Minimalistic multichip DIY LED build

If CHIP one is already locked in, then why not throw the kitchen sink in CHIP 2 and make a truly broad spectrum chip?

That looks interesting: It also avoids that 6500K "dead flat" spectrum that many folks have complained about with the CW and favor NW and WW for instead.

I would even be happy with:
1) 20 x 10000K
2) 20 x 455nm
3) 20 x 445nm
4) 10 x 420nm, 10 x 430nm
5) 20 x 5500K

This doesn't sound to bad to me either, although I have strong doubts about 430nm.. I'd even say maybe

1) 20 x 15000k
2) 20x455nm
3) 20x445nm
4) 20x420nm (or whatever)
5) 20x3000k

you'd get a nice filler white with the 10000k, a shot of warm with the 3000k and enough blue to do whatever, plus a big chunk of actinic for pop. I say 3000k because it's something that ac-rc actually sells.


I'd prefer 20 x 420nm for actinic, since it gives such strong fluorescent pop, and the photosynthetic action is close to the peak at 430nm. It would be ideal for adding pop without adding much visible blue light. Not a deal breaker for me either way.
 
Last edited:
Many of us are looking for something that can be versitile and replicate dawn, dusk, mid day, reddish skies, whatever. If we need more punch we will ad more chips. We are not looking for a pre-configured "dream ratio" that only needs to be ramped up and down for intensity.

^^This

I want versatility. I have yet to see anything other than a multichip that can blend the multiple colors needed for aesthetics, IMO, without having separation issues. I DON'T like red shadows. Even closely spaced arrays with no optics still give you a goofy separation at times. I want a chip that gives me options. Coverage isn't terribly great with such a small point source anyway so using two chips helps alleviate those problems as well.

Upon further review my previous suggestion sucks.

I think I like this best:

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K- possibly even 14K depending on how warm the other whites are and how blue their 14K is.

I'd rather go three white channels even if it means including a higher Kelvin white. For me anyway I'm confident 40 watts of blue will make my tank quite smurfy. If going with 2 white channels I'd probably want a 3000K white with 10000K.

As stated before this is all based off the rating I have in my head.
 
Last edited:
This doesn't sound to bad to me either, although I have strong doubts about 430nm.. I'd even say maybe

1) 20 x 15000k
2) 20x455nm
3) 20x445nm
4) 20x420nm (or whatever)
5) 20x3200k

you'd get a nice filler white with the 10000k, a shot of warm with the 3200k and enough blue to do whatever, plus a big chunk of actinic for pop.

I know 430nm is the peak for chlorophyll a, but at 420 it still has a very strong response, and 420 will have a stronger fluorescent effect. (~80% of the response @ 430nm) Besides, with 420 and 445 bracketing 430nm, there will be PLENTY of 430nm light.. The way it works is, there's still about 50% output at ~10nm off the baseline. So a 420nm LED still outputs 430nm light, but at 50% the level of the 420, and the 445 will similarly output at 435nm. So there's a good overlap.

^ This doesn't sound half bad. We probably do need to use a higher white leaning toward blue plus a warm white to offset the limitations of only being able to use 20 whites of one color in each leg. I'm not sure of 430nm either, but I'm in full agreement with 420nm. We just need to find out which color of theirs would look best (10k, 12k, 15k) to achieve the versatile full spectrum goal of the chip (from urine to smurf)

Its easy to mix and match with separate 3w leds until you find a good combo, but this will be a bit more difficult, and probably why no one has attempted it before.
 
^^This

I want versatility. I have yet to see anything other than a multichip that can blend the multiple colors needed for aesthetics, IMO, without having separation issues. I DON'T like red shadows. Even closely spaced arrays with no optics still give you a goofy separation at times. I want a chip that gives me options. Coverage isn't terribly great with such a small point source anyway so using two chips helps alleviate those problems as well.

Upon further review my previous suggestion sucks.

I think I like this best:

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K- possibly even 14K depending on how warm the other whites are and how blue their 14K is.

I'd rather go three white channels even if it means including a higher Kelvin white. For me anyway I'm confident 40 watts of blue will make my tank quite smurfy. If going with 2 white channels I'd probably want a 3000K white with 10000K.

As stated before this is all based off the rating I have in my head.

Yes, it covers a good range. Removing a leg each from the both 445nm/455nm will prevent that overpowering blue. You'll get enough blue from the other LEDs anyway.

I would seriously consider going with 14000K instead of 12000K as you might miss that extra bit of blue. The way it stands, the spectral range is very dynamic covering low to high: 3300K-14000K. I'd keep 3300K but fine with either 12000K or 14000K:

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K/14000K
 
This does not look bad on paper:
1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K/14000K


It is risky I guess, but that is what "chip 1" is for.
 
This does not look bad on paper:
1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K/14000K


It is risky I guess, but that is what "chip 1" is for.

It is definitely risky, as it all boils down to how bright those low ks can get. If they are a bright led, it could be too yellow to work with. It goes against the norm in something like my Cree build where you need twice the amount of blue to compensate the low ks to reach 14k or something aesthetically pleasing at all channels max. A multichip like this is no norm though :)

If this can hit that 10-14k range with color pop, yet provide the flex to 20k still through channel adjustment, this could look fantastic. It would be like having 3 MH bulbs in one, with T5 actinic all in one chip.
 
1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 5500
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 12000K/14000K

This chip looks to add more flexibility. I like it. I would prefer the 14000K and 420nm for my chip. I would use 3 of them on my tank. I am also planning to use the 8w sunset chips ac-rc has to replicate the dawn and dusk colors.
 
It is definitely risky, as it all boils down to how bright those low ks can get. If they are a bright led, it could be too yellow to work with. It goes against the norm in something like my Cree build where you need twice the amount of blue to compensate the low ks to reach 14k or something aesthetically pleasing at all channels max. A multichip like this is no norm though :)

If this can hit that 10-14k range with color pop, yet provide the flex to 20k still through channel adjustment, this could look fantastic. It would be like 3 MH bulbs in one, with T5 actinic all in one chip.

I'll verify how bright the low temps can get in lumens/watt.
 
I'll verify how bright the low temps can get in lumens/watt.

Will they produce 2 samples (full price) for you to test the chips before placing a large order? The chinese companies I used for custom led drivers and led fixtures did that. It would eliminate the chance of chip #2 being a total failure just in case. After all, it is an experimental high risk, high reward chip.

If the chip was a bust, I'd chip in to reimburse you for the sample. Actually, ill chip in for your test sample if u can get one.
 
I don't know if it would make a difference but for a "better safe than sorry" way of doing things would it benefit us to arrange the channels as follows:

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 12000K/14000K
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 5500
 
It is definitely risky, as it all boils down to how bright those low ks can get. If they are a bright led, it could be too yellow to work with. It goes against the norm in something like my Cree build where you need twice the amount of blue to compensate the low ks to reach 14k or something aesthetically pleasing at all channels max. A multichip like this is no norm though :)

I honestly don't expect it to be aesthetically pleasing at max on all channels. I think with the 12/14k leg as your base white you could drive the warmer channels just enough to give you accurate color rendering and then use the blues to either cancel out the yellow, or drive it even higher into a 16k+ range while using a decent amount of the headroom available. I would probably be doing the opposite and using more of the warm side with the 12/14k most of the time with just enough blue to keep it at 10-12kish. We would both drive the 420nm chip until we got the desired pop. This is of course assuming it all works the way we think it should. Ha
 
I don't know if it would make a difference but for a "better safe than sorry" way of doing things would it benefit us to arrange the channels as follows:

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 12000K/14000K
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 5500

This mix will look like a 7,500k Cree emitter. If that's what you want, you will be happy with this combo.
 
I honestly don't expect it to be aesthetically pleasing at max on all channels. I think with the 12/14k leg as your base white you could drive the warmer channels just enough to give you accurate color rendering and then use the blues to either cancel out the yellow, or drive it even higher into a 16k+ range while using a decent amount of the headroom available. I would probably be doing the opposite and using more of the warm side with the 12/14k most of the time with just enough blue to keep it at 10-12kish. We would both drive the 420nm chip until we got the desired pop. This is of course assuming it all works the way we think it should. Ha

Yep. In theory, this could hit everything I desire in a led build, in one chip. I could also wire the thing up in minutes, and have a beautiful shimmer without disco that my other build just cant do. In reality, it could also look god awful. lol

I like this if it can be done, and if it provides the flexibility we think it will. If channels have to be turned off just to get it to look right, its a bust or needs more rb.

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 12000K/14000K
4) 10 x 420nm, 10 x 430nm or 20 420nm
5) 20 x 5500
 
The 420-430 combo is a tough call. It may give a more rounded spectral outout, then again who knows with the bins and number of them.

I would say 14000K

1) 20 x 3300K
2) 10 x 445nm, 10 x 455nm
3) 20 x 14000K
4) 20 x 420nm
5) 20 x 5500

And for that matter, we can argue about the 3300K or 5500K and choose one. I want diversity, not all "warm". Replace one of them with the 10000K....

1) 20 x 3300K or 5500 (leaning toward 5500)
2) 10 x 445nm, 10 x 455nm
3) 20 x 14000K
4) 20 x 420nm
5) 20 x 10000K

Thoughts?
 
Im thinking with the 420 we will have more of a bump in the "lower visibility, but still make the corals pop" range. As you said with the inherent inconsistencies in LEDs we'd probably have a good spread from 405-470 with just the three blues. Considering these would be comprised of 40 LEDs those chances would increase wouldn't they? The 430 may be unnecessary. I agree with the 14K. I'm undecided between one warm white, two warm whites, and if one which one. Hmmm....
 
This mix will look like a 7,500k Cree emitter. If that's what you want, you will be happy with this combo.

Only if all channels are run at 100% which is definitely not the point of having a 5 channel emitter.


BTW, are there any spectra of these various "kelvin" rated chips? I see 10000K and 3300K etc throw around, but the spectra is what matters not the kelvin rating... look at any "warm" fluorescent bulb and you'll see why.
 
I couldnt find any of the whites but they do show them for some of the hybrid chips and the blues. I've also seen them linked in this thread. I agree it would be helpful. Perhaps ron could ask.
 
Only if all channels are run at 100% which is definitely not the point of having a 5 channel emitter.


BTW, are there any spectra of these various "kelvin" rated chips? I see 10000K and 3300K etc throw around, but the spectra is what matters not the kelvin rating... look at any "warm" fluorescent bulb and you'll see why.

The ebay pages of some of the chips from ac-rc have the datasheets linked to them. The 20W 3000K links to the Epileds website, but the link is broken.
 
Back
Top