Active or Passive Heatsink?

[minorhero]

Hello folks,

I am in the process of building a light for a vivarium. I have all the parts and pieces ready and I'm starting the assembly. My question is regarding the heatsink. Specifically its 45" of this one right here. The light uses 11 COBs. 8 of those are BXRE-65S4001-C-74‎ - 34V at 1A. The remaining 3 are L2C5-30901204E13C0‎ 35V at 700ma. Here is a picture of them affixed with thermal adhesive to the heatsink:

So all that said, do I need fans on this heatsink or can I go without them? If I need fans, how many should I use? And for that matter.. how do you figure this out?

Thank you guys for any assistance!

 

[Bullzeyebill]

I'm sure that members will contribute to your request. It would be good if f you removed the direct link to that company, and use an image site.

Bill

 

[Katherine Alicia]

a Very quick estimate looks like you`ll be doing no more than 400W total, and I`v seen 1KW heatsinks a third of that size, so you`ll probably be fine!

 

[badtziscool]

Are there any specs on that heat sink with regard to its thermal capacity? Is this heat sink and led assembly eventually going into an enclosure? If so, will the back side be exposed to the open air or will it also be enclosed? There's lots of factors that will determine if you need a fan(s) or not.

 

[DIWdiver]

That heatsink is rated "approximately 2.15" C/W per 3" length. A 45" piece would be 15 times that long, so 1/15 the temp rise, or 2.15/15 C/W.

Since you have around 400 W, you'd expect 2.15/15*400 = 57°C temp rise. This is how much hotter the heatsink will be than the surrounding air. To get the heatsink temperature, add this number to the ambient air temperature. Keep in mind that the LED die temperature is what you are actually interested in, and this is considerably higher than the heatsink temp.

Their FAQ page says that the rating is for natural convection (no fan) with 70°C rise. That's probably in an orientation where the heat would cause air to rise through the channels. For this to work, your 45" dimension would need to be vertical. Unfortunately there's a point of diminishing returns when adding length, so your performance would be considerably worse, even in this optimal orientation.

If the heatsink is in any other orientation, the temperature rise will be considerably higher.

My conclusion would be that you probably do need some fans, though only enough to ensure a gentle flow of air through the fins. No need for any noisy, high power fans.

This assumes the air will be a reasonable 'indoor' temp - say 25-35°C. If it is significantly warmer, you'll need to make adjustments.

 

[minorhero]

Sorry, I looked for a list of forum rules and didn't find any. I didn't realize you are not allowed to post links to websites.

Here is a picture from the website as suggested:

Here is the information from the website regarding the heatsink:

[FONT=&quot]Extruded Aluminum Heatsink[/FONT]
[FONT=&quot]Alloy: 6063-T5
Width: 4.230"[/FONT]
[FONT=&quot]Fin Height: .75"[/FONT]
[FONT=&quot]Base: .300"[/FONT]
[FONT=&quot]C/W/3": approximately 2.15[/FONT]
Weight per Inch: .19

The heatsink will be exposed to the open air and will not be in an enclosure. I will probably try to put a diffuser or possible a lens on the COBs themselves but otherwise the heatsink will be in the open with only some wooden supports on either end to hold it up off the vivarium glass top. Please let me know if there is any other information needed.

 

[minorhero]

That heatsink is rated "approximately 2.15" C/W per 3" length. A 45" piece would be 15 times that long, so 1/15 the temp rise, or 2.15/15 C/W.

Since you have around 400 W, you'd expect 2.15/15*400 = 57°C temp rise. This is how much hotter the heatsink will be than the surrounding air. To get the heatsink temperature, add this number to the ambient air temperature. Keep in mind that the LED die temperature is what you are actually interested in, and this is considerably higher than the heatsink temp.

Their FAQ page says that the rating is for natural convection (no fan) with 70°C rise. That's probably in an orientation where the heat would cause air to rise through the channels. For this to work, your 45" dimension would need to be vertical. Unfortunately there's a point of diminishing returns when adding length, so your performance would be considerably worse, even in this optimal orientation.

If the heatsink is in any other orientation, the temperature rise will be considerably higher.

My conclusion would be that you probably do need some fans, though only enough to ensure a gentle flow of air through the fins. No need for any noisy, high power fans.

This assumes the air will be a reasonable 'indoor' temp - say 25-35°C. If it is significantly warmer, you'll need to make adjustments.

Thank you! This is super helpful! Would 2 fans be enough you think or better to go with 4?

 

[DIWdiver]

Depends on your fans, where you put them, etc.

Remember, the calculations said that with natural convection, you were doing pretty good, so you are only trying to overcome the length and orientation factors which are acting against you.

As long as you get enough air flow to simulate natural convection (which isn't very much), you should be good. You want to get air flowing over all the fins. Your heatsink is a little over 4" wide. I would guess a single 4" (100mm) fan mounted near the middle of the heatsink would probably suffice. Or two 2" (50mm) fans side by side to cover the width of the heatsink.

If you mount multiple fans along the length of the heatsink, there will be 'null' points between fans where the opposing forces cause zero air flow. Near the null points there will be low air flow. It's possible these regions would be small enough to be negligible, but I wouldn't want to hazard a guess about that.

In an ideal world, you would have the tools and expertise to model all this, and run a simulation to find out if you are okay. I don't run with that crowd. I would probably buy a fan or two, set it up and run it, especially given how far you already are along that path. Given the mass of that heatsink, nothing will happen super fast, and you can shut it down if it's getting out of hand. The initial temperature tests would be with my hands as measurement tools. If I was happy with the results, that would be the end of it. If not, I might put a thermocouple or thermistor on the hottest sections, and see if that satisfied me. Only if that failed (which I wouldn't expect) would I start thinking about ducting, better fans, multiple fans, etc.

 

[minorhero]

Depends on your fans, where you put them, etc.

Remember, the calculations said that with natural convection, you were doing pretty good, so you are only trying to overcome the length and orientation factors which are acting against you.

As long as you get enough air flow to simulate natural convection (which isn't very much), you should be good. You want to get air flowing over all the fins. Your heatsink is a little over 4" wide. I would guess a single 4" (100mm) fan mounted near the middle of the heatsink would probably suffice. Or two 2" (50mm) fans side by side to cover the width of the heatsink.

If you mount multiple fans along the length of the heatsink, there will be 'null' points between fans where the opposing forces cause zero air flow. Near the null points there will be low air flow. It's possible these regions would be small enough to be negligible, but I wouldn't want to hazard a guess about that.

In an ideal world, you would have the tools and expertise to model all this, and run a simulation to find out if you are okay. I don't run with that crowd. I would probably buy a fan or two, set it up and run it, especially given how far you already are along that path. Given the mass of that heatsink, nothing will happen super fast, and you can shut it down if it's getting out of hand. The initial temperature tests would be with my hands as measurement tools. If I was happy with the results, that would be the end of it. If not, I might put a thermocouple or thermistor on the hottest sections, and see if that satisfied me. Only if that failed (which I wouldn't expect) would I start thinking about ducting, better fans, multiple fans, etc.

Thank you!! Really appreciate you helping me out.

 

[Dave_H]

Getting accurate estimate of the COB temperature rise is a challenge with
so many factors. I have not directly used thermal modelling software but
have worked with specialists running it, and used its results. It's expensive
and complex and requires detailed setup, otherwise GIGO. Looks like it's
out of scope for this project anyway...

Total consumption is 8*34*1.0A + 3*35*0.7A = 346W but that is not the power
dissipation; fortunately it is a lot lower. 346W assumes all the input power
goes to heat, and none to light. Actual power depends on the LED efficacy.
There is some theoretical maximum (100%) for white LEDs which is >300 lumens/W.
I don't know about these COBs but if (say) they were 150 lumens/W, roughly
half goes to light and the rest to heat. So dissipation is probably closer
to 200W. This is good.

You should also know the COB to heatsink thermal resistance which is
part of the overall equation, and thermal resistance of the LEDs to the
COB package.

It's good that COBs are spread out along the heatsink, and none too
close to the ends, for heatsink effectiveness.

Regarding heatsink, effective thermal resistance varies with airflow.
Some vendors supply a graph of this. With specific fans you'd need to
relate fan capacity (usually CFM) to heatsink flow (usually linear feet per sec.
or min.) to get the thermal resistance; it is not linear with airflow, with
smaller heatsinks I have used.

Fan(s) should blow down the length of the fins as opposed to across, for
best effectiveness.

If single fan blows down the length, COBs further down will run hotter due
to less airflow (unless using a closed tube) but also due to air being
heated along the way. From PCB work I know 10C-15C increase is not hard to
get with much less power, but for this case I would not have a good idea.

It may come down to setting up something "reasonable" and doing some
temperature measurements. Power dissipation is fairly high but the heatsink
is quite beefy and looks like it's on the right track.


Dave