Heatsinks for DIY LED lamps

salmonetin

Well-Known Member

Attachments

Last edited:

robincnn

Well-Known Member
I think the C/W standard we are used to seeing is based on a very large temp difference between the heatsink and ambient (70C), which increases the efficiency of the thermal transfer. They are also based on vertical orientation and passive cooling in a still room. But in our case the heatsinks are often only a few degrees above ambient and horizontally oriented, so the C/W is really just a guide.

Going back to 2008, KNNA recommended a minimum heatsink surface area/dissipation W. Since then, quality LEDs have improved efficiency a lot (risen from about 20% to 40% efficient) so we use a smaller number now. Approx 25cm²/W for active cooling and 75cm² for passive cooling. There is more to the story though, the shape of the heatsink, thickness of the baseplate, COB efficiency and air circulation in the space affects the outcome. The 75cm²/W figure is assuming about 40% COB efficiency, thick baseplate, wide short fins and slight air movement (ideal conditions).

Many LEDs are running in the 30-40% efficiency range but they can go as high as 70%, so if we work with surface area/W of heat we would have a more accurate guideline. 40cm²/W of heat for active cooling and 125cm²/W of heat for passive cooling. I have been reluctant to go that route because I don't want to confuse the issue.

Anyway using those guidelines you should get a heatsink temp that is just a few degrees above ambient and a temp droop from .5-4%. I try to measure the hottest part of the heatsink without letting the photon and infrared output of the COB interfere (often in between the fins, behind the COB). @Greengenes707 has pointed out that infrared thermometers can have a hard time with shiny objects. They can reflect infrared from other sources such as the COB or your own skin. Some IR thermometers seem to handle it better than others. I use a $15 baby food thermometer from Walmart and it does well with mildly shiny surfaces. If you are doing a lot of IR testing you can rough up the surface you are going to measure.
subjective..darn it. I was trying to beat the SDS 0.14 °C/W Tetras record. jk

'often in between the fins, behind the COB' makes sense. I will use my IR thermometer to do the same. Thank You Supra

Thanks for the excel. Was able to open with the password. Will try this too
 

SupraSPL

Well-Known Member
subjective..darn it. I was trying to beat the SDS 0.14 °C/W Tetras record. jk
You likely will. That design has a lot of COB power concentrated in a small area. But the Vero29s can take some heat so there is little to be gained by going overboard on cooling, much more effective to address current droop, reflectors/lenses and spread IMO.
 

DoctorDelta9

Well-Known Member
I'm looking at thermal grease. I know the pk-3 is what a lot of you are using. Anyone else had success with anything else? Also rough estimate on how much grease per cob? I need to know how much to order. After @SupraSPL post comparing thermal grease and bare I'm thinking that selecting a more expensive brand is not worth the money.
 

salmonetin

Well-Known Member
...vid only for examples or ideas...

...TIMs are used to ensure a strong thermal connection between the LED array substrate and the heatsink or fixture...

...In this video, we’ll cover three common styles of TIMs:

- Thermal Pads,
- Phase Change Materials, and
- Thermal Grease...


Saludos
 

robincnn

Well-Known Member
I'm looking at thermal grease. I know the pk-3 is what a lot of you are using. Anyone else had success with anything else? Also rough estimate on how much grease per cob? I need to know how much to order. After @SupraSPL post comparing thermal grease and bare I'm thinking that selecting a more expensive brand is not worth the money.
I thought artic silver was more popular here.
 

caretak3r

Well-Known Member
over the last year or two I've ordered a total of 6 arctic alpine 11 plus cpu coolers. I just received one a few days ago, and as I mounted a vero18 on it, I realized something was different. The raised platform that you attach the cpu/cob to, is now significantly smaller (maybe an inch or so less raised platform) The v18 still fits fine, but i used to be able to nearly fit 2 on 1 heatsink (with some of the plastic hanging over, but the metal bases of the cobs nearly entirely touching the aluminum). Just wanted to give you guys a heads up.
 

alesh

Well-Known Member
over the last year or two I've ordered a total of 6 arctic alpine 11 plus cpu coolers. I just received one a few days ago, and as I mounted a vero18 on it, I realized something was different. The raised platform that you attach the cpu/cob to, is now significantly smaller (maybe an inch or so less raised platform) The v18 still fits fine, but i used to be able to nearly fit 2 on 1 heatsink (with some of the plastic hanging over, but the metal bases of the cobs nearly entirely touching the aluminum). Just wanted to give you guys a heads up.
There are different revisions of almost all Arctic products.
 

testiclees

Well-Known Member
"embedded fan pocket" ???

does anyone have experience with or know how to calculate the efficacy of milling the sink to embed a fan.
 

SLITLOS

Well-Known Member
I would like to ask how flat are the products from HS-USA are? Not being able to order from the US on heavy items, freight and import taxes kill good toys, and leaves only the A-Exps.
But back to flat, I got 6 of the first picture and great, fast service by DHL(W/out ++ taxes) but after checking the HS's, the aren't very flat, see the rest of the pictures, also some defects are showing up
on further inspection, I basing this as I worked aircraft sheet metal and parts mfg., miles and miles of extrusions. I think I would have been money ahead if I had some bar stock milled into heat sinks, has anybody gone this route? I started a dispute claim with the storefront and will file with AE if my
HS are not normal.
Thanks
 

Attachments

churchhaze

Well-Known Member
I would like to ask how flat are the products from HS-USA are? Not being able to order from the US on heavy items, freight and import taxes kill good toys, and leaves only the A-Exps.
But back to flat, I got 6 of the first picture and great, fast service by DHL(W/out ++ taxes) but after checking the HS's, the aren't very flat, see the rest of the pictures, also some defects are showing up
on further inspection, I basing this as I worked aircraft sheet metal and parts mfg., miles and miles of extrusions. I think I would have been money ahead if I had some bar stock milled into heat sinks, has anybody gone this route? I started a dispute claim with the storefront and will file with AE if my
HS are not normal.
Thanks
HSUSA sinks are very flat, and have very good cuts.
 

stardustsailor

Well-Known Member
You likely will. That design has a lot of COB power concentrated in a small area. But the Vero29s can take some heat so there is little to be gained by going overboard on cooling, much more effective to address current droop, reflectors/lenses and spread IMO.
Well ...I place a challenge to that one ....
It takes more than what you think ...
(Tip : Firstly you should try to find a heatsink made out almost of pure aluminium ....Not gonna happen easily ..)

Little to be gained by going overboard in cooling ?
You might want to reconsider that one ...
Seriously ...

Afterall ,not all LED grow lights have to be the same .
Some are still having the physical size of a small "UFO" LED grow light ,
while at the same time, they pack tremendous output ...
It depends ,on what one wants to achieve .
How and where this fixture it's going to be used ...

Cheers.
:peace:
 

SupraSPL

Well-Known Member
I am not opposed to having excellent cooling, just pointing out that we have to be wary of the law of diminishing returns.

HeatsinkUSA is what we use for the most part, the alloy is 6063-T6, 209 W/m-K. Pure aluminum at 300Kelvin (80F) is 237 W/m-K. Not much potential for improvement there. Source
Aluminum thermal conductivity.png


If you are able to cool a Vero from Tc 50C down to Tc 35C, you gain 3% light/W. But if you try to push it further and cool down to Tc 30C, the total gain is 4%. If you had to increase fan speed to achieve that, the ellectrical gain may cancel out the output gain. If you had to spend more $ on heatsink to achieve it, the money may be more effective spent on running at a lower current.
Vero temp droop.png

If you run a Vero @1.4A (51W) instead of 2.1A (79W), you gain 7.6% light/W.
 
Last edited:

stardustsailor

Well-Known Member
I am not opposed to having excellent cooling, just pointing out that we have to be wary of the law of diminishing returns.

HeatsinkUSA is what we use for the most part, the alloy is 6063-T6, 209 W/m-K. Pure aluminum at 300Kelvin (80F) is 237 W/m-K. Not much potential for improvement there. Source
View attachment 3423444


If you are able to cool a Vero from Tc 50C down to Tc 35C, you gain 3% light/W. But if you try to push it further and cool down to Tc 30C, the total gain is 4%. If you had to increase fan speed to achieve that, the ellectrical gain may cancel out the output gain. If you had to spend more $ on heatsink to achieve it, the money may be more effective spent on running at a lower current.
View attachment 3423417

If you run a Vero @1.4A (51W) instead of 2.1A (79W), you gain 7.6% light/W.
Yeap ,correct ...
But ....

Lower current = more pieces of COBs ( + $$$ ) = larger heatsink surface = more aluminium = even more $$$ .
Higer current =less pieces of COBs (-$$$) =less heatsink material /surface (-$$$) = but better cooling system needed .

Generally ,when you "encase " COBs into a complete fixture ,by utilising more COB pieces and lower currents ,
overall weight and bulk of the fixture ,tend to become kinda of a limit ,themselves .....
There are situations that a neat and powerful , small sized fixture is needed ...
And not the whole Klingon Armada of Motherships ,hanged over the plants ...

Moreover , if one wants more COBs ,in a larger area ,then more neat ,small fixtures can be used , but driven at lower currents .
Still ,with superb cooling ,ensuring high outputs and long service life of each of the fixtures ...

2 Watts of fan power per fixture , ain't exactly cancelling out the output gain ...
But in fact ,they increase service life of COBs being cooled .



Cheers.
:peace:
 

SupraSPL

Well-Known Member
At least in the case of CPU coolers, fan power consumption easily erases the gain in temp droops. Almost every test was most efficient with fans running at 5V. An unexpected result but good news for those running at 12V.

I agree there is a time and place for enclosed fixtures and certain situations where you can't have them spread out. My original point is, a DIYer can achieve slightly better thermal management by spreading them out and will also enjoy efficiency gains by improving the spread in the canopy. You know me I don't mind the armada, at the end of the day it is about piles of nugs in the jar :)
 
Last edited:
Top