Diy led

swimmer4uus

Member
Updates:
Finished the wiring of my last couple drivers. I now can drive 12 strings of LEDs, which should cover the whites. I got some more chips to make adjustable drivers, along with more components, and some capicitors to smooth out the transitions between switching.
I ordered today some heatsinks for the 317's to keep their temperature in check while they are opperating. I'm also getting some generic LEDs from E*bay; one string of reds, greens, and yellows. I'll be testing out some different combos of these to get some awesome colorings to pop.
On a side note, I'll start on the temperature control circuit hopefully tonight that'll drive the fans telling them when to turn on and off
 

swimmer4uus

Member
Ran my LM317 based drivers for a burn in for about an hour. They got the mini homemade heatsinks I put on them too hot for my liking, so I've been looking at different drivers. On one of the bigger forums found some info. I've ordered some parts for a new driver, in the buck topology, hopefully everything goes smooth. Total cost for 10 is just under 50 dollars. These are dimmable, and have the ability to hook up to a controller. Anyways, here's the circuit I'll be working with, and the BOM.


In the diagram some of the Labels are wrong. Rsense is my R1, and Rs is my Rsense.
Hopefully more people will get interested in LEDs here on SWF. I'm almost never on here anymore because a lot of info is elsewhere
 

swimmer4uus

Member
One bank. 20 LEDs on


Vs Dual 96w pc's

That's down to 60 watts from 192 watts, with an increase in light intensity. And to think LEDs are "still out to jury"....ppsssshhh
 

swimmer4uus

Member
So I'm starting my NCP3066 based driver. I'm reading the circuit, and putting parts in place. Pretty simple so far but ran into some questions. Getting them answered and then will actually start soldering.

FYI for anyone interested. These are going to have almost the same exact result as buckpucks. Total cost plus shipping for 10....50 buck! And these have PWM dimming.
 

king_neptune

Active Member
Hows this build coming? im wanting to order my parts in a few days. Ill be doing a combo of yours and soundwaves.
Its going to be for a 125g 72" brick.
You still happy with your choice to DIY your own drivers?
Or in hindsight would you makea different choice?
 

swimmer4uus

Member
The DIY drivers, IMO, are the only way to go. Yeah, they don't look as spiffy, but you put them all in a project box anyways. I wish I would have known about the LEDs before I bought my Power Compacts a year ago. I'm not even running a 1/4th of my LED power, and I'm still bleaching corals from too much light. I'm waiting on my last few parts to come in, before I can call this this a wrap, and do a nice little write-up for SWF. Give me a couple mins and I'll post a new FTS
 

acrylic51

Active Member
Very Nice build Swimmer4uus....I'll be following along...I planned to DIY LED's over my 415 gallon plywood build
 

swimmer4uus

Member
Thanks acrylic
So here are a couple shots. I currently have only 30 White LEDs on the tank running. I'm playing around with some different color ratios, and am waiting for some cheapo red, green and yellow LEDs to come in, before I finish with the Blues. The LEDs are driven at ~1000mA. This is currently taxing my passive cooling with the heatsinks. When I finally mount the blues, I'll drop the current down to 500mA






 

scsinet

Active Member
Originally Posted by swimmer4uus
http:///forum/post/3258782
And these have PWM dimming.
Hmmm... I seem to remember someone saying something about how PWM dimming is the way to go... who was that...

Well I think you'll be happier having gone this route. I like this circuit... the DIP package on the chip makes it easier for DIYers.
If you send me a final schematic (once you've tested this), I'll see if I can create PCB artwork for you. If you are building a bunch of these, you may want to make PCBs instead of breadboarding. It's faster, eaiser, and your circuit will be more robust... plus it may come in handy for others down the road.
 

king_neptune

Active Member
ya. what he said ^_^
PCB? PWM? huh?
anyways....
just a rough outline of the parts i plan to order to get my LED array started.
I went to a major LED supplier that seems to be popular with the LED community, they had "Bundles" so I put a few together.
http://us-dc2-order.store.yahoo.net/...onId=ysco.cart
* LED Type : Cree XR-E
* # of LEDs : 18 (+$119.00)
* Color : Neutral-White
* Drive Current : 1000mA
* BuckPuck Options : Dimming w/ Pot. (+$4.00)
* Power-Supply : 24vdc1.7a (+$21.99)
At $181.94, ill order three of these.
Color choice for white was debatable. They offer Cool white,neutral white, and warm white. I dont know which would be best. So i went with Neutral. I dont understand what w/pot means. I assume poton pack, ive heard the word a couple times. I have no idea what it means(or if they are even needed). So an explanation would be nice). But at an extra $4, im not worried(unless its something I DONT want incorporated).
For the blue, same thing pretty much:
* LED Type : Cree XR-E
* # of LEDs : 18 (+$119.00)
* Color : Royal-Blue
* Drive Current : 1000mA
* BuckPuck Options : Dimming w/ Pot. (+$4.00)
* Power-Supply : 24vdc2.5a (+$29.99)
At $189.94 each, Ill get three.
The grand total is just over $1,100. Ill still need the Cooling fins,alum frame, thermal paste, and a few other odds and ends.(picking them out will be easy), and relatively cheap compared to everything else.
Total will be 54w/54b. Ill divide them amongst 2 panels, left and right. Aprox 32" in length each. Ill build a custom canopy, and am hoping to make something that allows me to raise/lower the array to allow me room to work inside the tank when I need too.(Like sticking my head over the water from time to time)
What I'm hoping to achieve is a quality DIY LED system, that rivals something I would have to pay 3x as much for to have solaris make. And theoretically, something Ill be much happier with. Also Dimming Is an absolute MUST! I want to independently control the lighting so make my own custom color schemes(as well as tune it down a bit, if im over burning the corals).
Now aside from the Color choice in white, and the "Whats a poton pack added for?".
I do have a few others:
I was going to order the 1000ma kit( does this mean the bulbs are rated at 1000ma?)
Or is the driver 1000ma, and all Cre-XRE bulbs are 1000ma anyways?
I was wanting to have this system running at 700ma, but bulbs capable of 1000ma lit by it.
My reasoning is that running the bulbs roughly 30% under max would improve the life of the bulb.
Ive read this from several sources, that you want to run them at less than max capacity, and it makes sense.. so I want to do it.
And finally, optics. Will I need to be getting optics with this many LED's? The tank is a standard 125g Brick. I think 22" deep(so realistically the light would only be penetrating 18" of depth at the most)
 

swimmer4uus

Member
Originally Posted by SCSInet
http:///forum/post/3267816
Hmmm... I seem to remember someone saying something about how PWM dimming is the way to go... who was that...

Well I think you'll be happier having gone this route. I like this circuit... the DIP package on the chip makes it easier for DIYers.
If you send me a final schematic (once you've tested this), I'll see if I can create PCB artwork for you. If you are building a bunch of these, you may want to make PCBs instead of breadboarding. It's faster, eaiser, and your circuit will be more robust... plus it may come in handy for others down the road.
Yes, I got on the PWM bandwagen

This would be sweet. Finals are at the end of the week, and next week as well. Hopefully my last parts will be in right when I finish my last test. I'd LOVE to get a PCB built for this, as I see more people using buckpucks (essentially what I'm building). Yes, boose is more efficient, but for the average person DIYing their builds, they buy off the shelf bucks. I'd totally be down for prototyping some PCB's in the buck topology so that people down the road can go off this.
 

swimmer4uus

Member
Neptune. I'd seriously suggest looking at DIYing your own drivers, and part together everything yourself. Although the "kits" now starting to pop up on the market are great for a quick fixture, I think you're going to be much happier if you piece together your own. How many total LEDs are you looking at getting? I'm gonna ramble on specifics of what you posted without quoting.
PCB means Printed Circuit Board. PWM means Pulse Width Modulation. Google brings up a fast explination of the two.
That kit, although it's a good one, I don't think you want. I would look at buying all the parts seperately if you could. That power supply will only let you do roughly 2 or 3 strings, as it's rated only at 2.5A. A simple 6.5A supply, the exact ones I have, will give you more capabilities, and with a 125g, you're going to want that. You would be able to run the "Kits" on one power supply, instead of the 3 they would want. Both for whites and blues. Give me a rundown of the number of LEDs you want, and the configurations you want and I can quickly do a parts search for you if you want.
A POT, is shorthand for a potentiometer. Basically, it's what gives you the ability to adjust light. They can control a lot of things, and where most widely used as volume controls. (Yes, that means the volume knobs, and such, on your car) The pot tells the buckpuck what load you want the LEDs to be driven to.
On the topic of drivers, you'll be able to CONSIDERABLY reduce the cost of your build if you can DIY your own. $20 for a driver versus $2. Since most LED builds are DIY, it's not much more work soldering the drivers.
 

king_neptune

Active Member
OK fair enough. I went from totally opposed to DIY my own drivers. To "on the fence"...leaning slightly towards "On-board". To put it blutnly, it was lazy with a mix of un-confidant. Your pictures really were what intimidated me.I dont know what this stuff means. But I got my dad on board now, and he will help me do all the wiring/solding ect. This is his area of knowledge(used to work on 747's). So if you seem to think DIY drivers are a safe investment(if not superior). Im on on bard!
Sooooo...since your the one who opened this can of worms...ill pick your brain a little more :)
Hers my plans:
54 of these.
# Royal Blue Cree® XR-E Star Features: 425mW Flux @ 350mA
# 1000mA - Max Drive Current
# 3.3Vf - Typ. Forward Voltage @ 350mA
# 450nm - Wavelength
and 54 of these.
# White Cree Q5 XR-E Star Features: 107 lm @ 350mA
# 1000mA - Max Drive Current
# 3.3Vf - Typ. Forward Voltage @ 350mA
# 6500K - Color Temperature
Thats 108 lights for a 125g brick. 1368 square inches of surface, so Im at 1 LED per 12.6 inches of surface. If im right, that means Ill have a well lit tank, that wont need any optics to penetrate to the lower levels(roughly 18" of water). ANd Ill be able to do corals all the way up to SPS. If Im gonna need more, now is the time to plan for it.
Im confidant that Im getting the "Favored" lights. So with these as a foundation to start with. What direction would you take it? My goal is to have the ability to dim white and blue separately.And run these at about 75% of their max rating(with the dimmer set to max). I would like to get by using only two power supplies, one for blue, one for white. And if possible, one dimmer for blue, one dimmer for white.
 

swimmer4uus

Member
So, let's break this down. The suggested power supply is ~24V @ 6.5A (suggested by me). You might be limited to strings of 5 LEDs (like mine), because of the different voltage drops of the LEDs. You can run safely 6 strings per power supply, and maybe push it to 8, but if you're going to invest in LEDs, I don't think it's worth the risk. So you're going to want to run with multiples of 5 or 6. Maybe 4 banks of 30 LEDs (both white and blue).
If you DIY your own driver, probably based on the NCP3066 like mine, the PWM signal will be able to do one of two things for you. The easiest option is to connect it to a knob controlled signal, allowing you dim the string down analog style. You can group any number of strings together, giving you control of blue vs white ratio. You can also leave it individual if you want, and dim each string down to the desired light. The second option is to connect the PWM signal to a micro controller, and it can be programmed to so all sorts of bells and whistles. Dim up for sunrise, down for sunset. Even saw a lighting storm replication. Either way is workable with PWM signals, and at any time you want to switch, it's really simple.
 

swimmer4uus

Member
Picture time!


The second picture is to compair. The left side with the blues, right side without. Right now the blue to white ratio is 1:2
IDK if I want to go more blue or not, as I like a more white, crisp coloration.
 

king_neptune

Active Member
Option A sounds simpler,and more in line with what I originally had in mind. I only wanted to manualy adjust it from time to time. Mainly if I want to look at the color of the corals, or adjust if I notice negative effets from too much intensity.
Otherwise Ill probably have the blues fire 30min prior/post the whites, at full power.And have the whites running at max capacity as the mainstream setting(700MA of course being max setting).
As for driving:
Since I will have two array's. That leaves me with 54 LED's per array, or 27W 27B. 5 LED's per driver sounds good. That means I need 11 drivers per array. That means a total of 22 drivers. 20 of these will be running 5 LED's each. And the last two drivers will be 4 LED's each(4W/4B). That sounds like an awful lot of drivers to DIY. With shipping/part's. How much would this end up costing me? I'm guessing around $200 and a headache to solder(not to mention time consuming) vs $400 and no effort.
Maybe im over anticipating, and its really a simple thing to DIY, and would only take a few hours to do 22 drivers.
Keep in mind, quality is my main goal. Not cheap and easy. I would think a pre-fabbed Driver would be better quality than one I can make from spare parts at radio shack(not calling your plans lame...I'm just worried that Ill be screwing it up, and I want to consider that something made in a factory might be more reliable than something made in my garage. Then again...these ARE made in china :)
Also how big are these DIY drivers? Would mounting them take up alot of space? I was hoping to keep the guts of the system inside a shoebox sized metal container, mounted between the two arrays.
 

swimmer4uus

Member
If done even on breadboards from the shack, you would DEFINANTLY get them in a small container like that. The drivers are super easy to do, I did one in about 10 mins. The drivers so far are around 2 bucks a piece. For sake of simplicity, bump those last couple LEDs up to full strings. You're going to want to keep uniformity when you design your electrical circuits. If you have 2 parrallel strings of LEDs, one 5, one 4, the 4 LEDs are going to be overworked, and will possible fail sooner than the 5's. When the 4 LEDs fail, then the whole workload is on the 5's, causing them to fail prematurely as well. Making all the strings the same is just way more secure.
 

king_neptune

Active Member
Ok, I kinda like where this is heading.
Can you give me a specific parts list so I can head to radio shack tomorrow and begin pricing supplies?
Im gonna regurgitate in my own words what I'm basically understanding:
I get two power supplies, one for blue,one for white. At 24v and enough amps to power 55 lights each. At 700ma each that means I need 38.5 amps. So get something that can handle slightly more amps in case of surge ect. Umm...40ish?
From there I go to the driver, in your case, a DIY driver.This is where most people go to buckpacks (w/potons if they want to have it dimmable).
Next is the LED's. I wire from each driver to a string of 5 LED's. In keeping uniformity as per your suggestion, Im getting 110 bulbs, aka 55w/55B.
This should all be done in series not parrallel, correct? parallel causes more stress, and potential failure.
If I missed something please let me know.
 
Top