on the hyper violet note look into Steve's LED for those, he has ones that support 1000mA. basically two chips in one. I only suggest it because I find that that wavelenght does more for coloration of corals than most people think and that would let you crank them up ranther than needing additional LEDs if you decided you wanted more of the Violet waveleingth. also he can build 3-ups that could include your red, cyan and blues to minimize the color banding potential, price wise they are not bad, and he uses Luxeon LEDs which are right up there with CREE
Input requested on DIY LED build
- Thread starter joshlawless
- Start date
Violets are very good for growth and color.
Steves is great mainly due to his 3up stars at dirt cheap prices. $11 for a custom 3up with like 11 selections of color. Also, he now offers Luxeon M leds, which are in fact MORE efficient than the current XT-E Crees.
As for the dual-chip violets that can run up to 1000ma, I wouldn't. The heatsink star doesn't look up to scratch, and I've read that you can only run them up to 700ma before they start smoking.
So you might as well just use 2 regular violets run at 500ma.
not at all the ones I have running from him are up to snuff. It's not the star that causes the problem, There is nothing special about a MCPCB star board. I've soldered numerous "button style" leds my self to make custom multi-chip emitters and the main problem with violets burning up has to due with the plastic dome lens used on them not being UV stable. there are several threads on this. light below 400nm is considered damaging UVA radiation, and with LEDs being such concentrated point sources if they produce any light in that range they tend to be self destructive if not packaged with the highest quality materials.
his use a quality plastic and are not outputing any true UVA light. I've tested and destroyed nearly every violet out there, and the only ones I'd recommend are those sold by Rapid LED and Steves LED currently. everyone elses I've burnt up in a matter of weeks. but I have my original Rapid violets still runing strong @ just under 700mA after 2 years and various models from Steves all still going good including the 1000mA model running at 1000mA! for just over 2 months now. if they were going to burn they'd have started to show signs by now.
I was merely suggesting that they combine two 700mA chips into one LED so they basically split the current. being so close together and taking into account variations in voltage they must be "de-rated" so they land on the 1000mA max. you woud be just fine running them at only 700 or only 500mA and would get more light than you would from a single chip led at the same current. LEDs are more efficeint at lower currents so you get more light from Two at 500mA than you could from one chip at 1000mA. if that makes any sence.
joshlawless
New member
You guys are absolutely blowing my mind. I love the idea of the custom 3-UPs at StevesLEDs (although the lack of optics options bums me out a bit). Here's my new thought on a repeating pattern:
The blue/white stripes are a 3-up with 2xRB and 1xNW.
The blue/yellow stripes are a 3-up with 2xRB and 1xWW (4000K, maybe 2700K, maybe one of each)
The circus-tent-looking one is a 3-up with cool blue, turquoise, and deep red
The violet one is self evident.
This gives me a ratio of:
10x RB (Royal Blue) --- corrected from 5, thanks asid61 ---
3x NW (Neutral White)
2x WW (Warm White)
3x UV (Violet)
2x CB (Cool blue)
2x DR (Deep Red)
2x TQ (Turquoise)
Less disco shadows on the sand=good. Less light reaching the bottom of the tank=bad.
Maybe for the neutral white 3-ups, I'll go with the CREE version (since 40 degree optics are available), and for the other, I'll go with the Luxeon ES version.
The blue/white stripes are a 3-up with 2xRB and 1xNW.
The blue/yellow stripes are a 3-up with 2xRB and 1xWW (4000K, maybe 2700K, maybe one of each)
The circus-tent-looking one is a 3-up with cool blue, turquoise, and deep red
The violet one is self evident.
This gives me a ratio of:
10x RB (Royal Blue) --- corrected from 5, thanks asid61 ---
3x NW (Neutral White)
2x WW (Warm White)
3x UV (Violet)
2x CB (Cool blue)
2x DR (Deep Red)
2x TQ (Turquoise)
Less disco shadows on the sand=good. Less light reaching the bottom of the tank=bad.
Maybe for the neutral white 3-ups, I'll go with the CREE version (since 40 degree optics are available), and for the other, I'll go with the Luxeon ES version.
Last edited:
That picture looks really trippy. 
That is a lot of white light. I think you calculated wrong though; you get a ratio of:
10RB: 3NW: 2WW: 3TV: 2CB: 2DR: 2T. Which is much bluer (~12k). Be sure to dim down the circus tents, because they are colors.
For optics, you can get optics (I think) that fit around an entire star. Check newark or mouser.
That is a lot of white light. I think you calculated wrong though; you get a ratio of:
10RB: 3NW: 2WW: 3TV: 2CB: 2DR: 2T. Which is much bluer (~12k). Be sure to dim down the circus tents, because they are colors.
For optics, you can get optics (I think) that fit around an entire star. Check newark or mouser.
joshlawless
New member
I do want to avoid putting too much blue in the tank - coral growth is great and all, but I want to like the color rendition, and I've always disliked the overly-blue look.
What if I swap out the Luxeon 3-ups with 2xRB and 1xWW for 1xWW, 1xNW and 1xRB?
That would give me:
8RB : 4NW : 3WW : 3UV : 2CB : 2DR : 2TQ
Still plenty of blue (especially if you count the cool blue), and I can dim the whites a bit if needed (although I'm eager to get a little of the warmer wavelengths into the tank.
Now that I've been educated a bit on the LDDs, I could run the WW on a separate channel from the NW, so I can back them off a bit if needed.
What if I swap out the Luxeon 3-ups with 2xRB and 1xWW for 1xWW, 1xNW and 1xRB?
That would give me:
8RB : 4NW : 3WW : 3UV : 2CB : 2DR : 2TQ
Still plenty of blue (especially if you count the cool blue), and I can dim the whites a bit if needed (although I'm eager to get a little of the warmer wavelengths into the tank.
Now that I've been educated a bit on the LDDs, I could run the WW on a separate channel from the NW, so I can back them off a bit if needed.
joshlawless
New member
More violet, still? About 13% of the emitters in that layout are violet. I'm all for fluorescence, but wouldn't more than 1/6 of the emitters start to look a little freshman-dorm-room-with-Bob-Marley-poster-esque?
I hear you on too warm a temperature. I've got some refinements.
The last layout was, by emitter count:
33% RB
17% NW
13% WW
13% UV
8% CB
8% DR
8% TQ
Here's a variant with more RB, NW, less WW, and about the same UV, CB, DR and TQ.
17 RB
10 NW
3 WW
6 UV
4 R
4 C
4 Bb
Which, by emitter count, is
35% RB (+2%)
21% NW (+4%)
6% WW (-7%)
13% UV (=)
8% CB (%)
8% DR (%)
8% TQ (%)
I hear you on too warm a temperature. I've got some refinements.
The last layout was, by emitter count:
33% RB
17% NW
13% WW
13% UV
8% CB
8% DR
8% TQ
Here's a variant with more RB, NW, less WW, and about the same UV, CB, DR and TQ.
17 RB
10 NW
3 WW
6 UV
4 R
4 C
4 Bb
Which, by emitter count, is
35% RB (+2%)
21% NW (+4%)
6% WW (-7%)
13% UV (=)
8% CB (%)
8% DR (%)
8% TQ (%)
joshlawless
New member
If they are all driven around their max capacity, that is -- it will be readily tunable up and down in color temperature. Still a better idea to design the hardware as close as possible to the desired temperature than to run some channels permanently at 20%, though.
Thanks for all the help! This has been an excellent learning experience!
Yep, why do you find that surprising? I swapped them into a fixture on my frag tank on April 14th, (replacing standard 700mA violets) so that I could crank it up to 1000mA (the rest of the chips are luxeon so that maxes everything out) as I needed a bit more light and the leds are all on one string. If you're at all concerned about violets burning just buy them from RapidLED, I've never read any reports of any of their chips burning up. as a rule of thumb generally you don't want to run LEDs over 70% of their max rating to get the best long term results.
I'm pretty sure that the 3-up 40 degree optics will fit Steve's 3-up chips as well (except for the anti disco version for those you'd need a dome optic) Steve's could probably confirm that.
Heck I have ~40% on the previously mentioned fixture. it gives a look similar to my display that uses VHO Actinic tubes.
6V
2RB
2TB
2WW 2700k 90CRI
2NW 5000k
joshlawless
New member
I'll try to get confirmation, but I'm skeptical, as the group buy 3-ups have holes drilled into the star to accept the optics:
Whereas Steve's have no such holes:
it's more about where the LED sits in relation to the optic. in eather case you have to glue them down. you might have to trim the pegs a bit, but if those don't fit I know they are available, steve's could likely direct you to the proper one for his chips, but typically people don't need them.
joshlawless
New member
Okay, based on everyone's very helpful input, I've got a final fixture design in mind, with a pricetag to boot!
I've bumped the length of the center fixtures from 36" to 40", to better ensure overlap between the center fixtures and the side fixtures.
On a single 40" heatsink, I can fit (using the pattern illustrated below)
21x UV dual-core stars (1000 mA)
12x custom 3-up stars with Luxeon RB, NW & WW on separate circuits (1000 mA/each)
26x custom 3-ups with Luxeon RB, RB & NW (1000 mA/each)
14x custom 3-ups with Luxeon CB, TQ, DR (1000mA, 1000mA, 700mA, respectively)
A channel of 64x RB emitters (Vf of ~3.05V) will be driven by 5 LDD-1000s (12 or 13 per driver)
A channel of 38x NW emitters (Vf of ~3.0V) will be driven by 3 LDD-1000s (12 or 13 per driver)
A channel of 12x WW emitters (Vf of ~3.1V) will be driven by 1 LDD-1000 (all on one driver)
A channel of 21x UV emitters (Vf of ~3.6V) will be driven by 2 LDD-1000s (10 or 11 per driver)
A channel of 14x CB emitters (Vf of ~3.3V) will be driven by 2 LDD-1000s (7 per driver)
A channel of 14x TQ emitters (Vf of ~2.9V) will be driven by 1 LDD-1000 (all on one driver)
A channel of 14x DR emitters (Vf of 2.0V) will be driven by 1 LDD-700 (all on one driver)
If every one of the LDDs were fully populated with emitters, such that the Vf for the whole string summed to 45V (the max recommended output for the LDD when the input voltage is 48V), then I would need a power supply that could provide 14.7A at 48V, or about 705W. By my math, however, the less-than-fully populated strings will consume just under 12A at 48V, necessitating around 570W. The Mean Well SE-600-48 seems to fit the bill, providing 600W at 48V while passively cooled (no fan noise!).
At current prices (getting the LDDs and PSU from powergatellc.com), I total this up to around $1000.
For that price, I'll have a 7-channel, full-spectrum, 40" 570W LED fixture. Why would anyone buy one off the shelf?
I've bumped the length of the center fixtures from 36" to 40", to better ensure overlap between the center fixtures and the side fixtures.
On a single 40" heatsink, I can fit (using the pattern illustrated below)
21x UV dual-core stars (1000 mA)
12x custom 3-up stars with Luxeon RB, NW & WW on separate circuits (1000 mA/each)
26x custom 3-ups with Luxeon RB, RB & NW (1000 mA/each)
14x custom 3-ups with Luxeon CB, TQ, DR (1000mA, 1000mA, 700mA, respectively)
A channel of 64x RB emitters (Vf of ~3.05V) will be driven by 5 LDD-1000s (12 or 13 per driver)
A channel of 38x NW emitters (Vf of ~3.0V) will be driven by 3 LDD-1000s (12 or 13 per driver)
A channel of 12x WW emitters (Vf of ~3.1V) will be driven by 1 LDD-1000 (all on one driver)
A channel of 21x UV emitters (Vf of ~3.6V) will be driven by 2 LDD-1000s (10 or 11 per driver)
A channel of 14x CB emitters (Vf of ~3.3V) will be driven by 2 LDD-1000s (7 per driver)
A channel of 14x TQ emitters (Vf of ~2.9V) will be driven by 1 LDD-1000 (all on one driver)
A channel of 14x DR emitters (Vf of 2.0V) will be driven by 1 LDD-700 (all on one driver)
If every one of the LDDs were fully populated with emitters, such that the Vf for the whole string summed to 45V (the max recommended output for the LDD when the input voltage is 48V), then I would need a power supply that could provide 14.7A at 48V, or about 705W. By my math, however, the less-than-fully populated strings will consume just under 12A at 48V, necessitating around 570W. The Mean Well SE-600-48 seems to fit the bill, providing 600W at 48V while passively cooled (no fan noise!).
At current prices (getting the LDDs and PSU from powergatellc.com), I total this up to around $1000.
For that price, I'll have a 7-channel, full-spectrum, 40" 570W LED fixture. Why would anyone buy one off the shelf?
that will be one impressively powerful light when it's all up and running! only 30 watts of head room on the PSU at full load is a little tight, but then you're not planning to run it at full tilt so it will probably be fine.
do you have plans to use the fans if they are needed after it is all up and running? I'm not familiar with using the makers heatsink in terms of how hot it gets. just keep an eye on temp. the luxeons and Crees can handle heatsink temps upwards from 150F, the violet chips are the weak link though and have a maximum operating temp of just 125F.
edit: never mind, i think you were talking about the PSU not needing fans. Just don't put them in a spot that has no air flow or they will over heat.
do you have plans to use the fans if they are needed after it is all up and running? I'm not familiar with using the makers heatsink in terms of how hot it gets. just keep an eye on temp. the luxeons and Crees can handle heatsink temps upwards from 150F, the violet chips are the weak link though and have a maximum operating temp of just 125F.
edit: never mind, i think you were talking about the PSU not needing fans. Just don't put them in a spot that has no air flow or they will over heat.
Skinnysloth
New member
Not trying to mess up your beautiful layout and build, but have you considered the Luxeon M Royal Blue 12w from Steve's? One M will put out as much light as 4 Luxeon ES single stars. For the 64 royal blue ES emitters, you can instead use 16 M. You can power 4 M with one 1000mA LDD.
Pros:
The Ms will cost less overall, use less energy, save you some time with soldering, and maybe some time if you need to trouble shoot.
Cons:
May not blend as well as the 3-ups. Not designed to fit the screw slots of the Maker's heatsink, but it can work with some creative positioning.
The Luxeon M 5000K NW will have similar efficiency and money saving benefits as well. It will also have the same negatives as mentioned above. Of course there's always the Bridgelux ES array.
I may be barking up the wrong tree as you might prefer the multiple 3up stars configuration, but thought I'd give you another option.
Pros:
The Ms will cost less overall, use less energy, save you some time with soldering, and maybe some time if you need to trouble shoot.
Cons:
May not blend as well as the 3-ups. Not designed to fit the screw slots of the Maker's heatsink, but it can work with some creative positioning.
The Luxeon M 5000K NW will have similar efficiency and money saving benefits as well. It will also have the same negatives as mentioned above. Of course there's always the Bridgelux ES array.
I may be barking up the wrong tree as you might prefer the multiple 3up stars configuration, but thought I'd give you another option.
joshlawless
New member
That's why I posted here - I'm looking for input! Many thanks for your comment. You haven't messed anything up, I'm still obsessively reworking the design. I'll check them out, and see if they'll work in rev.6.
joshlawless
New member
So, here's v.5 and v.6 side by side:
v.5 on the bottom includes, on a single 48" heat sink (yep, I lengthened them again):
24 x 3-ups with RB, RB & NW (1000mA/ea)
6x 3-ups with RB, NW & WW (1000mA/ea)
24x UV dual-core (1000mA)
12x 3-ups with CB, TQ, DR (1000mA, 1000mA, 700mA, respectively)
11x LDD-1000
1x LDD-700
1x Mean Well SE-600-48 48V 600W power supply
The channels are broken down as follows:
RB channel: 54x RB (1000mA, Vf 3.05) (need 4 LDD-1000s to drive 13/14 ea.) (about .92A ea.)
NW channel: 30x NW (1000mA, Vf 3.0) (need 2 LDD-1000s to drive 15 ea.) (about 1.0A ea.)
WW channel: 12x WW (1000mA, Vf 3.1) (need 1 LDD-1000 to drive all 12) (about .45A)
UV channel: 24x UV (1000mA, Vf3.6) (need 2 LDD-1000s to drive 12 ea.) (about .82A ea.)
CB channel: 12x CB (1000mA, Vf 3.3) (need 1 LDD-1000 to drive all 12) (about .89A)
TQ channel: 12x TQ (1000mA, Vf 2.9V) (need 1 LDD-1000 to drive all 12) (about .78A)
DR channel: 12x DR (700mA, Vf 2.0V) (need 1 LDD-700 to drive all 12) (about 0.35A)
v.6 is on top, and includes, on a single 48" heat sink:
12x Luxeon M Royal Blue
6x Luxeon M Neutral White
12x Luxeon ES Warm White 4000KL
24x UV dual-core
12x 3-ups with CB, TQ, DR
11x LDD-1000
1x LDD-700
1x Mean Well SE-600-48 48V 600W power supply (could get away with a 500, but don't want to pay for active PFC)
The channels on v.6 are arranged as follows:
RB channel: 12x RB "œM" (1000mA, Vf 11.4) (need 4 LDD-1000s to drive 3 ea.) (about .78A ea.)
NW channel: 6x NW "œM" (1000mA, Vf 11.4) (need 2 LDD-1000s to drive 3 ea.) (about .78A ea.)
WW channel: 12x WW (1000mA, Vf 3.1) (need 1 LDD-1000 to drive all 12) (about .84A)
UV channel: 24x UV (1000mA, Vf3.6) (need 2 LDD-1000s to drive 12 ea.) (about .82A ea.)
CB channel: 12x CB (1000mA, Vf 3.3) (need 1 LDD-1000 to drive all 12) (about .89A)
TQ channel: 12x TQ (1000mA, Vf 2.9V) (need 1 LDD-1000 to drive all 12) (about .78A)
DR channel: 12x DR (700mA, Vf 2.0V) (need 1 LDD-700 to drive all 12) (about 0.35A)
There's a reduction in total overall output with v.6 compared to v.5. I replaced 9 royal blue ES emitters with 2 "M" emitters - so the reduction in output for royal blue is about 10%, and replaced 5 neutral white emitters with 1 "M" emitter, for a reduction in neutral white of about 20%, and doubled the warm white ES emitter count from 6 to 12. I think this is still enough light for the tank, considering there will be two such 48" fixtures, and two 24" fixtures with exactly half the emitters outlined above.
Any thoughts on the arrangement of the 30mm Luxeon M "star" from Steve's LEDs? I assume I can just trap any roughly opposing edges of the PCB with screws in the Maker's sink.
v.5 on the bottom includes, on a single 48" heat sink (yep, I lengthened them again):
24 x 3-ups with RB, RB & NW (1000mA/ea)
6x 3-ups with RB, NW & WW (1000mA/ea)
24x UV dual-core (1000mA)
12x 3-ups with CB, TQ, DR (1000mA, 1000mA, 700mA, respectively)
11x LDD-1000
1x LDD-700
1x Mean Well SE-600-48 48V 600W power supply
The channels are broken down as follows:
RB channel: 54x RB (1000mA, Vf 3.05) (need 4 LDD-1000s to drive 13/14 ea.) (about .92A ea.)
NW channel: 30x NW (1000mA, Vf 3.0) (need 2 LDD-1000s to drive 15 ea.) (about 1.0A ea.)
WW channel: 12x WW (1000mA, Vf 3.1) (need 1 LDD-1000 to drive all 12) (about .45A)
UV channel: 24x UV (1000mA, Vf3.6) (need 2 LDD-1000s to drive 12 ea.) (about .82A ea.)
CB channel: 12x CB (1000mA, Vf 3.3) (need 1 LDD-1000 to drive all 12) (about .89A)
TQ channel: 12x TQ (1000mA, Vf 2.9V) (need 1 LDD-1000 to drive all 12) (about .78A)
DR channel: 12x DR (700mA, Vf 2.0V) (need 1 LDD-700 to drive all 12) (about 0.35A)
v.6 is on top, and includes, on a single 48" heat sink:
12x Luxeon M Royal Blue
6x Luxeon M Neutral White
12x Luxeon ES Warm White 4000KL
24x UV dual-core
12x 3-ups with CB, TQ, DR
11x LDD-1000
1x LDD-700
1x Mean Well SE-600-48 48V 600W power supply (could get away with a 500, but don't want to pay for active PFC)
The channels on v.6 are arranged as follows:
RB channel: 12x RB "œM" (1000mA, Vf 11.4) (need 4 LDD-1000s to drive 3 ea.) (about .78A ea.)
NW channel: 6x NW "œM" (1000mA, Vf 11.4) (need 2 LDD-1000s to drive 3 ea.) (about .78A ea.)
WW channel: 12x WW (1000mA, Vf 3.1) (need 1 LDD-1000 to drive all 12) (about .84A)
UV channel: 24x UV (1000mA, Vf3.6) (need 2 LDD-1000s to drive 12 ea.) (about .82A ea.)
CB channel: 12x CB (1000mA, Vf 3.3) (need 1 LDD-1000 to drive all 12) (about .89A)
TQ channel: 12x TQ (1000mA, Vf 2.9V) (need 1 LDD-1000 to drive all 12) (about .78A)
DR channel: 12x DR (700mA, Vf 2.0V) (need 1 LDD-700 to drive all 12) (about 0.35A)
There's a reduction in total overall output with v.6 compared to v.5. I replaced 9 royal blue ES emitters with 2 "M" emitters - so the reduction in output for royal blue is about 10%, and replaced 5 neutral white emitters with 1 "M" emitter, for a reduction in neutral white of about 20%, and doubled the warm white ES emitter count from 6 to 12. I think this is still enough light for the tank, considering there will be two such 48" fixtures, and two 24" fixtures with exactly half the emitters outlined above.
Any thoughts on the arrangement of the 30mm Luxeon M "star" from Steve's LEDs? I assume I can just trap any roughly opposing edges of the PCB with screws in the Maker's sink.
