Heatsinks for DIY LED lamps

captainmorgan

Well-Known Member
Hey Supra,what do you think of the new 12" heatsink from heatsinkusa? I was thinking of 3 Z4's at 1400 ma on a 12" x 12". I'm currently running 3 Z4's at 1400 ma on Arctic 11's over each plant in my current set up and they kick butt. Thought a single large heatsink might simplify that set up. Would it work passively in case of a fan failure? I was thinking one fan in the center.
 

SupraSPL

Well-Known Member
It is always good to have lots of variety to choose from, but after my experience with the 10.08" profile, I realized that the fins are essentially running the wrong way (it requires a gigantic fan to reach all the channels) and the channels are short. The same thing goes with the 12" profile. I figure that blowing the air down as long of a channel as possible is a more efficient use of fan power. Also, a longer heatsink gives us a better spread. I have come to realize that with the 10.08" profile as well, I needed a bit more spread. 156w of CXA3070s on a 12" X 12" would be very high intensity concentrated in a small space. I figure ideally they would be spaced 16-20" apart?

So when it comes to passive cooling, I like the 12" better than the 10.08" because it has shorter fins, I assume that would make for better convection and make better use of the circulation fan in the grow space and of course it would give a better spread.

Using the 90cm²/W figure for passive cooling, we would need 25" length of the 12" profile for (3) CXA3070s at 1.4A. BUT, I think that we could get away with much less if we get assistance from the circulation fan and I have some of a similar profile on hand to test that theory soon. (8.46" X 12") with a pair of CXAs at 1.45A (106W) passive cooling (45cm²/W) and we will see how that performs. If that doesnt work I will try adding the tiniest fan and see what effect that has.

Again I think they would be spaced too closely, but it would be for testing purposes.
 

captainmorgan

Well-Known Member
It is always good to have lots of variety to choose from, but after my experience with the 10.08" profile, I realized that the fins are essentially running the wrong way (it requires a gigantic fan to reach all the channels) and the channels are short. The same thing goes with the 12" profile. I figure that blowing the air down as long of a channel as possible is a more efficient use of fan power. Also, a longer heatsink gives us a better spread. I have come to realize that with the 10.08" profile as well, I needed a bit more spread. 156w of CXA3070s on a 12" X 12" would be very high intensity concentrated in a small space. I figure ideally they would be spaced 16-20" apart?

So when it comes to passive cooling, I like the 12" better than the 10.08" because it has shorter fins, I assume that would make for better convection and make better use of the circulation fan in the grow space and of course it would give a better spread.

Using the 90cm²/W figure for passive cooling, we would need 25" length of the 12" profile for (3) CXA3070s at 1.4A. BUT, I think that we could get away with much less if we get assistance from the circulation fan and I have some of a similar profile on hand to test that theory soon. (8.46" X 12") with a pair of CXAs at 1.45A (106W) passive cooling (45cm²/W) and we will see how that performs. If that doesnt work I will try adding the tiniest fan and see what effect that has.

Again I think they would be spaced too closely, but it would be for testing purposes.

I had a fan idea that might solve that. If I put a cover on top of the fins with a chamber on one end then mounted one of these fans from Steve's to the chamber and blow air from one end to the other along the fins. Heatsinkusa is actually only about two hours away from me and about one hour from a place I visit every now and then.

http://shop.stevesleds.com/Pressurizing-Cooling-Fan-with-Power-Regulator-Pressurizing-Cooling-Fan-w-plug.htm
 

SupraSPL

Well-Known Member
That is true, and from experimenting in my veg area, you can simply hang a fan some distance above a bunch of heatsinks and it can cool them all very well while also providing some circulation for the canopy.
 

captainmorgan

Well-Known Member
I like the idea of ducting (sucking or blowing) air through a chamber but baulk at the thought that the heat will be additive along the heatsink (or light 'bar'), hotter for each cob away from the inlet. This may not be an issue effecting efficiency that much other than in theory?
With that 12" x12" heatsink and only 3 COB's it wouldn't be channeling the heat from one COB over another. Two COB's would be on one end near each edge and the third would be in the center on the other end. I pretty sure that fan moves a good amount of air. Haven't sketched out the spacing yet,it may end up 12x14 or something like that.
 

churchhaze

Well-Known Member
This is the same reason I like the 10" profile. The 12" profile looks like an extended version of the 10" profile.

Given the same total surface area, shorter fins will have lower thermal resistance than longer fins simply because resistance is proportional to length. There are diminishing returns on increasing fin surface area by using taller fins.

Also, what you're saying about convection is also probably true. You'd need wider spacing of the taller fins to have similar temperature and shape hot air pockets between the fins.

That being said, I think that the 10.08" (and 3.95") profile is meant to be oriented so the fins are pointing sideways, so that cool air is sucked from below.... Obviously that's no good for LED growers unless it's side lighting.

So when it comes to passive cooling, I like the 12" better than the 10.08" because it has shorter fins, I assume that would make for better convection and make better use of the circulation fan in the grow space and of course it would give a better spread.
 

bicit

Well-Known Member
Just a hunch, but I think the 4.85 profile from HSUSA is going to be a sweet deal for most DIY builders. It's significantly cheaper than all of the other profiles for the amount of surface area. I wonder how thick of a base is really necessary to provide adequate cooling?
 

getsoutalive

Well-Known Member
Wanted to make sure that this was understood by everyone on the board, have not really seen it discussed much. It is possible to quite accurately and very simply determine the proper size required of a given heat sink design with the data available on the board here.

Most heat sinks will have a stat listed as Thermal Resistance or C/W (or in the case of HUSA C/W/3). This is the expected rise above ambient, assuming static air, per watt of heat applied.

So for instance, you have a single 3070 3K AB chip and you want to drive it at 1050ma. From the spreadsheets provided by @SupraSPL and others, we know that the chip will be approx 45% efficient at 50c. That means that it will be converting 55% of the 38 watts it is being fed into heat. So our chip will be applying about 21 watts of heat to our sink.

For use as a passive sink, the 5.375" model is one of the best choices from heatsinkusa.com. It has a C/W/3 of 1.15. That means that a three inch piece of sink will rise in temp 1.15 degrees above ambient for each watt applied.

If we assume an ambient of 25c, we can expect a three inch piece of the 5.375 to remain below 25+21*1.15 or about 50c. Again, this is in a static air environment. Your canopy fans will help to significantly reduce the effective c/w and keep operating temps much lower. But you can rest assured that even if your fans failed, your chips would be safe. Of course, ambient is not static and additional headroom on a heatsink is never a bad thing.
 

Bueno Time

Well-Known Member
Tested out a 50mm fan on one of my 2.08" profile veg light just put in the middle of the bar blowing down into the channels with blue painters tape over the top of the fins forcing the air down the length of the bars and cranked up the dimmable driver. Left it for about half hour or so and the one without the fan was hot while the one with the 50mm fan at 12v was cool to the touch about skin temp or less on the side fins and the bottom plate near and between the LEDs themselves was probably comparably as cool to the touch as my 3070 ABs @1450ishmA on Alpine 11s at 12v, so effectively cooled. The one without the fan installed the temp kept rising until I stopped the test as the one with the fan would not gain any additional heat after the first couple minutes of testing the temp stopped rising while it was getting hotter and hotter on the passive heatsink over time.

I was impressed enough with my ghetto test that I ordered 8 more of the same 50mm fans last night, going to try 1 on each bar on my new flowering light build, if not cool enough to my liking then two 50mm per heatsink bar. Pretty sure 1 will be effective enough though cooling ~38w per 32" long 2.08" profile at max and less if dimmed.
 

SupraSPL

Well-Known Member
@getsoutalive I agree that the 5.375" is a nice heatsink profile with a thick base and relatively cheap/surface area. But the 5.88" profile has a lot more distance between the fins, so I have been thinking that profile might ideal for passive cooling. I have never paid much attention to the C/W statistics because I figured there were too many variables in our situation, such as the thermal resistance of the LED package, drive current relative to maximum, quality of the thermal interface, the orientation of the fins, ambient air movement and type of LEDs used. For the example you gave, that is only 28cm²/W, which is 1/3rd what I normally recommend for passive cooling. Of course times are changing now that we can use white COBs at high efficiency and completely skip the reds, so I think our surface area numbers can be reduced much lower than 90cm²/W, thankfully.
 

getsoutalive

Well-Known Member
Sure, there are many ways to accomplish the same or sometimes very similar tasks, mine was simply an example for the sake of the math.

The 5.88" has a 1.75 c/w/3" so a nearly square 6" slab would keep that same 3070 above closer to 45c. The three inch piece would cost $7.56 and weigh 1.11 lbs while the 6" would run $10.44 and weighs 1.5 lbs. Its the thicker base and extra mass that allows for the smaller surface area on the 5.375.
 

Bueno Time

Well-Known Member
Wow, 50mm fan cooling almost 3', that sounds like a very efficient solution BT :leaf:
Ya, I tried it on a 24" heatsink of my veg light but it was running at 1-1.1w dissipation per inch of 2.08" profile and staying cool with the 50mm so Im going to try on 32" long at 1.15w dissipation per linear inch of heatsink. Then I can see if the single 50mm is still effective for the extra 4" in each direction which I think will be fine just have to test it on the actual setup now.

Still undecided on the K temp of V10s. I had bought all the different V10 80CRI K temps to help pick a spectrum but now after thinking more I am realizing its not the best way to choose the spectrum by what it looks like to my human eye as I was looking at the Bridgelux spectrum graphs more and seeing that the 4000K looked really white to me I thought it didnt have enough red but now I see that the 3000K and 4000K 80CRI versions have about the same red just a bigger blue peak on the 4000K so it looks white to my eye like not enough red is there but really its just got extra blue that balances the red more to the human eye. Hmm... And then I saw a spectral analysis sheet of Vero 18 2700K 80CRI vs Vero 18 3000K 80CRI supposively from Bridgelux that the 2700K has more red and blue than the 3000K and more overall output in nearly all areas of the curve by a small amount but still find that odd somehow the more warm color having more output and more blue...

Im getting there though just a little more research and thought before I commit to ordering.
 

PurpleBuz

Well-Known Member
Tested out a 50mm fan on one of my 2.08" profile veg light just put in the middle of the bar blowing down into the channels with blue painters tape over the top of the fins forcing the air down the length of the bars and cranked up the dimmable driver. Left it for about half hour or so and the one without the fan was hot while the one with the 50mm fan at 12v was cool to the touch about skin temp or less on the side fins and the bottom plate near and between the LEDs themselves was probably comparably as cool to the touch as my 3070 ABs @1450ishmA on Alpine 11s at 12v, so effectively cooled. The one without the fan installed the temp kept rising until I stopped the test as the one with the fan would not gain any additional heat after the first couple minutes of testing the temp stopped rising while it was getting hotter and hotter on the passive heatsink over time.

I was impressed enough with my ghetto test that I ordered 8 more of the same 50mm fans last night, going to try 1 on each bar on my new flowering light build, if not cool enough to my liking then two 50mm per heatsink bar. Pretty sure 1 will be effective enough though cooling ~38w per 32" long 2.08" profile at max and less if dimmed.
Impressive the cooling you got from a small fan. What 50mm fan was that ? How many watts was it drawing ? Trying to get a handle on how much air flow is going through the heatsink....
 

SupraSPL

Well-Known Member
I had a pair of Vero 29s and a pair of Rosewill RCX-Z1 heatsinks so it was another good chance to check the effect of polishing the heatsink and using an expensive thermal paste. I drove them at 2.583A and here are the results:
DSC07860a.jpg

Stock heatsink with stock paste
Cold: 38100 lux - 39.7 Vf
7.5V: 36900 lux - 39.1Vf (3.15% temp droop)
5V: 36600 lux - 39.03Vf (3.94% temp droop)

Polished to 600 grit US and used Prolimatech PK3 paste.
Cold: 39000 lux - 39.6 Vf
7.5V: 37700 lux - 38.99Vf (3.33% temp droop)
5V: 37400 lux - 38.98Vf (4.1% temp droop)

So the stock heatsink seems to have slightly outperformed the polished heatsink but I think that is just the margin of error in the "cold" lux and Vf, because they are tricky to measure. But the important thing is, another data point showing there is no point in sanding or polishing the heatsinks, even at high dissipation >100W. So I am convinced I wasted a lot of time sanding, but it was great exercise and I am very glad we do not need to do it :)
 

Abiqua

Well-Known Member
I had a pair of Vero 29s and a pair of Rosewill RCX-Z1 heatsinks so it was another good chance to check the effect of polishing the heatsink and using an expensive thermal paste. I drove them at 2.583A and here are the results:
View attachment 3320724

Stock heatsink with stock paste
Cold: 38100 lux - 39.7 Vf
7.5V: 36900 lux - 39.1Vf (3.15% temp droop)
5V: 36600 lux - 39.03Vf (3.94% temp droop)

Polished to 600 grit US and used Prolimatech PK3 paste.
Cold: 39000 lux - 39.6 Vf
7.5V: 37700 lux - 38.99Vf (3.33% temp droop)
5V: 37400 lux - 38.98Vf (4.1% temp droop)

So the stock heatsink seems to have slightly outperformed the polished heatsink but I think that is just the margin of error in the "cold" lux and Vf, because they are tricky to measure. But the important thing is, another data point showing there is no point in sanding or polishing the heatsinks, even at high dissipation >100W. So I am convinced I wasted a lot of time sanding, but it was great exercise and I am very glad we do not need to do it :)

Just picked up Pk1 as it was 1) cheap as shiat 2)same thermal conductivity as Pk2 and Pk3, just that thermal resistance is .017 .015 .013, going from pk1 to pk3 :peace:

Would be interesting to see how resistance comes into play.....
 

FrozenChozen

Well-Known Member
So the stock heatsink seems to have slightly outperformed the polished heatsink but I think that is just the margin of error in the "cold" lux and Vf, because they are tricky to measure. But the important thing is, another data point showing there is no point in sanding or polishing the heatsinks, even at high dissipation >100W. So I am convinced I wasted a lot of time sanding, but it was great exercise and I am very glad we do not need to do it :)
I knew it wasn't just laziness on my part! cleaning/wiping is all I have ever done!
 
Top