I have hijacked enough threads doing this so I wanted to start a new one and see what input I could get on this design:

9 K2 LEDs - I ordered 9 blue and 9 white so we can test both
3 1ohm/10w resistors
1 5ma 12v Driver
9 heatsinks

Pretty simple design actually. Found a place that calcs out the circuit for you and prints it. I think so long as I don't overheat the LEDs and don't burn up the resistors, all will be well. Depending on how it looks, I will likely have to buy some reflectors and/or lenses to maximize the directed light, but I can probalby try something cheap from lowes first.

Comments/suggestions welcome. Tell me if I am missing something because I haven't soldered anything in 5 years.

Attached is the diagram showing the basic wiring.

Is this going to be moonlight or Full on light?

I am going to wire both and see what happens. The blue is for my moonlight setup, but I figured what the hell... I would see what 9 whites look like. I also want to see there par rating at around 20". I think I can build one bulb for $6 bucks not including driver costs. Depending on what calcs are used, 99 of these should be equal to or greater than the PFO unit. You would need about 500 watts to push it or three dirvers pushing 175w about $200. I would say it would take me about $1000 plus several hours work to make a full PFO unit not including a cool microprocessor. That is probably the reason they went with the Type III LED as it is substantially cheaper for 70% the output. On the other side, you can drive these K2s up another 30% reducing your lifespan on the bulb.

Keep us posted i'd like to see it when your done.

Cameron,
You and I are on the same page here. Your K2s are more money but I think worth the cost. After meeting with some electrical engineers that I know, my mind has been racing. All the things I can do with a good microprocessor and a modular input. How about a PH sensor that displays on the LCD of the light panel, Monitors the PH 24/7 and warns of any trouble. How about a USB connection input so you can update the microprocessor to different programs, follow the cycle of the moon in any part of the world, etc.. Endless on what you can do. I have read alot about the PAR ratings of the LED lights and I am interested to see what you get with your K2s. Please keep me updated.

I'm along for the ride... Are you going to use a project board, etch your own pcb, or just go with wires for now?.. It's been a while since I've etched PCB's, but I can give you some tips if your looking to go that route... I've got a few large sheets of single sided unetched board left over that could go towards the cause...



Big D

The difficulty in the microprocessor (outside of writing the code and I haven't written assembler professionally for almost 15 years) is wiring the circuit. Suddenly you jump from a simple line of series circuits in parallel to something much more complicated. I can solder around, but not very good at the EE stuff. Course you have engineers to fall back on who do this stuff for a living (or at least in school).

Interestingly, hooking up a mindstorm processor would probably be very cool. You would instantly have access to huge library of programming and building home made sensors don't look that difficult. Plus it would be a lot easier to integrate into a PC environment since a bunch of people have been working on this for years. Throw some float switches in as sensors and you could make a very cool water changing system, more sophisticate top-off, etc.

The difficulty in the microprocessor (outside of writing the code and I haven't written assembler professionally for almost 15 years) is wiring the circuit. Suddenly you jump from a simple line of series circuits in parallel to something much more complicated. I can solder around, but not very good at the EE stuff. Course you have engineers to fall back on who do this stuff for a living (or at least in school).

Interestingly, hooking up a mindstorm processor would probably be very cool. You would instantly have access to huge library of programming and building home made sensors don't look that difficult. Plus it would be a lot easier to integrate into a PC environment since a bunch of people have been working on this for years. Throw some float switches in as sensors and you could make a very cool water changing system, more sophisticate top-off, etc.

;) Exactly... Then you pitch the whole idea to coralife so they can jump into the new market and Boom, free lights for anyone who helped out along the way! LOL :)

The ideas just keep coming with what you can do. How about a powerstrip built into the back so you can plug you power heads into your light and the microprocessor acts like a wavemaker. How about a base unit that all the add-ons are modular so you can add what you want when you want it without having to replace a lamp everytime, just plug the mod part in and update the microprocessor.

God, I am a geek arn't I?!? LOL just seems like something that would be awsome in the marketplace for a heck of a lot cheaper then what PFO is putting out.

I'm along for the ride... Are you going to use a project board, etch your own pcb, or just go with wires for now?.. It's been a while since I've etched PCB's, but I can give you some tips if your looking to go that route... I've got a few large sheets of single sided unetched board left over that could go towards the cause...I will start with a breadboard since I have one laying around. After that it was going to get real sloppy real fast assuming I can get it to work.

I am going to epoxy the LED to the silicon insulation layer then to the heatsink careful not to the epoxy between the LED body and silicon. I bought some cheap T03 heatsinks so each LED will get its own. I will do all 9 LEDs this way. Then I will arrange them in groups of three. I will solder 1 to 2 then 2 to the 3. The 3rd one will get the 1ohm resistor. I will repeat this 2 more times so I have 3 series circuits with 3 LEDs each. Then I will wire the three in parallel with each other. The wiring will be simple 22 gauge wire probably a foot between each LED assembly so I can stretch the entire thing from one end of my tank to the other. If that works out reasonably well, I will head down to lowes and pick up a few reflectors that might work for this project. After that, beg someone with a PAR meter (say mojo) to come over and we play around with the clustering and reflectors just to see what we get.

I was thinking about using a T5 relector and placing an entire 12 LED strip in it. That would be 60w of blue lighting which is roughly the same as 20 of the PFO LEDs. On the otherhand PFO probably uses optics to focus the light a certain distance. I know I would just to increase the PAR rating at that distance for marketing purposes.

Might be something might be nothing, but if it works as easy as I think it can I might be able to build some really stable lighting for super cheap. If nothing else, I will have learned what blows up an LED or better yet have a nice set of moonlights for my aquarium.

BTW, If you didn't mind spending 13 bucks a bulb this gets a lot easier. You can buy the K2 Star, K2 Star Reflector and K2 Star optic then just glue it to the heatsink. It would also look a lot nicer. Check out this

K2 Star LED: http://www.ledsupply.com/05027-pb14.php
K2 Star Optics: http://www.ledsupply.com/l2-opk2-1-003.php

This sounds very promising, especially for the nano guys. A complete light without heat issues for 200ish $$, and no ice probe, etc...

I wonder what it would take to build one for this 24" cube I'm getting.... :D

I am going to try one more time to post my attachment. Not sure why it isn't taking. Here is the very simple circuit diagram.... hopefully.

I went through all this about 1.5 years ago. Schematics for everything - internet capable, web interface- the works. Had some code written for the fundamentals and got some basic stuff working for a very modular platform. Just too many things going on.

But... you're on a different topic than Cameron. He's "just" building an LED light - please start the aquarium controller in a different thread.


;) Exactly... Then you pitch the whole idea to coralife so they can jump into the new market and Boom, free lights for anyone who helped out along the way! LOL :)

The ideas just keep coming with what you can do. How about a powerstrip built into the back so you can plug you power heads into your light and the microprocessor acts like a wavemaker. How about a base unit that all the add-ons are modular so you can add what you want when you want it without having to replace a lamp everytime, just plug the mod part in and update the microprocessor.

God, I am a geek arn't I?!? LOL just seems like something that would be awsome in the marketplace for a heck of a lot cheaper then what PFO is putting out.

New LED tech announced 1/23/07 by Philips... 115 lumens per watt. 12-18 mos a 2000mA LED will deliver 502 lumens. I think the ones today are somewhere between 50-100 lumens at 1000-1500mA.
___________


Philips Lumileds claims performance records with 115 lm/Watt LEDPress release, January 23; Carrie Yu, DigiTimes.com [Wednesday 24 January 2007] http://www.digitimes.com/images/spacepx.gif
Philips Lumileds announced on January 23 new performance records for its high-power white LEDs, claiming its new white LEDs break the 100-lumen-per-Watt mark.
Operating at a 350-mA current level, the LEDs deliver 136 lumens for a light source efficiency of 115 lumens per Watt at a correlated-color-temperature (CCT) of 4685,000. At 2000 mA, the LEDs deliver 502 lumens at a corresponding 61 lumens per Watt.
Philips Lumileds achieved the record results for white LEDs by combining several new and innovative technologies it has developed, according to the company. The first devices using these technologies will be introduced in a new generation of products during this quarter. These new technologies will continue to proliferate in new and existing products throughout the next 12-18 months.
The light output performance announced today and available to the industry in the near future is 17 times greater at the same power than was available in 1999 when Philips Lumileds introduced the first high-power LED, stated Philips. While performance numbers continue to increase for low-power LEDs, operating at lower currents such as 20 mA, high-power LEDs are required to deliver the quality and quantity of light required for today's and tomorrow's lighting applications, according to the company.
Philips Lumileds: Specification of high-power, white LED
Item
Details
Current (mA)
350
2,000
Lumens
136
502
Lumens per Watt
115
61
Watts
1.2
8.3
Source: Company, compiled by DigiTimes, January 2007

Sounds like we're getting closer! I fully expect that in 5 years these will rival MH for output. I can imagine being able to dial in your color, dim for dawn and dusk, etc, etc. Very exciting stuff!

I hadn't heard about the new LEDs. If there output is real and not some gimmick, that would revolutionize lighting across the board. Flashlights, home lighting, outdoor lighting, headlights, etc would all convert to that kind of output. Very impressive.

oh yeah I can get wholesale on Solaris now! Im ordering one to play with tommorow. I'll let you guys know what I think of it. Tony is supposed to bring over a par meeting and test it against my halides.

I'm along for the ride.

Fishy, Will you let us take it apart and reverse engineer it after you get it?!? That would save me quite a bit of time. Seeing if they used DC power as opposed to AC power, etc...

I can't wait to find out what the PAR is. Please measure at various depths because I am wondering if they focused the light to deliver PAR at a certain depth.

LED lights came into today along with the resistors I ordered. I believe the heatsinks will be here Monday. I have a huge tank move this weekend, but hopefully I can find some time to play with this thing next week.

let me know if I can help with the creation process, would love to play with some of that stuff...


/d

So the light output on these leds is supposed to be amazing but what about the light spectrum output of these? And how does the lumen output compare to say.....a MH. Oh and one more thing, do leds put out heat or will they heat up the tank? Lol from the sounds of it I need to take a trip to camerons house and see all his cool DIY since I live right down the block. hahaThere are four LEDs that are intersting from a spectrum standpoint. Blue which is 460nm, second is Royal Blue at 440nm and then Cyan at 490nm. White is interesting as it is rated at Kelvin and that depends on how much juice you throw at it. It can do from 4500k to 10000k. I will be juicing it to around 5500k - 6500k.

Lumens can't really be used to equate the two. MHs put out a rated 22000 lumen at 250w, but most actually throw around 15000 from what I have been seeing in tests. 12 white LEDs only put out a max of 140 lumen. Lumens are really the light you can see so these lights won't be anywhere near as bright to humans. However I can mix in a lot more focused blue in an LED bank which should give better PAR ratings. The Solaris unit uses 25 3watt LEDs which is by my math is about the same as 13-18 5watt LEDs. Trick is to match the same color LEDs as they do so you can get similar par numbers.

LEDs do put out heat, but it is minimal and disapated through the air. Virtually no heat will make it into the tank.

As for cost, my LED driver supports 60W so I should be able to run 9 of these LEDs full out per driver. The K2 bulb cost me about $6, heat sink/silicon buffer $1.25, resistor $.25, wiring say $.25 and the driver is $20 bucks. That makes for about $95 bucks per bank. Later I am going to switch to a good PC power supply as they are cheaper and can easily run 54 of the K2s.

I had success and failure with 3 K2s. I haven't gotten my drivers yet, but I couldn't wait so I just wired three up with a small 12v 1500ma power supply. These things are really bright. The pics don't show much (no making fun of the crappy solder job either) but they do show the beginnings. Right after I got everything up, looked like I burned a bulb up. I don't know why yet, but probably a failure to heat sink properly. I will look into it tomorrow. The last picture I believe shows the last white LED blowing out.

Lookin good so far! Tagging along to see what happens.

Have to rethink how I mount the K2s to the heatsink. I think I am not getting a good enough heat transfer.

lookin good... can't wait to see more

Cameron, this might sound like a dumb question, but can you use CPU heat sink transfer cream (i do not know what the real name for it is)? I would think that would be some of the best mounting cream out there.

yeah its called artic silver or something.


I have some in my pc room..

/d

Arctic Silver would probably work great, though remember that you need to use a very, very small amount. The primary metal in it is silver. Not sure what would happen in salt water.

You can get thermal pads that you can cut that have adhesive on them that would probably work more easily.

Last night I think I found the perfect solution which many have eluded to already. I am going to use Arctic Alumina Thermal Epoxy. I bought 15grams of the stuff so I should have enough to give it a good coating. I also found copper heatsinks for 25 cents each and some 20mm 15degree cheap optics. If they work, I believe pennies before 1963 may work as well. So here is the new cost break down for those interested in a 9 bulb config:

9x Blue K2 LEDs - $2.86 or White K2 LEDs - $2.99
9x Optics - $0.87
9x Heatsink - $0.25 (if they work)
1x Thermal Adhesive - $6.80
1x Driver - $19.99
3x Resistor - $0.16
1x Wire - $1.00
1x Solder - $2.25

For a blue array - $66.34 and another $20 bucks in shipping.

I think if you build two arrays with a mix of white/blue you would be pretty close to what one bank on the Solaris unit puts out. So for around $400 and some labor you could outfit your tank with LED lighting that is probably better than 4x48" T5s and approaching 2 MH lights.

Very exciting... to me at least.

Very exciting to me as well. Very glad my thought about Arctic Alumina was not as dumb as I thought. My calculations are about $475 with shipping to reproduce the $2,300 LEDs of Solaris. Now this is without the microcontroller but alteast you have some lights for a cost simular to 2 X 150W Halides. But you do not have the bulb replacement, the heat problems, nor the electric bills of Halides! Very cool.

Quit burning out bulbs so you can build me one! ;)

I just had an idea/question.....I have a nano that the ballast and bulbs are out on. Now......how much would it cost to retro fit leds into the hood of a 12 g nano. This could be amazing since you could have the low heat output of leds with the brightness of a PC or even metal halide. Cameron let me know if you would be interested in helping me with this once you get your fixture done.

Assuming I get this done, my plan is to create some really easy to follow directions so anyone can DIY. Everything from where to buy and how to do it. Truth is I am not talented at soldering and such so if I can do it... pretty much anyone can.

Beyond replacement lighting at the very least these things make very cool moonlights and accent lighting. You can even get a spot optic that would allow you to highlight certain objects (corals) in your tank with a special hue to make them really pop. Throw that on the end of some locline and you could point a royal blue or cyan hue on practically anything.

So what would be the rough cost for putting these in a nano 12 g hood and how many would you need?

I would guess about $200 since you would likely want 18 of over the top. You could start with 9 and see what that looked like though. I don't know what the PAR is going to be and I am going strictly on what Solaris units have tested at. They have spent a lot more time tinkering so I may not be able to reproduce those numbers easily.

Ouch.......

That looks like a moster of a soldering pen. I've been using a 13w really sharp point. I got it for working on xbox's but I love it. If I remeber it right there aren't much of legs on those LED's are there? I'm not the best at soldering and the small wattage gives a bit of a safety zone. I'm trying to picture what they looked like without the heatsink on them when you showed them to me when I came buy and bought some stuff from you. Oh nad would there be any advantages using the LED driver vs. just a CPU power supply? What are the physical demensions on the driver?

Billy

The soldering gun is just one I have laying around. If I bought one, it would be a smaller unit for sure and yes the legs are VERY small.

I think a CPU power supply which is essential a 12v driver would work fine actually. The reason they are so cheap is that they make so many. If my other drivers don't get here before my new mounting items I will likely use a PC power supply.

Personally I like "ballasts" that are built for this type of system. They use less power (depending on how many you use) and generally speaking are more compact. Otherwise, a good 12v PC powersupply will work great. You do have to jumper the ATX connector to power it on though.

i was trying to think of a way to put them under a hood of a Nano Cube or Aquapod. I know it complicate things , but then cooling is a major issue. what about using a watercooling system but have them directly epoxied to the copper pipe or would this not be enough for heat transfer?

A proper heatsink will do the job. I can not even wrap my mind around a watercooling system for lights. You epoxy the lights to a copper pipe running water and I think the epoxy will not transfer the heat that you are trying to get away. I could be wrong but it is not the best conductivity in the world. Moisture in a aquapod canopy might be an issue too. Atleast in a normal canopy you can have air flow and about 6-8" of space.

Arctic Alumina Thermal Epoxy.

Review
http://www.mikhailtech.com/articles/tim/asaluminaepoxy/

Its non-electrical conducting but thermal conducting epoxy. I figure that you could have more LED's in a tighter space (almost touching each other). The heat transfer on this epoxy is comparable to Arctic Silver 2. so I was thinking soldier the wires to the led slap some of the epoxy to the LED put it on the copper pipe and in 5 minutes you have adhesion. Not only does the epoxy glue it insulates from electrical shorts but transfers heat. so less heat shrink or electrical tape.

As for moisture its not that bad under the hood but in the splash guard. I'm really not running my lights much because I just have sand and water in my pod (different story). There is a lot of condensation on the bottom of the splash but not inside.

I have a plan for "water resisting" these as well. After I make sure each terminal with LED is working, I am going to mount the optic with silicon caulk. I am going to bury the LED in the caulk. Silicon is a pretty good thermal conductor so it shouldn't overheat the LED and it will provide great resistance to the salt. Should be as good or better than most enclosed units built today. Also if heat is a concern, you could always wire in a small fan on the circuit to blow over/around the heatsinks. However, I don't think heat will be a problem so long as the LED is secured properly (my last one wasn't).

Holy crap...

http://www.lck-led.com/product_info.php?cPath=78&products_id=219

I want one and another project for my list

http://www.reefcentral.com/forums/showthread.php?s=&threadid=979567&perpage=25&pagenumber=1

Frewl, Wow I stand corrected... Thanks for the info, I would have never guessed they mady conductive epoxy!

Cameron, Pretty nice links, that LED array is nice but how would you adjust the optics on it?

here are some that I've made. Controlled by DMX-512.