consider. I for one have not seen any of them in person so can only rely on the descriptions and experience of those of you who have.
 
Whose chip is that? Ah Orphek.

See this is why I feel the warmer whites are needed. Notice they have just blue/white chips, but this XP line with added colors is for "better florescence"? Why use green and red separately when you can just incorporate them in with warmer whites?

Aren't the warmer whites lacking in those spectrum?
 
I doubt there are any commercial units that use 1/5 3300k and 1/5 5500k with no visible blue (465-485). I respect that you may indeed like the output, but I'm only trying to help you understand what the finished product will look like.

Just for clarification. I am not advocating the use of 2 channels of low Kelvin whites anymore. This is the design I'm supporting right now and I'm replacing a Kelvin rating with a description.

1)14000K
2) 420nm
3) A reddish warmer lower Kelvin white
4) 445nm+455 nm
5) 10000K

It would only include one leg of warm white and, unlike most builds, would have two channels of higher Kelvin whites. Remember even a cool white cree only goes up to 8000K. Once i have seen this manufacturers 10k and 14k I may think we need a 20k emitter instead of 14k, or I may think we should include more visible blues. I'm not saying what we want is possible with this design but I also cant say it is not possible.
 
The supplier I was talking to offered to make two custom configuration multi chips for me, the buy in was a little more than I wanted though...

Min. order was 50 pieces at about $60.00 each... He offered one similar to the first one (Orphek Style), and another using 5 watt cree emitters... Like below.
 

Attachments

  • securedownload.jpg
    securedownload.jpg
    90.4 KB · Views: 5
Aren't the warmer whites lacking in those spectrum?

Generally speaking a lower Kelvin rating will mean the peaks are pushed more towards the red side of the spectrum. The higher Kelvins generally push more into the blue side of the spectrum. However Kelvin ratings are subjective and there are many mixes of light that can create the same color to our eyes. With these variables in play it means spectral plots and our eyes our the best judges. Spectral plots should ensure we are providing the light necessary to promote photosynthesis. The plots may also help guide us aesthetically when looking for general characteristics of an LED but until you have seen it in person it is pure speculation.
 
The supplier I was talking to offered to make two custom configuration multi chips for me, the buy in was a little more than I wanted though...

Min. order was 50 pieces at about $60.00 each... He offered one similar to the first one (Orphek Style), and another using 5 watt cree emitters... Like below.

I've been thinking of the compact Cree pcb myself, and then using a reflector and lens to achieve the same effect. The only problem with Crees, are that they do not make a violet, and you would need to design a completely custom pcb to incorporate the different sized emitter from a different company into it.
 
Not taking anything away from RON trying to work this, but I had already designed a PCB for tighly packed discrete emitters (mix of cree osram and china) and was about to place an order when I became involved in this thread. I think several folks have been working behind the scenes to accompolish what we are talking about here but have (for obvious reasons) kept it private. As predicted by Wilson... There is no way to organize this in this public forum, as there is no way to keep it from being commercial or viewed as commercial, even if the "vendor" collects the money and we (the designers) reap nothing but the benefit of the chips.

In any case WRT the PCBS, I was simply going to reflow them myself on a hotplate and that still may be what I do. One of my customers is a PCB fab house, though I have never used them for my prototypes (don't want to waste their time or goodwill). My initial design (spread out) was on FR4 and would have been thermally ok. The dense design (similar to the cree photo above) would need to be on MC substrate.
 
Back
Top