Minimalistic multichip DIY LED build

[CaptiveReef]

Definite sales to be made !!!!!!!

Definite sales to be made !!!!!!!

You should go into business!!!! There is a market for your lighting designs in the aquatic business, especially Hobbyists with deep tanks, as well as public aquariums.
Very impressive lighting !!!!!


CaptiveReef

 

[andygold]

I read through the first dozen plus pages or so....and then the last few...
Most of what I read seemed to be geared towards tanks of 40 gallons and up. What about those of us with nanos.....

Also, I realize this is a marine forum with marine tanks, but I figured the gurus here would know the answer for those of us that are also running FW tanks.

Any thought as to which emitter would be best on a FW fish and invert only (Live bearers, Mystery snails, shrimp...no plants) tank?

What color spectrum would we need to light the tank, but not cause algae to bloom like crazy?

Any specific multichip/driver recommendations for FW nanos 3 to 10 gallons?
Same question for marine nanos of the same sizes.

How does one figure the necessary wattage per tank size? I've seen the formula for corals...LxW/16=P, and P*3=#of 3W emitters, but this formula does not take depth into account. What does one do about depth in regards to the formula? Also, how does the formula relate to a fresh water tank?

For those of us running both marine and FW nano tanks, does the above formula still work for a marine nano, or would the lesser depth cause too much light and burn the softies? Would one just throttle the light down a bit to overcome, or would you recommend just using 3W LEDs on a nano? Same answer for the freshwater nano?

 

[Lassef]

The HEAT transfer to an ideal heatsink is purely a function of the electrical efficiency of the emitter. Lower overall electrical efficiency means less energy is converted to photon energy and more to heat at the die.

The real world differs due to the heat density of the multi-chip vs the single emitter. When heat is spread over a larger footprint, it is easier to move away from the critical area. The high density chips need much better thermal management (heatsinking) due to the density.

I have measured both close to the chips (directly under) and a bit away. There is no significant difference between the temperatures.

I put 9 pcs 10 watts multichip at a 12 mm thick aluminum plate. (150 * 250 mm) On the upper side I placed a graphic card cooler with the largest attaching area I found. The cooler had heatpipes. But it only covered about 10 % of the plate´s surface.

I was worried about the heat transfer because even I can my physic so I did a test array and measured very carefully. It is only 1-4 degrees difference between different places on the aluminum plate - including the metal base on the chip. Its no problem.

With the higher wattage - heat pipes works very well. The computer guys have developed very good equipment for heat transfer from small, high wattage processors - let us use their knowledge.

Sincerely Lasse

 

[bhazard451]

The only thing the DIM4 lacks is a lunar phase. I may source out another DIY Arduino controller (perhaps the Jarduino) to get the lunar phase and add DC pump speed control.

As for the multichips, I can't post the source on RC. The version with 30% UV costs almost double that of the version with 10% red and only 10% UV. There seems to be a lot of red in the 14k chip so the 660nm chips don't seem necessary. There is a value in using them during dawn and dusk for unique colours in the fish and coral.

I didn't know there were multichips available like this. I'd be interested in trying them out.

 

[bhazard451]

Who did you measured that? PAR, Lumen or what? Figures? Did you run them at the same current?

Sincerely Lasse

Said 14k Epistar ran at ~60 lumen/w @ 630mah. Looks terrible and I'm removing them from the fixture because it looks so poor.

Said 14k Epistar/Bridgelux from another chinese fixture ran at ~90-100 lumen/w at 580mah. It was also twice as bright and much more visually appealing.

Unless you know exactly what bin (or brand even) that you are getting, the quality varies greatly when you buy the chinese leds.

 

[BeanAnimal]

I have measured both close to the chips (directly under) and a bit away. There is no significant difference between the temperatures.

Respectfully, you did not read what I posted. I was not trying say your were not able to move the heat.

You may be able to cool both setups effectively regardless of heat density and if they both have the same electrical efficiency, then they both conduct the same amount of heat per watt to the heatsink. If they differ in electrical efficiency, then by physical law, they conduct different amounts of heat to the heatsink.

The point was, that the amount of heat generated (conducted to the heatsink) is purely a function of the electrical efficiency. An 20% efficient emitter will produce .8W of heat conducted to the heatsink for every .2 watt of light it radiates. A 35% efficient emitter will conduct .65W of heat to the heatsink for every .35w of light it radiates.

 

[Lassef]

Said 14k Epistar ran at ~60 lumen/w @ 630mah. Looks terrible and I'm removing them from the fixture because it looks so poor.

Said 14k Epistar/Bridgelux from another chinese fixture ran at ~90-100 lumen/w at 580mah. It was also twice as bright and much more visually appealing.

Unless you know exactly what bin (or brand even) that you are getting, the quality varies greatly when you buy the chinese leds.

As I know it Epistar and Epileds is based at Taiwan. Bridgelux in USA. According to the multichip, there is different producers of the final product. At ebay you can find different sellers - the more serious present today BIN and other figures - it is just for you to make your choise.

Sincerely Lasse

 

[ronreef]

@Ronreef
Which hybrid royal blue or attnic

10000K and 445-447nm blue hybrid.

 

[Lassef]

Respectfully, you did not read what I posted. I was not trying say your were not able to move the heat.

You may be able to cool both setups effectively regardless of heat density and if they both have the same electrical efficiency, then they both conduct the same amount of heat per watt to the heatsink. If they differ in electrical efficiency, then by physical law, they conduct different amounts of heat to the heatsink.

The point was, that the amount of heat generated (conducted to the heatsink) is purely a function of the electrical efficiency. An 20% efficient emitter will produce .8W of heat conducted to the heatsink for every .2 watt of light it radiates. A 35% efficient emitter will conduct .65W of heat to the heatsink for every .35w of light it radiates.

I´m sorry - I try to discuss with the other guy that see a lot of problems to transfer the heat.

I am fully aware that much of the electrical energy is to heat already in the diode. Of course, efficiency is of great importance for how much becomes heat and how much gets into photons. But my experience says that the small difference that it is among the better multi-chips and Cree-chips do not have any great significance.

You wrote also

The real world differ due to the heat density of the multi-chip vs the single emitter. When heat is spread over a larger footprint, it is easier to move away from the critical area. The high-density chips need much better thermal management (heatsinking) due to the density.

I was trying to say that this is not a problem today either - it has computer people already solved for us

Sincerly Lasse

 

[Lassef]

Here are some close-ups of mounting the 50W 16000K multichip using Arctic Silver thermal adhesive. The thermal adhesive requires about 8 hours cure time. Note the standoffs between the aluminum angle and the gooseneck bracket. This prevents the bracket from interfering with installation of the fan.

https://picasaweb.google.com/103048...&authkey=Gv1sRgCPihk6v6o6PkGQ&feat=directlink

Ron

It has going so fast the last days so I miss a lot of information in this thread. Very good construction Ron - I hope you continue with how the build progres in this thread.

Sincerely Lasse

 

[bhazard451]

Ok, now that I know certain hybrids exist, and that heat isn't as big of an issue as I thought it was, I'd like to try to build a multichip light to test with.

Based on personal preference, I would probably need 3 50w hybrid multichips and also separate violet chips if they cannot be included with the following colors:

Warm White
Cool White
Royal Blue 440-460
Violet 420nm
Red (very slightly or none at all because of the warm whites)

I would say around 2-3 drivers, with white on one color channel, and blue/violet on the other. Would be a D series Meanwell driver so I can test on an Apex, and then make a larger build for my 120 gallon tank. Which Meanwells work best for these? 36v 2a each?

I would need a 12x12 finned heatsink at first for the nano. Would 120mm fans alone be ok for cooling on it, or would 3 intel coolers be needed?

I'm also trying to determine the costs between a multichip and 3w setup. Maximum I could tolerate is $450 per 36" in length. Any higher than that and I can just buy a Maxspect R420r and save myself the time. My 180w DIY cost around $350 in comparable parts.

 

[ronreef]

Said 14k Epistar ran at ~60 lumen/w @ 630mah. Looks terrible and I'm removing them from the fixture because it looks so poor.

Said 14k Epistar/Bridgelux from another chinese fixture ran at ~90-100 lumen/w at 580mah. It was also twice as bright and much more visually appealing.

Unless you know exactly what bin (or brand even) that you are getting, the quality varies greatly when you buy the chinese leds.

I believe you mean product not LEDs. I also believe you mean Taiwanese not Chinese. Could be that the final product is Chinese made not the LEDs.

 

[bhazard451]

I believe you mean product not LEDs. I also believe you mean Taiwanese not Chinese. Could be that the final product is Chinese made not the LEDs.

That's the thing. Real Epistars are from Taiwan. There is no guarantee you are getting real anything. AC-RC seems to be legit and provides the spec sheets, but not all sellers are like this.

Another certain company that shall remain nameless tried to pass off their own Guangdong produced leds to me as Cree XP-G/XP-E. Unless you had a Cree of your own to compare with, you would never know by just looking at it.

Unless you are getting exact bin specs, and an honest seller, results will vary greatly when dealing with chinese companies.

 

[ronreef]

That's the thing. Real Epistars are from Taiwan. There is no guarantee you are getting real anything. AC-RC seems to be legit and provides the spec sheets, but not all sellers are like this.

Another certain company that shall remain nameless tried to pass off their own Guangdong produced leds to me as Cree XP-G/XP-E. Unless you had a Cree of your own to compare with, you would never know by just looking at it.

Unless you are getting exact bin specs, and an honest seller, results will vary greatly when dealing with chinese companies.

Yes. Understood.

 

[ronreef]

Another few shots of the build. This time, a ruler is included in the shots as a reference:
https://picasaweb.google.com/103048...&authkey=Gv1sRgCOe02_LC2fbiBA&feat=directlink

As you can see, it's not a large build. The Arctic Cooling Accelero L2 Plus is rated for 120W. Heat dissipation should be sufficient as long as the fan is going. Having the build mounted on a Kessil gooseneck provides flexibility to angle the light. Over my 24"x24"x21" tank, I'll be able to fit a few over the tank if needed, and with lots of head room to spare. Being able to angle the light is a plus.

 

[mr.wilson]

The HEAT transfer to an ideal heatsink is purely a function of the electrical efficiency of the emitter. Lower overall electrical efficiency means less energy is converted to photon energy and more to heat at the die.

The real world differs due to the heat density of the multi-chip vs the single emitter. When heat is spread over a larger footprint, it is easier to move away from the critical area. The high density chips need much better thermal management (heatsinking) due to the density.

Multichip users employ active cooling while array users use passive cooling (with fans in most cases). The active cooling makes up for the efficiency and footprint issues.

As far as efficiency goes, a 100w multichip will replace a 120-140w array from my experience. The combination of large dome lenses and single light source make it a better choice IMHO.

It would be nice to run these multichips without a fan, but we aren't quite there yet. I have a few spare chips I will use to test some advanced cooling materials. BTW, I accidentally drilled a hole in one of the heat pipes on my Tuniq 120 Tower CPU cooler. There is no liquid inside??? I don't now if it evaporated, leaked out earlier, or if they simply don't use liquid???

 

[mr.wilson]

To complete the journey into OT land...

12x24 Hammond NEMA enclosure, 18 circuits, 18 breakers, 18 relays for 18 inputs from automation devices, 36 status indicators to status of 18 breakers and 18 outputs with 18 HOA switches for override and 6 GFCIs with dedicated indicators and... of course 18 current monitors.

Why? I have no idea, but I enjoy projects like this. Some of the bits and pieces..

How does this compare to microcontrollers such as Arduino?

 

[tomservo]

mrwilson: That heatpipe probably contained water under reduced pressure (IE no air inside only liquid water and vapor). Most of the CPU type heatpipes use water, although ammonia and alcohol are also used. edit: Acetone is apparently pretty common too.

 

[smoothdog]

BTW, I accidentally drilled a hole in one of the heat pipes on my Tuniq 120 Tower CPU cooler. There is no liquid inside??? I don't now if it evaporated, leaked out earlier, or if they simply don't use liquid???

Many heat pipe designs have a wick or wicking design engraved inside the pipe used to move the fluid once it condenses. This may prevent it from just leaking out through a hole. Alternatively, some fluids used, such as ethanol, may just evaporate.

 

[BeanAnimal]

How does this compare to microcontrollers such as Arduino?

This is just the "switch box" that holds the high voltage components and relays. It is (for lack of a better analogy) the relay strip that plugs into typical hobby controller like an RK.

The difference is, that my receptacles are spread over the room via conduit that will feed back into this box and attach to the 18 relays. Each relay can be switched to ON, OFF or AUTO by corrosponding front panel switch. In ON mode the relay is bypassed (overriding the outside control logic) and 120V sent to the receptacle. In AUTO mode, the relay is inline and controlled by outside logic (the aquarium controller, PC, whatever). OFF mode, disables the receptacle, regardless of what the controller does.

It is just a fancy (expensive as crap) switch box with lights