DIY LEDs - The write-up

[Jay1982]

Thanks DWZM, I was looking at the other heatsinks from heatsinkusa. I'll be getting two panels that are 7.23 inches thick and put them one behind another if I go with 2 inches of spacing.

I think I have everything planned so I'll be purchasing the stuff tomorrow. I'm thinking of getting the rapidled DIY retrokits. My only grief is that the meanwells they supply are LPC's which upon investigation don't seem to be dimmable...

 

[Dannyrotorpilot]

Is there a Sale/Trade section in this forum? I need to sell the 3 ELN-60-48's as they are not dimmable and cant dim them with the reefkeeper system. I accidentaly purchased them and need to get 3 more of the D models. Anyone interested? Brand new in box!

 

[SWINGRRRR]

Is there a Sale/Trade section in this forum? I need to sell the 3 ELN-60-48's as they are not dimmable and cant dim them with the reefkeeper system. I accidentaly purchased them and need to get 3 more of the D models. Anyone interested? Brand new in box!

Please don't get this thread closed for something as stupid as this. Read the stickies!!!!!

 

[akindbro4u]

While I agree 100% that cooling is important - beyond keeping the LEDs from insta-death, the cooler they are the longer they'll last. But - melt the solder on the star? It strikes me that the LED itself would vaporize long before that would happen.



If you're serious, shoot me a PM. I can walk you through doing it yourself, no need to hire someone. I did that prototype run of the "ELN dimming shield" mentioned in the other post and I've got lots of extra PCBs (minimum order was 10). Technically there are some flaws in the PCB layout but they're still totally usable, so if people want them I'll give them away for the cost of shipping. Beyond that, you just need to buy an Arduino and a RTC and get your hands on some code (which I can provide, or others - several people have written this stuff.)

Save one for me I will shoot you a pm in the am, when I can get on the computer. posting pm's mobile is no fun. lol

 

[n0rk]

Does anybody have experience with the Recom RCD-24 driver Cutter are selling? I'm looking at these instead of Buckpucks as they're A) cheaper, and B) able to provide me more than 1000mA for my XP-G whites. The thing that I'm sticking on is whether or not I'll need some kind of PWM voltage booster to run these from Arduino, or if they'll do 0-100% dimming from the standard output voltage? The board I'm looking at is the ATmega128.

 

[tahiriqbal]

Does anybody have experience with the Recom RCD-24 driver Cutter are selling? I'm looking at these instead of Buckpucks as they're A) cheaper, and B) able to provide me more than 1000mA for my XP-G whites. The thing that I'm sticking on is whether or not I'll need some kind of PWM voltage booster to run these from Arduino, or if they'll do 0-100% dimming from the standard output voltage? The board I'm looking at is the ATmega128.

They have been around for quite a while now and seems to be good option but you need to be good with your soldring iron since it needs being installed on bread board or PCB style setup.

Why don't you go with Coral Growers Special LED Engine such as 40mm 7 up PCB with 3 x XPE Royal blue & 4 x XP-G Cool white R5 BIN. Since they come with Maxflex5 Boost Driver(Programmable), it would save you lots of hassel from doing extra soldring etc.

 

[n0rk]

They have been around for quite a while now and seems to be good option but you need to be good with your soldring iron since it needs being installed on bread board or PCB style setup.

Why don't you go with Coral Growers Special LED Engine such as 40mm 7 up PCB with 3 x XPE Royal blue & 4 x XP-G Cool white R5 BIN. Since they come with Maxflex5 Boost Driver(Programmable), it would save you lots of hassel from doing extra soldring etc.

I've been soldering for years so it's no real drama, I've developed quite a hand for it the reason I'm not going the LED engines is twofold - firstly I want to be able to position distinct colour groups in order to highlight specific corals with specific colours, and also, I want to be able to run tight optics as I'll be mounting high. It's just not fiscally viable to do it that way all things considered.

 

[der_wille_zur_macht]

That driver will take the PWM signal from the Arduino no problem, based on the datasheet cutter have for it. It also describes it as available in a pinned or wired form, so it would be virtually interchangeable with a buckpuck.

Though, IMHO, if you're handy at soldering, you should just DIY your drivers. You can probably save money. You mentioned those being cheaper than buckpucks - I'm not seeing that. The price for the dimmable version at 1200mA is coming up as $23.40 USD.

 

[wfournier]

That driver will take the PWM signal from the Arduino no problem, based on the datasheet cutter have for it. It also describes it as available in a pinned or wired form, so it would be virtually interchangeable with a buckpuck.

Though, IMHO, if you're handy at soldering, you should just DIY your drivers. You can probably save money. You mentioned those being cheaper than buckpucks - I'm not seeing that. The price for the dimmable version at 1200mA is coming up as $23.40 USD.

Output voltage is listed as 1.2-35v though so I'm guessing you could get up to 10 (might be cutting it close) LEDs per driver rather than 6 for a buckpack.

 

[serpentman]

I dont know about the XR-E RBs, but I used the XP-E RB, and my tank is severly blue. Thats running the whites at 500mA and the RB @750mA. Its real blue to me and I used a 50:50 with 84 LEDs total over a 135. Id liken it to a 20K+.

Are you running the XP-G R5 cool whites? Depending on the driver, I believe running them at 500mA would represent 1/2 to 1/3 their max power which explains why the blues are so dominant. However, by dimming to achieve color, wouldn't this be sacrificing output.

Since your tank is so wide front-back you might want to do two of the narrow heatsinks (the 5.whatevers) parallel as I suggested above - that would be cheaper. You could even do 6 of them, such that you end up with 3 "pairs."

This is what I have planned for my 36" wide tank. Depending on your aquascaping, I think it offers the most flexibility to tweak coverage as well.

 

[Troy McClure]

Hi, I've been following this thread for quite a while, I'm still in the very early stages of planning my 1423l (375 gal) build, and I'm certain I will be selling my 4 x 400w mh's and building a custom led fixture. Being that I would be using a lot of led's and efficiency will be my main concern, I'm wondering if anyone has done a comparison PAR test between the XP-G R5 and the XP-E RB. Obviously the more XP-G cool whites you run the higher the par, but I'm curious how many less led's you could use to get the same PAR readings, but with maybe a 10k to 12k look, rather than using more led's (by adding more royal blues to the mix) to get a 14k+ look.
I will be buying some led's from Cutter soon to try over my 46 gal, but I will be unable to do any PAR tests at this stage, so any feedback would be appreciated.

Ben

 

[Dannyrotorpilot]

Ok so I have my first setup together and wow those XP-G's are very bright. I have a mix of 24 XR-E Royal Blues and 12 XP-G cool whites. Running them both at appx 700mA, the whites almost overpower the blues. I am using the Medium ripple lens on the Blues and the Frosted Medium Lens on the XP-G's and seems like a good balance for now till I mount them on the tank. Will post pics soon of the finished product.

 

[serpentman]

From what I have gathered from folks that are running XP-G CW's, it seems that even with a 2:1 ratio of blue to white, at the same power it seems the whites are close to overpowering the blues.

I plan to run the XP-G's at 1000mA. Do you recommend I go with a ratio of 3:1?

 

[der_wille_zur_macht]

Hi, I've been following this thread for quite a while, I'm still in the very early stages of planning my 1423l (375 gal) build, and I'm certain I will be selling my 4 x 400w mh's and building a custom led fixture. Being that I would be using a lot of led's and efficiency will be my main concern, I'm wondering if anyone has done a comparison PAR test between the XP-G R5 and the XP-E RB. Obviously the more XP-G cool whites you run the higher the par, but I'm curious how many less led's you could use to get the same PAR readings, but with maybe a 10k to 12k look, rather than using more led's (by adding more royal blues to the mix) to get a 14k+ look.
I will be buying some led's from Cutter soon to try over my 46 gal, but I will be unable to do any PAR tests at this stage, so any feedback would be appreciated.

Ben

No one's done side-by-side PAR tests to my knowledge, but the specs are pretty clear - XP-Gs will produce around 30% more light at a given drive current, but they're also spreading that light a little wider.

Experimenting with your small tank first will be the best thing you can do! I'd guess that for a 10kk look, you might want 50:50 XP-G cool white to XP-E royal blue.

Output voltage is listed as 1.2-35v though so I'm guessing you could get up to 10 (might be cutting it close) LEDs per driver rather than 6 for a buckpack.

True! Somehow I missed that.

Ok so I have my first setup together and wow those XP-G's are very bright. I have a mix of 24 XR-E Royal Blues and 12 XP-G cool whites. Running them both at appx 700mA, the whites almost overpower the blues. I am using the Medium ripple lens on the Blues and the Frosted Medium Lens on the XP-G's and seems like a good balance for now till I mount them on the tank. Will post pics soon of the finished product.

Have you tried swapping the optics around a bit?

From what I have gathered from folks that are running XP-G CW's, it seems even with a 2:1 ration of blue to white, the whites are close to overpowering the blues.

I plan to run the XP-G's at 1000mA. Do you recommend I go with a ratio of 3:1?

For every post you see suggesting that 2:1 royal blue XP-E/XR-E to cool white XP-G is too much white, you see another post claiming it's too much blue. It's really a personal preference thing, and will also depend on which bin of both you get, whether you have XP-E or XR-E blues, optics, etc.

 

[serpentman]

For every post you see suggesting that 2:1 royal blue XP-E/XR-E to cool white XP-G is too much white, you see another post claiming it's too much blue. It's really a personal preference thing, and will also depend on which bin of both you get, whether you have XP-E or XR-E blues, optics, etc.

Yeah, that led to some of my confusion. Although I am not a 20K guy, I would like to bridge somewhere between a 14K and 20K appearance.

I guess I am going to have to bite the bullet and rig a test array to be sure. I'm just trying to get the ratio close as I would like to hit the quantity price breaks to save a little $.

 

[SWINGRRRR]

Are you running the XP-G R5 cool whites? Depending on the driver, I believe running them at 500mA would represent 1/2 to 1/3 their max power which explains why the blues are so dominant. However, by dimming to achieve color, wouldn't this be sacrificing output.

I am running the R5. The current is what was suggested on here. I don't want to run them up to high, because don't you sacrifice energy at with only a minimal gain to far above 350mA? I am not dimming them by the current pot in the driver. They are set and forget. I am dimming by the ELN and dim circuit.
Lets see if this will show up:
http://www.coralskyled.com/forum/viewtopic.php?f=2&t=2

 

[sfsuphysics]

Quick question on LEDs... what would happen if you wired 2 in parallel with say 700mA drivers?

Now I know in an idea situation with simple Ohmic devices (yeah I know they don't exist in reality), that the voltage across the two would be constant, but the current would get split evenly between the two (assuming similar resistances).

But, what's the idea situation? Is there enough variance in LEDs that you'd see one get 500mA while the other would get 200mA? Or are they fairly close to 350mA apiece?

 

[der_wille_zur_macht]

Now I know in an idea situation with simple Ohmic devices (yeah I know they don't exist in reality), that the voltage across the two would be constant, but the current would get split evenly between the two (assuming similar resistances).

That's what SHOULD happen, but people (including me!) have observed a several-percent difference in voltage drop from one LED to another of the same model and bin, at a given current. A few percent in voltage drop can translate to tens of percents of current difference in parallel, which means potentially frying one while the other is under-driven if you're driving them close to the limit, or other problems if you're not.

Plus, parallel has the disadvantage of being less safe - if one of two parallel LEDs fails open, the other one will instantly get twice the current! This can be designed around with careful placement of fuses, but you'll still be fighting voltage drop differences.

 

[serpentman]

I am running the R5. The current is what was suggested on here. I don't want to run them up to high, because don't you sacrifice energy at with only a minimal gain to far above 350mA? I am not dimming them by the current pot in the driver. They are set and forget. I am dimming by the ELN and dim circuit.
Lets see if this will show up:
http://www.coralskyled.com/forum/viewtopic.php?f=2&t=2

Very interesting thread and can see the benefits either way. I think it ultimately comes down to whether your preference is efficiency or total light output. From an efficiency standpoint, it actually sounds like it makes more sense to use twice as many LED's and run them at 350mA rather than push them to 1000mA. However, the overall electrical savings may not justify the cost of the additional LED's.

 

[sfsuphysics]

That's what SHOULD happen, but people (including me!) have observed a several-percent difference in voltage drop from one LED to another of the same model and bin, at a given current. A few percent in voltage drop can translate to tens of percents of current difference in parallel, which means potentially frying one while the other is under-driven if you're driving them close to the limit, or other problems if you're not.

Plus, parallel has the disadvantage of being less safe - if one of two parallel LEDs fails open, the other one will instantly get twice the current! This can be designed around with careful placement of fuses, but you'll still be fighting voltage drop differences.

Hmm, yeah I do recall hearing about that whole voltage issue, which is the primary reason you shouldn't use a voltage source to run them.

I wonder if its simply a matter of manually testing LEDs to find those with the closest tolerances. I'm just wondering about doing this because it seems that, the Crees specifically, are most efficient (lumens per watt) at 350mA, so why not use more to achieve that.

Also when an LED goes bad, does it leave a short or an open circuit? Because if it shorts out shouldn't all the current go through the dead LED and the other one would be safe? Even if it does leave an open circuit 700mA should be fine for the additional LED, it'll just be noticeable brighter, as long as you plan for a heat sink that can handle it before hand.