DIY LEDs - The write-up

[SimonSKL]

There is so much info on DIY LEDs that I feel my head is ready to explode! Other than some basic knowledge in soldering and electrical connections, I know very little about electronics. But from what I have read and thought I understand, I am going to design and then build my first LED fixture for my reef tank.

I have a 92g corner tank which is somewhat of a challenge to light the tank. This is my tank today lit by 1x250W, 1x150W HQI, and a dual actinic T5 at 78W.

I plan to replace all of my existing lightings with 48 CREE Q5 LED using 4 Mean Well ELN 60 48D PS/Driver. I have a RKE controller and I heard there is a ALC module with two 0-10VDC ports that can dim LED arrays and/or mimic sunrise and sunset pattern.

Here is my Sketchup of my design and I welcome any comment and positive criticism on how I can make it better.

.

I plan to use 1" aluminum angle as heatsink and will have two 120mm 12VDC fan blowing air from the top. The exhaust air will exit from 12x 1" holes on the sides and back.

This is how the LED array will cover the tank. 48 CREE Q5 (24 Cool White and 24 Royal Blue). The aluminum angle strips will be anchored to a 1/4" plywood, The LED stars will not touch the plywood as there will be a 1/4" space between the aluminum angle and plywood. 1" dia holes will be drilled where the LEDs are located. The 1" hole will allow me to attach optics if I decide to use any of them.

Here is a schematic of how I would wire my 4 strings of LEDs with 12 on each string driven by 4 MW dimmable drivers.

This drawing shows how I would connect the MW drivers and RKE ALC to the LEDs.

I don't have a good understanding on how to wire the dimmable drivers to the ALC. What I want is to have the RKE to provide sunrise and sunset lighting as well as allowing me to adjust the light intensity all the white and/or blue LEDs. Any comments on this will be welcome.

Compared with others, my 48 LEDs array is on the low side of light intensity. After hearing some of you mentioned about certain SPS started to bleach under LED and had to turn down the current, I feel 48 is a good number to start.

 

[widmer]

Looks like a good plan to me. ~150w of the Q5 LED might not be quite as bright as your 478w of halide/T5. You may want to consider switching out the Q5's for XP-G's, to get ~1/3 more light out of them and have less heat to deal with. Also, it looks like you're averaging ~6" of length of the L-beam per LED? I wonder if this will be enough surface area to transmit the heat away...Could you substitute the U-beam for 50% more surface area?

 

[laverda]

Simon
It is so hard to light a corner tank with conventional lighting. I think LEDs are ideal for your tank because, they give you so much more flexibility. Why not thin outside the box a little and bend the L or U channel to follow the curve of you tank? You could easily do that with most U channel I think.
It looks to me like you need one more long row of 12 LEDs toward the front. Plus I you think you need more light then the 48 LEDs. I also think U-channel would be better then the L as far as heat dissipation.
If I understand correctly your planning to have the L strips mounted above a thin sheet of plywood with holes in it. One concern about the plywood is, I think it will hinder the air flow. I would just but it all on top of a sheet of acrylic spaced far enough away to allow for optics. That should give more air flow and protect the LEDs and wiring from most of the moisture.

 

[ReefEnabler]

Personally I love the MC-E's I just installed over my tank They give the best shimmer because it's basically 4 XREs in one tiny spot, whereas if the 4 spots are spread out, the shimmer is more scattered and less dramatic. That said you need to get the M bins to get close to XRE Q5 efficiency.

Finally replaced all my luxeons with Crees with Carlco optics and MAN the crees are brighter. Still planning on adding 6 more buckpucks loaded with crees in a few days, will post some pics then.

I'm also gearing up to dim the whole setup with my Profilux (wille has been giving my tips via PM).

Profilux by default uses a 1-10v dimming interface, but you can adjust it to be 0-6v. The buckpucks need a 0-5v ctrl, so you have to set your maximum dimming to be 85%.

Using these settings, the profilux will dim between 0.08 - 4.96v as measured with my multimeter observing a 5min fade period 0 - 85%.

The buckpucks are inverted, so 0.08v will be the MAX on the buckbucks. Since its not 0v, 11ma will be lost on the 700ma pucks, not really a big deal.

 

[widmer]

That's a really good point about the MC-E's creating better shimmer effect... Have you got a camera which could capture that well? Also, are you finding it difficult to cool them?

I would think that at some point Cree would put together some sort of MC-E type thing but with the XP-G dies. With four of the XP-G dies, it would be somewhere between 5 and 6 times as intense as a standard XR-E...

Why isn't anyone here making headlights for their cars out of these things yet? If wikipedia is right and xenon lamps are ~50 lumens/w then like 4 of the XP-G's or 6 of the XR-E's on a buck puck would easily replace the 35w xenon headlamp found in high end cars...

O nm it's being done already...
http://www.electronicsweekly.com/Articles/2009/05/20/46132/led-headlights-within-the-year.htm
http://gizmodo.com/210326/first-led-headlights-appear-unleash-astonishing-styling-possibilities

And this you can buy, can't imagine how on earth it wouldn't fry from its own heat within like one minute though...
http://www.sourcingmap.com/white-smd-led-car-auto-headlight-fog-light-12v-p-42018.html

/digression

 

[ReefEnabler]

I am running two strings of LEDs on each 1.5" x 12" x 1.5" heatsink, a total of ~48 watts on each, right now just have an 80mm fan and its keeping the heatsink pretty cool.

I can probably snag an iphone video, it might show the shimmer.



I really wish they'd come out with a quad "MC-G" type die in Royal Blue That would be sweeeet.


Also I just tested the dimming cycle of the profilux on one of my buckpuck strips using some alligator clips. The dimming was pretty smooth. When it fades from "almost off" to "completely off" there's a noticeable pop, but it's still not bad. Didn't realize it would be this easy to set up.

Just find the Dimming 1 wire (brown) and Ground wire (red) from the profilux extender cable, and hook the brown wire to the Buckpucks CTRL line, and the Red wire to the LED (-) line, and make absolutely sure you set the voltage to fall in the 0-5v range. Make sure you test it with a multimeter first since if the voltage goes too far over 5v it could fry the pucks.

 

[ReefEnabler]

Ok, it seems the LEDs turn off completely when the profilux voltage is set to 71% of 6v, which probably equates to around 4.1v with the resistance of the long cables.

Set it to 70% and they're on but barely, like dim moonlights. I'll just have to live with that little pop, guess there is no true 0-100% dimming with this hackjob Still pretty sweet.

 

[ReefEnabler]

*dp*

 

[ReefEnabler]

This crappy iphone video shows the shimmer with T5s on, but doesn't really do it justice. came out way too blue and contrasty. The back and sides of the tank aren't really getting much coverage with these optics, which is the main reason I'm adding more LEDs soon. That's when I'll be re-soldering all my connections with nice DC plugs and wiring the Profilux to all the pucks... can't wait


http://www.worstkind.com/aquarium/shimmer.mov

 

[sammy113]

Simon:

48 LEDs on that tank won't do much. I'm currently designing an LED system for a tank same to yours. I'd go higher and dimm it down if needed. You can acclimate your corals, you just need to do it with patience.

Another concern is you won't be able to cool down your LEDs well with aluminum angle. I'd use it in smaller applications. Consider better having Heatsink strips. What I'm doing is attaching the LEDs to an aluminum sheet that will have a few heatsinks on top to create more mass to remove heat.

92 Corner tank is 30" high so you'll need tight degree lenses. Then you'll need more LEDs to cover the entire tank. I'd say stay somewhere close to 72 LEDs...

Here's the basic layout i'm doing:

 

[der_wille_zur_macht]

I would think that at some point Cree would put together some sort of MC-E type thing but with the XP-G dies. With four of the XP-G dies, it would be somewhere between 5 and 6 times as intense as a standard XR-E...

That would be cool, but at some point I think we'd hit a point of diminishing returns as far as having light and heat concentrated on a single chip. Note that the MC-E is rated with a max of 700mA, for instance. An XP-G version would be more efficient but we'd still run in to the same problem, just "later on."

Ok, it seems the LEDs turn off completely when the profilux voltage is set to 71% of 6v, which probably equates to around 4.1v with the resistance of the long cables.

Set it to 70% and they're on but barely, like dim moonlights. I'll just have to live with that little pop, guess there is no true 0-100% dimming with this hackjob Still pretty sweet.

Check the datasheet - the REAL dimming range is smaller than 0-5v. The curve drops off at the low end and flattens out at the high end.

 

[g8gxp]

Ok, it seems the LEDs turn off completely when the profilux voltage is set to 71% of 6v, which probably equates to around 4.1v with the resistance of the long cables.

Set it to 70% and they're on but barely, like dim moonlights. I'll just have to live with that little pop, guess there is no true 0-100% dimming with this hackjob Still pretty sweet.

what is your setup like? my P3 comes on at 11% with 2 meanwells per channel.

 

[skippyreef]

Ok, it seems the LEDs turn off completely when the profilux voltage is set to 71% of 6v, which probably equates to around 4.1v with the resistance of the long cables.

Set it to 70% and they're on but barely, like dim moonlights. I'll just have to live with that little pop, guess there is no true 0-100% dimming with this hackjob Still pretty sweet.

Can you explain what you used to wire teh profilux and how you did it??? I have this controller and a dedicated lap top to run things.

 

[bstohrer]

Simon,

Looks well thought out. I agree with some of the others that you may wish for higher intensity. Easier to add the LEDs now versus later.

I think the angles should work. I put together 12 XREs on a 1"X1"X0.125" channel at 3" center to center. I ran the setup at 1000 ma and monitered the temperature with a small probe at the back of one of the stars through a tiny hole drilled through the channel and filled with Arctic Silver. With no airflow in a 70 degree room the temperature rose to 140 F in about 10 minutes and leveled off. When I placed an 8" fan about 15 feet away on low speed blowing perpendicular to the channel (very gentle air flow) the temperature rapidly dropped to 104 F (40 C) and stayed there. Even with no airflow, this is well below Cree's junction temperature limit. Although back of star temps. are going to be a bit lower than true junction temperatures, I think we're reasonably close. I think with good airflow you'll be fine. But just to be on the safe side you could go with larger angles or channels if they will fit.

Also, think about angling your angles towards the back to keep more of the light in the tank and off/out the front glass.

Bob

 

[der_wille_zur_macht]

I ran the setup at 1000 ma and monitered the temperature with a small probe at the back of one of the stars through a tiny hole drilled through the channel and filled with Arctic Silver.

Any chance you measured the temp in any other locations or with any other methods at the same time?

Out of curiosity, it would be interesting to see how measurements made the way you describe correlate to measurements some other people are taking, but until someone does multiple measurement methods on a single build, we won't know.

 

[bstohrer]

Any chance you measured the temp in any other locations or with any other methods at the same time?

I did not measure any other areas other than by touch. The channel felt uniformly warm. I didn't really notice hot or cold spots by touch. The area between stars felt about as warm as the area very close to the star.

I still haven't dissasembled the setup yet, so I could run some more tests. My objective was to see if I could get comfortable using alternatives to the large heatsinks others are using. I really can't use large heatsinks sue to height and weight constraints inside my hood. In summary, I am comfortable with the alternatives and will go that route.

Any suggestions regarding other locations you would like to see measured?

 

[der_wille_zur_macht]

Mainly I'm interested in comparing your method to what customcolor did - he pointed an IR thermometer right at the LED itself. It's really no big deal, just a curiosity thing. The more data the better for other people trying to build things!

 

[ReefEnabler]

Can you explain what you used to wire teh profilux and how you did it??? I have this controller and a dedicated lap top to run things.

1) cut VSL extender cable and strip off outer insulation.
2) strip Brown and Red wires to expose metal
3) on the profilux side, set your desired L-port to 0-6v operation.
4) set a dimming cycle from 0 - ~80%
5)bust out multimeter w/ alligator clips, hook to brown and red wire
6) plug VSL extender cable into profilux/extender junction
7) watch the voltage go through a few dim periods, from 0- 5v then from 5-0v etc to make sure the voltage stays in range. If not, adjust your illumination percentages

If everything looks good at this point you can hook it up to a puck. So far this is all I did:

1) attach screw-terminal alligator plugs to VSL cable brown and red wires.
2) clip Brown wire to buckpuck CTRL wire
3) strip abit of insulation from LED (-) wire and clip on the VSL red wire (GND).

Then you are good to go I do not recommend the alligator clips for any actual usage, only for a quick test. I will not be leaving my LED (-) wire with part exposed either when I wire the LEDs with DC jacks it will provide some room to solder onto.

I havent tested hooking up multiple pucks to one dimming circuit yet.

When I finish with all that I will post a proper writeup with pics.

 

[TheFishMan65]

Angling LEDs

Angling LEDs

My plan is to mount the LED to a strip of aluminum and then bend the strip so that most of the light falls in the tank. So given a lens angle and how wide the LEDs cover (left to right or front to back) how do I design this. Here are some excel formulas I think work, but first let me explain the output:

OUTPUT - {3 columns by 6 row
Lens Angle . 30 ........ degrees
Width ...... 48 ........ inches

Diameter ... 96 ........ inches
ArcLength .. 50.26548246 inches
Deflection . 6.430780618 inches

DESCRIPTION
Lens angle is the angle of your lens
Width is how far your strip spands

Diameter is where the LEDs will converge
Arc Length is how long to cut the strip to fit in your width
Deflection is how far the bar should bend to (if the strip fits in the opening this should be how far up it is)

FORMULAS
Please tell me if I am wrong, been a while since I had trig, but I think they are right.
B4 = B2/2*(TAN(RADIANS(B1/2))+(1/TAN(RADIANS(B1/2))))
B5 = PI()*2*B4*B1/360
B6 = =B2/2*TAN(RADIANS(B1/2))

Issue
I just realized that this will not produce a column of light. If the tank is the diameter (above) away then all light will be concentrated in the middle. At two times the diameter you will be the same width as the strip plus spread. I will try and develop some formulas for this, but I will throw this out so it can be digested and any errors found.

 

[OTCHU]

hello

Finally, ready to order my Leds .... need 18 XRE Q5 and 18 XRE RB.

Did check on dealextreme and did not see the lens and XRE RB ... do I miss them or they don't sell them ?

Ledsupply is pretty princely ... Do you know somebody else who sell them ?

Thank you guys