Heatsinks for DIY LED lamps

[Abiqua]

Really interesting to see the Rosewill's performance. I know you mentioned it previously, but to see it layed out. It is an interesting tidbit.

Fyi, I honestly thought the Rosewill's I linked, were similar to the one's that you had actually tested [aka in the graph]. Still, with your testing, I would have no problem trying out the NewEgg version. Those look quite like the A11's, but definitely curious to see if performance could be just that nth greater....

 

[SupraSPL]

Also worth mentioning, I did not prepare those heatsink surfaces in any way. I used the stock paste, except for the Intel, which used PK3. In earlier tests at 1.4A, I saw zero difference between a stock A11 with stock paste and a sanded, polished A11 with PK3. So we can actually gain more efficiency by optimizing our fan speed than by preparing the surface. But, at very high currents that may not be the case.

 

[Dloomis514]

If the heat sink fan references a cpu, you can usually find the heat dissipated. Most are over 50 watts which works for us

 

[Abiqua]

If the heat sink fan references a cpu, you can usually find the heat dissipated. Most are over 50 watts which works for us

http://en.wikipedia.org/wiki/Comparison_of_Intel_processors
http://en.wikipedia.org/wiki/Comparison_of_AMD_processors

I really like anything from Socket 423/478/775/T/M/P etc. for Intel stuff. 478/775 are usually what I look for first. P4's were power hungry and all the way up the Conroe Quad [Kentsfield] pushed 135+ watts. Xenon based chips are higher...
Not a huge user of AMD in the past, but anything from Thunderbird to QuadCore's are powerful enough...

I would watch with some of the newer mid power cpu's as they have been getting more efficient power wise and decreasing in size. Some of these coolers should be sought out with caution. The older stuff until from a point, was power hungry....>P3 Intel >Thunderbird cores AMD

 

[AquariusPanta]

HeatSinkUSA is bringing out a new 12" wide HS in the coming month. O fuck yeah!!!

 

[Bueno Time]

Thinking about testing out how much more cooling capacity the 2.08" profile has when actively cooled with 50mm fan(s) vs passive. I would blow the fan down into the fins and have a cover over the top to direct the airflow from the fan through the fins and out the sides of the heatsink bars, like the Alpine 11 but much longer and skinnier. Im going to test 1 50mm fan on a 23" long bar and one without the 50mm fan purely passive cooling, both bars are on one single string on a dimmable driver and I cranked it up to around 50w dissipation (close to 60w on kill-a-watt) the other day and forgot about it for about 20 minutes or so and the bars were pretty warm kinda hot but not burning hot either. Still too hot for my liking, trying to test out different theories before I do my next build, still the plan is 35x Vero 10s plus reds/blue most likely just the ones I already have made they seem to put out plenty of extra red for my little tent.

Anyway Im working out the heatsink part of the equation, if I didnt care about spending a lot I would just get the 3.5" profile but thats overkill and could be good for passive cooling my intended setup but I am a cheap ass and passive cooling is much more effective allowing the use of less aluminum (cheaper). If the 50mm fan test goes pretty well on the 23" bar of my veg light then I will try 2x 50mm on 32" long bars with 7 Vero 10s each @ ~200-215mA each (max).

Sorry if everyone getting sick of me rambling about my testings for a new build but figure its relevant if the 50mm fans work well on the 2.08" profile it could possibly be helpful to others too. Im not going to have hard data in numbers like Supra but I can tell you if it helps a lot or a little lol.

 

[AquariusPanta]

Thinking about testing out how much more cooling capacity the 2.08" profile has when actively cooled with 50mm fan(s) vs passive. I would blow the fan down into the fins and have a cover over the top to direct the airflow from the fan through the fins and out the sides of the heatsink bars, like the Alpine 11 but much longer and skinnier. Im going to test 1 50mm fan on a 23" long bar and one without the 50mm fan purely passive cooling, both bars are on one single string on a dimmable driver and I cranked it up to around 50w dissipation (close to 60w on kill-a-watt) the other day and forgot about it for about 20 minutes or so and the bars were pretty warm kinda hot but not burning hot either. Still too hot for my liking, trying to test out different theories before I do my next build, still the plan is 35x Vero 10s plus reds/blue most likely just the ones I already have made they seem to put out plenty of extra red for my little tent.

Anyway Im working out the heatsink part of the equation, if I didnt care about spending a lot I would just get the 3.5" profile but thats overkill and could be good for passive cooling my intended setup but I am a cheap ass and passive cooling is much more effective allowing the use of less aluminum (cheaper). If the 50mm fan test goes pretty well on the 23" bar of my veg light then I will try 2x 50mm on 32" long bars with 7 Vero 10s each @ ~200-215mA each (max).

Sorry if everyone getting sick of me rambling about my testings for a new build but figure its relevant if the 50mm fans work well on the 2.08" profile it could possibly be helpful to others too. Im not going to have hard data in numbers like Supra but I can tell you if it helps a lot or a little lol.

I wanna see your 35x VERO 10 COB fixture when it's all done!!!

Feel free to dump some photos of the project here

 

[SupraSPL]

For the Intel w/copper base, you have 5V being the lowest percentage or the most efficient amongst the fan's potential range. Was that a typo/error/misreading or a phenomena?

And yes, the intel fans that I have are strange birds. Maybe yours are stranger though.

Sorry I missed your reply somehow. Strange I know, but not a typo. With the cooling performance and fan power factored in, it was most efficient at 5V. To some extent that is because the power draw was wonky at 7.5V and 9V. Or maybe you could say it was wonky at 5V. There must be some kind of circuitry in those fans interfering somehow. I did not see any kind of temp sensor in the airstream.

 

[Dloomis514]

My take away from the 5v issue is the COBs studied don't need that much air flow over a heatsink to work satisfactorily. But they do need airflow or risk frying

 

[SupraSPL]

That about sums it up. A slight bit of air movement makes a huge difference for CPU coolers. The heatsinks with wider spacing and shorter fins might be able to get along better with low airflow, benefiting more from the circulation fan. Heatsinks with vertically oriented fins should do even better still due to greater convection.

The circulation fan makes a significant difference for passive cooling. For example I tested two identical heatsinks in the same flowering tent. One was running at 85cm²/W and the other was at 95cm²/W. You would expect the 85cm²/W to be running hotter, but it was getting more airflow from the circulation fan so it was running 2 degrees cooler than the 95cm²/W heatsink. The implication is that if we design the system for passive cooling and make sure the circulation fan is doing double duty, we could get away with a lot less surface area, greatly reducing the cost and weight of the build. That may be especially true for the heatsinks with wide spacing and short fins.

 

[DonPetro]

That about sums it up. A slight bit of air movement makes a huge difference for CPU coolers. The heatsinks with wider spacing and shorter fins might be able to get along better with low airflow, benefiting more from the circulation fan. Heatsinks with vertically oriented fins should do even better still due to greater convection.

The circulation fan makes a significant difference for passive cooling. For example I tested two identical heatsinks in the same flowering tent. One was running at 85cm²/W and the other was at 95cm²/W. You would expect the 85cm²/W to be running hotter, but it was getting more airflow from the circulation fan so it was running 2 degrees cooler than the 95cm²/W heatsink. The implication is that if we design the system for passive cooling and make sure the circulation fan is doing double duty, we could get away with a lot less surface area, greatly reducing the cost and weight of the build. That may be especially true for the heatsinks with wide spacing and short fins.

I am all for passive cooling. Was thinking 4 or 5 vero10 on a 10" x 18" slab from hsu for my 19" x 24" x 48" cab with a clip-on fan directed above it.

 

[SupraSPL]

What drive current do you have in mind? Napalm and I were just working out a design for Vero 10s vegging and it is surprising how cheap we were able to get away with for passive cooling using the heatsinkUSA 2.08" profile. It is the cheapest surface area they offer I believe (because of the thin base plate .1").

 

[DonPetro]

What drive current do you have in mind? Napalm and I were just working out a design for Vero 10s vegging and it is surprising how cheap we were able to get away with for passive cooling using the heatsinkUSA 2.08" profile. It is the cheapest surface area they offer I believe (because of the thin base plate .1").

I was thinking of driving them at 300ma. I was also thinking about bars using the 3.5" profile for a 4x8 space.

 

[DonPetro]

What drive current do you have in mind? Napalm and I were just working out a design for Vero 10s vegging and it is surprising how cheap we were able to get away with for passive cooling using the heatsinkUSA 2.08" profile. It is the cheapest surface area they offer I believe (because of the thin base plate .1").

I would be interested to see what you come up with.

 

[SupraSPL]

Yep same current I had in mind. So at 300mA they dissipate about 8W ea. At 85cm²/W we would need 680cm².
That is about 8" of the 2.08" profile for each Vero 10, cost $2.56.
If we use the 3.5", we would need 5.25" per Vero 10, cost $6.

It would be an interesting test to see how they compare. The 2.08" has shorter fins but closer together. So I am not sure if one would outperform the other. Also the 3.5" has a much thicker base .3" vs .1".

 

[DonPetro]

Yep same current I had in mind. So at 300mA they dissipate about 8W ea. At 85cm²/W we would need 680cm².
That is about 8" of the 2.08" profile for each Vero 10, cost $2.56.
If we use the 3.5", we would need 5.25" per Vero 10, cost $6.

It would be an interesting test to see how they each perform. The 2.08" has shorter fins but closer together. So I am not sure if one would outperform the other. Also the 3.5" has a much thicker base plate, .3" vs .1".

I didn't notice the cost if 2.08"; thats encouraging. Could do 5 on a 40" bar for $12. That brings cost down significantly.

 

[Dloomis514]

"The implication is that if we design the system for passive cooling and make sure the circulation fan is doing double duty"

Sounds a lot like a remote (not local) active cooling fan.

Why not have the room fan blow directly on the cob(s).

 

[SupraSPL]

I tried something along those lines in the vegging room and it worked great. I bought some heatsinks that turned out to have less surface area than advertised (an honest mistake and the retailer made good on it, CustomHydroNutes.com, awesome people) So the heatsinks were running hotter than expected. I hung a high RPM 140mm fan a few feet above all the heatsinks and ran it a 5V. It created a nice breeze in the canopy and knocked the temp of all the heatsinks down to almost ambient.

Next experiement will be to see if I can do away with my 65W ceiling fans in the flowering rooms. They work great but I suspect I can do better with less watts and cool the heatsinks better. Not that they really need it at 32C, but I will take what efficiency I can get.

 

[Dloomis514]

What drive current do you have in mind? Napalm and I were just working out a design for Vero 10s vegging and it is surprising how cheap we were able to get away with for passive cooling using the heatsinkUSA 2.08" profile. It is the cheapest surface area they offer I believe (because of the thin base plate .1").

I wonder what the cost drivers are for various heat sinks, pounds of aluminum seem to the be the biggest factor. Die cost, labor, power etc would be pretty constant i bet. In fact i bet profit is bigger than those items combined.

 

[SupraSPL]

Ya you can see that expressed comparing the surface area/cost of the various heatsink prolfiles. The 2.08" profile has a lot of surface area for the money, but a much thinner base than most of the others.