DIY LED Cluster?

Yodeling

Premium Member
I'm getting ready to start my DIY LED project, and thinking about the LED arrangement. I'm thinking the closer the LEDs are to each other, the better (blending and shimmer). So I'm wondering if the experts here would recommend packing about 50 LEDs closely together in a cluster. By my calculations, it would come out to be about 6x6 inches area. Heat is the main issue I'm worried about. I was also thinking of using a single reflector similar lumenarc for each of these clusters. Thoughts?
 

reefermad619

New member
I'd ditch the 50 LEDs and instead use bigger led chips. Look up "evil cluster" its what you want to do but with far less soldering. Which to me is well worth it.
 

ReefLedLab

In Memoriam
You will have serious problem with such dense cluster. I think, has no necessity to design such big cluster. I sure, for you will be useful this article http://www.advancedaquarist.com/2014/4/lighting

In any way please pay attention to thermal design. Only SinkPAD MCPCB can provide good enough thermal conditions for dense LED cluster. Please note - no any usual MCPCB, even made from copper, can't give similar result.
 

oreo57

Active member
You will have serious problem with such dense cluster. I think, has no necessity to design such big cluster. I sure, for you will be useful this article http://www.advancedaquarist.com/2014/4/lighting

In any way please pay attention to thermal design. Only SinkPAD MCPCB can provide good enough thermal conditions for dense LED cluster. Please note - no any usual MCPCB, even made from copper, can't give similar result.

Have to agree on the heat issue ..as an example "makers" heatsinks can handle 60LED per sq. foot.. 5x the area you are talking about.. and that is ignoring any thermal transfer issues.. as mentioned above..

AFAICT you can compensate by having a much denser 'base" or by also limiting drive current which to a certain degree defeats the purpose.. but hey.. pushing the envelope is part of the game.. ;)

sinkpads are efficient at pulling heat from the LED "core" but this is really most important w/ overdriving .. once the heat is to the "heatsink" it is a matter of efficiently removing it from there..

IF you gave me a choice of using sinkpads plus Al heat sink or "regular" pads and a copper heatsink. keeping sq in and not overdriving I'd bet more on the latter Cu+ normal....

would be an interesting experiment..

http://en.wikipedia.org/wiki/Thermal_management_of_high-power_LEDs
 
Last edited:

ReefLedLab

In Memoriam
IF you gave me a choice of using sinkpads plus Al heat sink or "regular" pads and a copper heatsink. keeping sq in and not overdriving I'd bet more on the latter Cu+ normal...
Yes, we done lot of tests of usual MCPCB. Please read about first version of 12x LED assembly, that has reviewed in article. After change design of usual MCPCB to SinkPAD we have received approximately 25C temperature fall on crystals of LEDs.
 

oreo57

Active member
Yes, we done lot of tests of usual MCPCB. Please read about first version of 12x LED assembly, that has reviewed in article. After change design of usual MCPCB to SinkPAD we have received approximately 25C temperature fall on crystals of LEDs.

All well and good.. but I'm not seeing where they were attached to the heatsink (probably were) when you measured them. not seeing the drive current or not seeing that you used copper heat sinks anywhere and I assume they were all aluminum..
copper is much better than Al in thermal conductivity.

not trying to criticize your efforts just looking at variables..

That said ... sinkpadded LEDs heat sinked to copper would be even better.. ;)

Though expensive..

All commercial design is almost always a tradeoff ....

25C is impressive though..
 

Yodeling

Premium Member
Thanks folks! The heat issue is what I was afraid of, and it looks like perhaps the solution to producing shimmer lies elsewhere. Btw, producing halide-like shimmer is one of two main reasons I'm doing this. I currently have decent LED fixtures and they have so many advantages over halides, but the shimmer is not there (or is different).

So I like the idea of the "evil cluster", and using larger LEDs. Lucky I haven't invested in the LEDs themselves yet, and the LDD drivers I purchased should be able to drive the Bridgelux BXRA's or similar large chips (just glanced at the specs). Btw, feel free to suggest better alternatives to the Bridgelux. I already scrapped the idea of using the big 100W multi-chips as the only source of light because of lack fine-tunability and also complexity of cooling, but I'm thinking using large chips for white light, and smaller ones for UV/RB/Blue/Cyan/etc should still produce similar shimmer to the traditional 10k Halides with VHO actinics. Correct me if I'm wrong on any of this.
 

oreo57

Active member
There are these:
lowres_LED_nov25_015__43522.1385941842.1280.1280.jpg
 

oreo57

Active member
Yes, we done lot of tests of usual MCPCB. Please read about first version of 12x LED assembly, that has reviewed in article. After change design of usual MCPCB to SinkPAD we have received approximately 25C temperature fall on crystals of LEDs.

Getting back to heatsinks a bit........

Heat sinks manufactured using the method and a combination of aluminum alloys with a copper base have been shown to have superior thermal performance (at approximately four times better thermal conductivity) than more typical, commercial extruded, machined, and die-cast heat sinks.

http://www.digikey.com/en/articles/techzone/2011/oct/advances-in-led-thermal-management
 

ReefLedLab

In Memoriam
All well and good.. but I'm not seeing where they were attached to the heatsink (probably were) when you measured them. not seeing the drive current or not seeing that you used copper heat sinks anywhere and I assume they were all aluminum..
copper is much better than Al in thermal conductivity.
It is very important point. I'll try to explain. Since the crystal is quite small (a power LED's crystal usually has the surface of only 1-2mm2 or 0.0015-0.003in2), its generated heat density is quite large. In fact, it is about 30 times as large as the heat density through the soleplate of a household iron! So removing heat from LED heatspreader is critically important. We should use best possible technology for do this, at least for dense LED assemblies.

After removing the heat from LED heatspreader we should to distribute it evenly across whole heatsink. You right, material with best thermal conductivity will better for this. But we have technology, that provide the MORE best thermal conductivity than all material, include copper. It is heatpipes. It's thermal conductivity is more than 100 times more than copper. We have used for testing this heatsink http://reefll.com/index.php?route=product/product&path=65_69&product_id=69
In short word, combination of SinkPAD MCPCB and heatsink with direct contact of heatpipe to MCPCB will provide the best possible thermal mode for LEDs.

That said ... sinkpadded LEDs heat sinked to copper would be even better.. ;)

Though expensive..
Agreed, it is not cheap. But good thermal mode for LED will provide not only long lifetime, but also best efficacy. Depend on type of LEDs the difference may be 10-50%. Also, with using the best way, described above, we can design fixture with passive, i.e. fanless, cooling. By the way, temperature of LED crystal on SinkPAD MCPCB, installed on that heatsink with heatpipe, on 100% current is around 60C. I think, it is awesome :)

All commercial design is almost always a tradeoff ....
Exactly right. But if our target not a cheapest prime cost, but best solution, we should use best technology. To be honest, it is not very pricey ;)
 

ReefLedLab

In Memoriam
Getting back to heatsinks a bit........
Agreed with Digikey. Most of extruded heatsink is use not pure aluminium, but aluminium alloys with worst thermal conductivity. Because of this, pure copper will have thermal conductivity up to 4 times better. But I think, full-copper heatsinks or heatsinks with copper bottom it is not good way. In my opinion, holy graal of designing the LED fixture it is passive cooling. Only this way can provide 100% reliability. For this way we should not use any type of usual heatsinks, even if with copper bottom. This solution will heavy, heatsink will big, fixture will ugly. But if we will use proper heatsink, we can construct such fixtures: http://www.reefcentral.com/forums/showthread.php?t=2407970
 

zachts

New member
If you want dense point source lights using LED no mater what chip or array you decide on heat disipation is the bigest problem and the more LEDs in tighter space makes it dificult to DIY due to needing a thermal cooling solution designed for such use. There are a few available such as the nuventix synjet assemblies or the heat pipe coolers like ReefLedLab mentioned that are capable of tremendouse cooling for point source arrays even in passive mode, or Rediculous amounts of cooling using the synjet air movers or fans (incidently the synjets are designed for LED and rated for 100,000hours of operation unlike a fan) the big assemblies are espensive but if you are looking to create dense point source lights to simulate a MH these cooling solutions might work for you. Some of them even look pretty sleek just hanging by themselves.

Just a thought.

something like these monsters could cool a whole lot of ReefLedLab arrays, or any other type for that mater.

Mid size passive cooling, more than enough to rival a 250watt MH in similar foot print as a pendant:
http://www.nuventix.com/led-heatsink-r186-70/

Crazy amounts of cooling to rival a 1000watt MH with the synjet coolers:
http://www.nuventix.com/led-heatsink-r150-170/

These are also nice (budget freindly) and can easily handle 36volts worth of LEDs at 1 to 1.5amps with passive cooling depending on the LED chips used(add a fan or synjet and you can easily double the cooling power:
http://www.nuventix.com/downlight-led-heatsink-48/
 
Last edited:
Top