DIY LEDs - The write-up

Status
Not open for further replies.
I don't have anything yet, but i'm planning on building something for a 10g nano and if i like it then i'm going to move to my main display... That being said, i spoke to someone today from my club, DFWMAS, and they stated that you can't keep SPS under LED lighting. That's not true... right?
 
dogstar74; That heat sink is SOOO over kill!! LOL You could cover it, packed with stars. You should ditch those rubber bands before you go another minute. They are rubber and rubber is attacked by air and heat.. They could all fail suddenly, and soon, dropping that nice driver into the soup. Use a piece of wire or tie wraps until you get your taps. BTW you could go buy a single tap.

Otherwise nice job there!



james3370; That single U- channel is a great idea!



That's not true... right?

Nope not true.
 
I've been lurking in this thread since day one and finally have a few questions. I'm nowhere near building a LED rig myself, but the concept is still fascinating to me.

These questions are nothing new and more of a gut-check to make sure I understand some of the concepts involved:

<a href=showthread.php?s=&postid=14835792#post14835792 target=_blank>Originally posted</a> by kcress
dogstar74; That heat sink is SOOO over kill!! LOL You could cover it, packed with stars.

Clearly his sink is overkill, and there are probably examples at the other end of the spectrum floating around, too. This raises a question - how do we know how much heatsink we need? I've seen references in threads to the brightness and life of LEDs depending on junction temperature. Can we measure that easily in mocked-up designs to experimentally arrive at a good choice for a heatsink?

I'm guessing that when constructing a "traditional" light fixture, such as done by the OP in this thread, overkill is probably the easiest way to provide enough heatsink, and it's arguably not that expensive or cumbersome to do. But one of the things that attracts me to LEDs is that the LED itself is so small, it creates the possibility of all sorts of creative lighting designs. When exploring these creative designs, how do we know if we're moving heat away from the chip fast enough?

I'm also struggling to understand how one would decide how many LEDs to put over a tank. I suppose one method is to just copy the commercial fixtures, in terms of lumen-density (i.e. if a commercial filter has 10 LEDs per square foot and they're 100 lumens each, then we need 1,000 lumens/square foot). Back a few pages, stugray posted this:

Lets say I have ~1000 Watts of 15k Metal Halide lighting ( nice round number )
From what I can tell online, numbers for "effective lumens per watt" for MH is ~70 lumens/watt.
This assumes 80% reflection efficiency.

So my tank of 1000 watts should have ~70,000 lumens

Now for a CREE XR-E Q5 driven at 0.75A outputs ~220 lumens with no "reflection factor".
These are 3.0 Watt devices being run at 2.25 Watts.
So that is ~98 lumens/watt.

So to get 70,000 lumens I need:

70,000 / 98 = 714.

Here, he determined that 714w of Cree XR-E Q5 LED is roughly the same lumens as 1000w of 15k MH. I can follow the math, but I'm guessing this ignores some potentially important factors? How do you account for adding optics to the LEDs into the mix? Comparing different reflectors for the MH? Other things I'm not thinking about?

On a related note, it seems that many of these LEDs are spec'd with different outputs at different currents. For example, the white Cree XR-E Q5 is ~220 lumens at 750mw, and 107 lumens at 350mw. If you look at the curves these numbers create, there's a point of maximum efficiency (i.e. highest lumens per watt). It seems that the trend is to drive these LEDs near their maximum current, rather than their point of maximum efficiency. Is this simply because it's perceived to be cheaper to buy fewer LEDs and drive them higher, vs buying more LEDs and driving them more efficiently? How do you choose the current to run a particular LED at?

To apply this with stugray's math above, should one buy 300 XR-E's and run them at full steam, or 500 and run them at maximum efficiency? Is there anything to be considered here other than just weighing the economics of one option over the other?

And, while the examples of small tanks are great, are there any lurkers in this thread that have done this on a large scale (i.e. more than 10 or 12 square feet of tank surface?) The examples posted so far typically seem to be smaller than the OP's fixture.
 
<a href=showthread.php?s=&postid=14835792#post14835792 target=_blank>Originally posted</a> by kcress
dogstar74; That heat sink is SOOO over kill!! LOL You could cover it, packed with stars. You should ditch those rubber bands before you go another minute. They are rubber and rubber is attacked by air and heat.. They could all fail suddenly, and soon, dropping that nice driver into the soup. Use a piece of wire or tie wraps until you get your taps. BTW you could go buy a single tap.

:lol: I know NOW that it's overkill, but I had no idea when I was ordering parts. You know they say the second build is always the best.

As for the rubber bands, don't worry, it's not over the tank right now. I am still using the VHO setup until I can get the driver properly mounted with screws and protected behind an acrylic sheild.

As far as buying a single tap, well... I won't bore you with my financial dillemas! :mad2: :rolleye1:

But rest assured I'm not going to dangle this setup over my SALT water tank in its current state.

Another note: I saw in the store the other day a flashlight called the Coleman MAX. It uses these same Cree XR-E Q4 or Q5 LEDs and is run by three AAA batteries. I noticed that the LED is mounted deep in the light with a plastic shield on the front for protection. No sign of a heat sink anywhere. So I really question needing a HS altogether. I mean, if you can stuff one of these beasts down into the middle of your hand, the casing, and the batteries and cover with a piece of acrylic to fully encase it and make it water resistant, then I kinda think our free floating heat sinks are WAAAYYYYY overkill, and I'm definately not going with this same design when I build one for my 20L. Why buy 30 inches of HS material when the U tube aluminum is such a brilliant idea?!

Cheers.
Aaron
 
Last edited:
<a href=showthread.php?s=&postid=14836470#post14836470 target=_blank>Originally posted</a> by der_wille_zur_macht
These questions are nothing new and more of a gut-check to make sure I understand some of the concepts involved:

Truthfully, while I can respect anyone's academic desire to better the outcomes, I for myself, went into this project for the simplicity of it. If and when others want to fine tune this project, then I will probably make upgrades. But for my first foree into using LEDs, I am happy with the outcomes. Some of the questions you pose are really good, but if it comes down to 7 years of use out of these lights at 700mA or 10 years with 350mA, well I guess the cost of 6 LEDs will have paid for itself exponentially over just the 7 years to the point that I'm not too worried about it. I just needed a fixture now and this posed a great alternative to buying 20 dollar VHO bulbs.

Now my theory would probably change if I was trying to light a 24 inch deep SPS keeping 300 gallon tank. That would require a whole new set of parameters and even at 7 bucks an emitter, the cost would soon be staggering. But for now, it's just a nano, and I'm happy with my Over large heat sink, over driven emitters and poorly soldered wiring connections!:lol:

Cheers
Aaron
 
<a href=showthread.php?s=&postid=14836890#post14836890 target=_blank>Originally posted</a> by dogstar74

Now my theory would probably change if I was trying to light a 24 inch deep SPS keeping 300 gallon tank.

See, thing is, that's most likely my next tank (and one I would want to keep for at least a decade or two), hence my concerns. At that size, initial setup price differences are magnified, but so are ongiong costs.

And let me just say that even if no one bothers to answer my questions, I appreciate all the effort the various folks experimenting with this are putting in to documenting their results on this forum. That'll make it trivial for us late-adopters to build a functional system.
 
<a href=showthread.php?s=&postid=14835792#post14835792 target=_blank>Originally posted</a> by kcress
BTW you could go buy a single tap.

that's what i did. decided on 6-32 screws at lowes & then over in the tool dept, they sell the correct drill & the tap together in a "kit" for about $5

<a href=showthread.php?s=&postid=14835792#post14835792 target=_blank>Originally posted</a> by kcress
james3370; That single U- channel is a great idea!
thanks :) i was looking at the hollow aluminum square tubing & it was kinda pricey for this lil project & then i stumbled across the U-tubing...... :D
 
<a href=showthread.php?s=&postid=14837178#post14837178 target=_blank>Originally posted</a> by der_wille_zur_macht
See, thing is, that's most likely my next tank (and one I would want to keep for at least a decade or two), hence my concerns. At that size, initial setup price differences are magnified, but so are ongiong costs.

And let me just say that even if no one bothers to answer my questions, I appreciate all the effort the various folks experimenting with this are putting in to documenting their results on this forum. That'll make it trivial for us late-adopters to build a functional system.


I look forward to seeing what you decide for that tank. Please be sure to keep us updated!

I considered going all LEDs on my 220, but just realized the amount of work to wire up that many LEDs and pucks over a large area with a 30" deep tank would be alot of commitment, in addition to $$$. So for now I went with T5s and LEDs for shimmer and supplementation (my LEDs are the real actinics, since I am not happy with any T5 actinics).

So for me this is a little foray into the LEDs. If I like what I see, I will consider going all LED in the future. But part of me is also waiting for super-high-wattage LEDs to become more available so that less LEDs are required. And I don't want to use narrow optics either, I want diffused lighting coming in at angles as well as downward rays.

How you want the light to enter the tank changes things. Narrow optics means downward rays and better % of light entering the tank, so less total light is required... but also requires LEDs spaced at small but regular intervals. Going for diffused lighting means you need MORE light, but dont have to be as evenly spaced (aka what Soundwave did). So its hard to come up with a formula for how many LEDs, as theres just so many options.

Tonight I finish installing the last of my T5 ballasts, and then hopefully this weekend I finish mounting the rest of the LEDs (gotta cut more heatsinks up, got 3 more pucks to wire and thats it).
 
I've decided to add a nano reef into my system just to try an LED lighting setup.

I figure, I put them on my motorcycle and loved them, I put them in my 75 vette and love it,
I might as well put them on my fishtank too.
 
i am really looking to change my lighting in my tank
this thread is really interesting.
i am in the middle of tearing down and redoing my tank (72lx30wx24h) 225 gallon
trying to go as low watts as i can using Hydor Koralia Water Circulation Pumps.
and looking into led lights
what about these led lights?
http://www.reefbuilders.com/2009/04...rethinking-implement-aquarium-leds/#more-6339
would these be enough to light a tank my size?
thanks for any input
 
ive been to the shop that sells them. hes changing over his halides to led
they look pretty good
but i like to over research stuff before i buy
i dont think i could build them like you guys ( so i thought i would ask the some of you guys that are building them )
just wondering if they could light up a large tank
 
<a href=showthread.php?s=&postid=14838326#post14838326 target=_blank>Originally posted</a> by wuelfman
ive been to the shop that sells them. hes changing over his halides to led
they look pretty good
but i like to over research stuff before i buy
i dont think i could build them like you guys ( so i thought i would ask the some of you guys that are building them )
just wondering if they could light up a large tank

This place isn't too difficult to get to from SMF is it? I'm traveling to Sacramento next week (final destination Stockton) and will have to swing by and see those lights in action. Worth the stop?
 
<a href=showthread.php?s=&postid=14838843#post14838843 target=_blank>Originally posted</a> by DeathWish302
This place isn't too difficult to get to from SMF is it? I'm traveling to Sacramento next week (final destination Stockton) and will have to swing by and see those lights in action. Worth the stop?

not at all, come out of the airport and get on 5 south then pick up 50/99, then 99 south.


http://maps.yahoo.com/;_ylc=X3oDMTE...-#mvt=m&lat=38.580932&lon=-121.493816&zoom=13

Starting on AIRPORT BLVD
Take ramp onto I-5 S toward SACRAMENTO/YUBA CITY - go 9.7 mi
Take the SAN FRANCISCO/FRESNO exit onto US-50 E toward FRESNO (CA-99) - go 2.5 mi
Take the RENO/FRESNO exit onto CA-99 S toward FRESNO - go 3.6 mi
Take the 47TH AVE WEST exit onto 47TH AVE - go 0.4 mi
Turn Left on MARTIN L KING BLVD - go 0.3 mi
Turn Left on FRANKLIN BLVD - go 0.1 mi
 
<a href=showthread.php?s=&postid=14836470#post14836470 target=_blank>Originally posted</a> by der_wille_zur_macht
This raises a question - how do we know how much heatsink we need?


You have enough heat sink if you are meeting your device junction temperature requirements.

I've seen references in threads to the brightness and life of LEDs depending on junction temperature.

Yes, all semiconductors have life times. They are essentially insanely pure elements that have highly controlled other elements mixed uniformly into the crystal lattices that define the ICs characteristics. Overtime these 'mixtures' will change because of migration of the various elements. As the temperature increases this process speeds up. It doubles for every 10 degrees. So an LED in a sealed bag may last 300 years before it would have issues with eventual migration changes. But heat that bugger up to 140C and you are now talking just a few years before the process results in marked changes,(this is why the brightness decreases with time). Hotter yet, and you will get physical melting and immediate destruction.

The manufacturers do accelerated studies and provide numbers for lifetime. If you run them at a lower temperature the lifetime climbs dramatically.

Can we measure that easily in mocked-up designs to experimentally arrive at a good choice for a heatsink?

No. An engineer would get all the details and do the extensive calculations needed to assure that the LED would stay cool enough. Then would still have to test and confirm the results. Easier is to just use overkill as a DIY.


When exploring these creative designs, how do we know if we're moving heat away from the chip fast enough?

You need a way to measure the LED's temperature. Makers use miniature temp probes. Most LEDs are specified to have an actual junction temperature,(the die inside), according to the temperature of something on the outside. For semiconductors it is often the case temperature. The standard, shaking out for LEDs, is the lead temperature. Often a specific lead.

The easiest method now days is to just use a temp gun. Get it right down to the lead and read the maximum detected temperature. Consider the lead to be that temperature. See how that relates to the data sheet. This will tell you if you're design is adequate or not.

I'm also struggling to understand how one would decide how many LEDs to put over a tank. I suppose one method is to just copy the commercial fixtures, in terms of lumen-density (i.e. if a commercial filter has 10 LEDs per square foot and they're 100 lumens each, then we need 1,000 lumens/square foot).

That is a fairly reliable method. Keep in mind commercial units are going to be pushing the limits on everything. Just bright enough, just cool enough, etc.


How do you account for adding optics to the LEDs into the mix? Comparing different reflectors for the MH? Other things I'm not thinking about?

Short of a bunch of math you just try it. Even engineers designing this stuff would try everything out to try to find any gotchas.

It seems that the trend is to drive these LEDs near their maximum current, rather than their point of maximum efficiency. Is this simply because it's perceived to be cheaper to buy fewer LEDs and drive them higher, vs buying more LEDs and driving them more efficiently? How do you choose the current to run a particular LED at?

It's not perceived! It's real. The public is notorious for only caring about the initial purchase price. They don't generally give a twit about efficiency nor do they even understand it. LED hardware is quite expensive! To compete makers must get the maximum they can out of the minimum hardware and still meet a product family warranty. The buyer sees: PAR then PRICE. They compare those values, so that is what the makers focus on.

Luckily DYIers can go beyond that. ;)
 
Thank you kcress for laying a good foundation here on what exactly LEDs are, and how they work. That was a very good read.
 
I'll chime in that I did some limited testing with my heatsinks. Each had 24 of the Q5 LEDs. The heatsinks were 5.25" X 10". The heatsinks got to about 104 F without a fan, and ran at about 75F with the fan running.

Because of this, I initially mounted the heatsinks in my hood without the fans. But then I found the temp went up to about 120F before I switched it off and decided to wire in the fans. Now everything is kept at around 78F in the hood.

I think the heatsinks are probably overkill from a technical standpoint, but based on kcress' wisdom, it's certainly better to err on the side of caution with a DIY build. The cooler we can keep things, the longer it will last.
 
<a href=showthread.php?s=&postid=14840552#post14840552 target=_blank>Originally posted</a> by widmer
Thank you kcress for laying a good foundation here on what exactly LEDs are, and how they work. That was a very good read.


+1, some very interesting perspective in there.
 
Thanks folks! Always nice to be able to pass knowledge to folks seriously interested about something.

Yossarian99; Welcome to RC! Thanks for that small but very demonstrative bit of info. It's a typically expected result that shows how having contingencies like optional fans can work to make projects more successful. Fans could mean you can run your LEDs harder and still have good lifetimes.

Yossarian99, how did you measure your heatsink temps?
 
Status
Not open for further replies.
Back
Top