DIY LEDs - The write-up

Status
Not open for further replies.
Foxy, the CAT4101 essentially fills the same purpose as you are proposing the LM317 for - they're both linear regs that can be run in constant current mode. The big differences are that the CAT4101 is made to accept a separate PWM control signal, so you don't have to PWM the power line yourself (which is probably not as simple as it looks if you want good control) and, even more importantly, the CAT4101 has a .5v minimum dropout, while the LM317 is something like 2.5v. This means it'll run more efficiently, and it'll be easier to run it "hard" (near it's current limit) without needing heatsinking. An LM317 run anywhere near it's 1.5A current limit needs a HUGE heatsink PLUS active cooling PLUS careful PCB design PLUS an ideal environment. A CAT4101 run at 1A only needs heatsinks or active cooling if your PCB sucks or you're running in a bad environment.

I detect a little bit of an over simplification. First, I totally agree about having built-in pwm, which is why I'm interested in trying it. Second, you're better off putting that MOSFET in series between the resistor and the ADJ (and running the regulator on the low-side), for PWM on the LM317. rather than PWM-ing the power line. Third, at the same voltage both chips must dissipate the same number of watts. 24V x 1A = 24W. The LEDs drop ~3.5V so 6 dissipate ~21W leaving 3W. The resistor dissipates 1.25W, leaving 1.75W for the chip regardless of which one... conservation of energy. Now, because lower drop-out if you trim the output voltage of your supply lower, then you CAN decrease dissipation of the chip.... just to be clear to those trying this.
 
Are there any drivers that can run a string of 25 or more leds at 750mah? I would rather not have 10-12 drivers if I don't have to. This is for a 72" 210gal build. TIA
 
25leds @ 3.5V gives a voltage of 87.5... there are some 90V MOSFETs out there, and it is possible to hack-up a very simple constant-current supply from such a MOSFET and potentiometer, so you could in theory do this... a pair of them would be more stable. I've not heard of commercial solution specifically for LEDs at this voltage, but maybe others have.
 
Now, because lower drop-out if you trim the output voltage of your supply lower, then you CAN decrease dissipation of the chip.... just to be clear to those trying this.

IMHO you're nuts if you don't trim the voltage supply to match - we've found a *lot* of variation in individual LED voltage requirements, ranging (IIRC) from 3.2 to 3.8v per LED. Since you're going to trim anyway, (again IMHO) it makes sense to use the more-efficient driver, that wastes less energy heating up the surrounding air.

Another point: the CAT4101 is ridiculously simple to solder (even if it is technically SMT). If you go way back to post 292 you'll see where I made one using my isolation router and through-hole components. It's been running solidly ever since at ~700mA and only gets very slightly warm to the touch.

I only run mine at 700ma because I don't want to prematurely kill the LEDs, and looking at the efficiency/life expectancy/current curves, it seems like a good compromise current. It also fits in well with the 1A max of the CAT4101. If I really wanted to go higher, I'd simply parallel the drivers, but I don't :)

Simon
 
I totally agree about supply trimming, but with a couple of caveats. I wanted to be clear to those reading this, that you ONLY get that efficiency and power dissipation improvement if you trim. Second, that only works if all your strings run at the same forward voltage.

MY tank is using 5 colors of leds, and from different manufacturers though in multiples of 6 each. Many of them have different voltage requirements, and so it is not possible to trim for them all unless I buy several 24V supplies, at which point I should really buy Meanwells and be done with it (You can get 48V units for $22 US).

That's the great thing about current-control, those voltage differences don't matter any more. They just come out as slightly high or lower dissipation by the driver.
 
New design - please give me your thoughts

New design - please give me your thoughts

I have designed the layout below (top diagram is a top view for LED placement and the second shows the light spread using 45-degree optics) for a typical 90G tank (48"x18" surface area) and would like to hear opinions on whether or not this is a good design. I am planning to use Cree XP-G whites and XP-E royal blues in a ratio of 24/32. The diagram below is the positioning I mapped out using 45-degree optics. I'm also planning to use the Meanwell constant current drivers (all from Rapid).

The questions I know I need to ask:

1) I'm thinking that I may not need them out quite so far to the edges. And if I don't need them out so far, should I still keep the same LED count and just move them closer together or just remove the outer rows/columns altogether?

2) Are 45-degree optics a good choice? Rapid also has a 65-degree option but I liked the idea of a tighter spread. But what do I know? :)

3) What other questions do I need to ask?

Any and all help would be appreciated!

<a href="http://s563.photobucket.com/albums/ss79/rguyler/?action=view&current=LEDFixtureInfo.jpg" target="_blank"><img src="http://i563.photobucket.com/albums/ss79/rguyler/LEDFixtureInfo.jpg" border="0" alt="Photobucket"></a>
 
Last edited:
1) I'm thinking that I may not need them out quite so far to the edges. And if I don't need them out so far, should I still keep the same LED count and just move them closer together or just remove the outer rows/columns altogether?

There is overlap at the center so that would imply more at the edges, but then you get loss from what spills from th tank. So IMHO equally space them.

2) Are 45-degree optics a good choice? Rapid also has a 65-degree option but I liked the idea of a tighter spread. But what do I know?

How high are you mounting them? That determins the spread as much as anyhting. See post 4994 IIRC on page 200 (25 post / page).
 
Optics Question

Optics Question

I plan on starting a 144 LED fixture over my 180 SPS tank. I need to keep the LED fixture within the current hood which will only give me a maximum height of about 10 inches above the water. Since the fixture will to so close to the water, I was planning on using C channel spanning the entire surface area to spread light over the entire tank. It will be a mix of 16 XP-G NW, 32 XP-G CW, and 96 XP-E RB.

Since the fixture will be so close to the water will optics be necessary? I can find 80 degree for the XP-E LEDs but could not find them for the XP-G LEDs. The tank is 24 inches deep. Would 65 degree be to narrow?

TBR701
 
FishMan, thanks for the help. I could not fine any 80 degree optics for the XP-G, only the XP-E. Does anyone know any place that might carry them?
 
I've read that some (or maybe even all) of the XP-E and XP-G optics are interchangeable. Maybe the 80-degree XP-E optics will fit the XP-G?
 
LED Lamp Design

LED Lamp Design

Dear Forum,

I am truly amazed to see such a great forum here! I think this might actually be the best source of general LED lighting information I've seen anywhere.

Some friends of mine asked me to design them some custom LED lighting for reef tanks, so I wanted to try to find out some more information about them. I design custom open-hardware LED light fixtures for a living (http://saikoled.com) for hobbyists and DIY folks, and from reading up about the biology and physics involved here, I think that these tanks are just about as perfect an application of LED technology as any I can think of, so I'm very excited about the project!

What I am wondering is if perhaps one of you kind folks could help me answer a few questions:

1. Why do you think LEDs are better than metal halide? My impression is that MH emit a huge amount of power as UV which is essentially useless (and actively dangerous), and that they don't dim well. However, otherwise they are more efficient than an LED and are cheaper. One that seems particularly odd to me is that I keep hearing that *20,000K* MH bulbs are too blue! At 20,000K, it should be putting out the majority of it's radiation at 140nm, way, way past the visible range and far into the range where it's extremely dangerous to look at or even shine on your body!

2. How much can you get 10,000 lumen commercial LED fixtures for? I am absolutely confident that I can make an extremely nice one that I would happily sell for under $1500, likely less if I am making more than one or two. Thus, it's sort of a waste of my time to make my own if I can point my friends to one that is just as good and costs less!

3. What are good sizes for me to design for? I could make a 2' long, 5-6" wide fixture that puts out 10,000 lumens, or a 4' long that puts out 20,000 lumens, and so on. But I'd like to design them to be as close as possible to what a general aquarium user might actually want.

4. Is there anyone around who might be interested in collaborating to do this commercially? I have a lot of experience with building these fixtures, including dealing with problems that most newbies don't think about, like case design, heat dissipation (the more powerful the LED, the more the die heats up and must be cooled rapidly), spectrum tuning, nice user interfaces, wifi control, etc. I would love to find someone who knows about aquariums to work with.

Best wishes and a happy new year,
Brian Neltner
 
HI Guys and Gals, im setting up a 55G rimless 32x20x20 and i was hoping to build a DIY led fixture for it.

I started reading through the posts at the start of the tread and it seemed fairly easy, but now not so much.

I'm keeping a mixed reef and want to get as much color out of my corals as possible w/o using anything other than led's. And in the future i would like to hook the lights up to a controller to sim dawn/dusk.


Would one of you diy guru's mind helping me out, at this point i seriously needs some help.

Also as of right now i haven't ordered any part of the lighting system, was hoping for guidance before i spent any green on it
 
a maximum height of about 10 inches above the water. Since the fixture will to so close to the water
I think that you can use a splash screen and place the fixture 2" above water. In this case optics won't be necessary for sure.

A photo of the hood could help us.
 
I would almost always recommend optics, regardless of fixture height, especially when using the XP-series LEDs.

Optics will greatly improve the efficiency of the fixture by grabbing all of the light that's shining off to the sides, and the light that's hitting the water at too much of an angle to penetrate significantly, and will help to channel it down into the water.

Even if you don't feel like you need them due to the shallowness of your tank or fixture height etc, use them anyway and then you get to run your LEDs at a lower drive current and save more on the power bill etc.

The only situations where I would be hesitant to recommend optics, or would say really wide optics only are when either all of the LEDs are grouped into a very small cluster or when there is a lot of space between each LED.
 
I would almost always recommend optics, regardless of fixture height, especially when using the XP-series LEDs.

Optics will greatly improve the efficiency of the fixture by grabbing all of the light that's shining off to the sides, and the light that's hitting the water at too much of an angle to penetrate significantly, and will help to channel it down into the water.

Even if you don't feel like you need them due to the shallowness of your tank or fixture height etc, use them anyway and then you get to run your LEDs at a lower drive current and save more on the power bill etc.

The only situations where I would be hesitant to recommend optics, or would say really wide optics only are when either all of the LEDs are grouped into a very small cluster or when there is a lot of space between each LED.

This is very good advice =) Optics are really good for improving apparent brightness... the only place they aren't is if you're including UV LEDs. I am looking at using some 400nm purple/UV LEDs, and the PMMA most lenses are made of would absorb a lot of the light. Otherwise, the PMMA generally only absorbs about 20% of the light passing though (i.e. 80% optical efficiency).

Brian
 
Status
Not open for further replies.
Back
Top