Heatsinks for DIY LED lamps

[SupraSPL]

When they test the heatsink's thermal resistance for passive cooling, they rate it when vertically oriented and at a temperature 80C hotter than ambient. That large temperature difference creates a pressure differential that allows the hot air to convect upward and cool air to enter at the bottom and side of the heatsink.

In our situation we have the heatsink oriented horizontally and we want the heatsink only a few degrees above ambient, so there is not enough pressure differential get the air convecting until it reaches a high temperature.

Also, in our application the circulation fan can assist the heatsink, making it semi passive and making the circulation fan multipurpose.

When using active cooling, even the slightest breeze can knock the heatsink temps down close to ambient, so we dont need a lot of fan power but adding a fan and fan PSU adds some cost, complexity, wiring and can reduce electrical efficiency slightly. The cost of the fan/psu and wiring can partially make up for the extra cost of using passive.

 

[klx]

Thanks for dumbing it down for me.

I was keen on passive cooling for the simplicity and also noise reduction but I understand what you are saying. I guess the only way to find out for sure is to grab one and test it out.

Thanks again.

 

[ttystikk]

haha nice could you hear the penny drop in my brain just then lol...cheers bro. just so I got this fully locked down, whats the formula to work out how many "heat watts" a COB will produce?

I don't pretend to know all the math, so I use a shortcut; the efficiency rating of the chip in the specific combination you're considering.

For instance, if you drive a CXB3590 at its full rated output of 100W, it's about 44% efficient. Different bins and color temperatures will affect this number, do it's best to drill down to a specific chip. Anyway, the other 56% of the wattage the chip uses is converted to heat; 56W worth.

Continuing the same example, driving that same CXB3590 at 50W nets you 56% efficiency, and now only 22 heat watts to be concerned with. Another way of thinking about this is that now you'd be running two chips for the same 100W, but you only have 44 heat watts to remove.

Finally, running the CXB3590 at only 23W nets the big spender 63% efficiency;
23 x (1-.63)= 23 x .37 = 8.5 heat watts each... times four chips to get back to 100W, hence the need for Mr money bags!
4 x 8.5 = 34 heat watts total. This is why it's so easy to run passive heat sinks together with scads of very softly driven chips.

FWIW, I chose a very common combination of 4 CXB3590 chips on one 200W driver, for 56% efficiency. I chose this combo as a cost effective (4 chips and driver are approaching $200) middle ground that still boasts over 50% efficiency. Considering how common it's become amongst the RIU DIY crowd, one could call it the smallblock Chevy of the COB LED growing world, lol

 

[ttystikk]

Cheers mate, you really need to be running a COB DiY school lol...if ya ever in New Zealand look me up man I got ya covered!!

You're in it. We're ALL classmates, bro!

 

[doz]

For my first build, I went with heatsinks from HUSA. Next build, I am thinking about trying something different. Found these on ebay and looks like they will work fine. Frostytech did a tech on CPU which rated them @ 46c on a 125w heat load. Should be plenty considering I will only have a 22w heat load (roughly).

GlacialTech Igloo 7223 (look it up on ebay).

They will cost $44 for 12 of them shipped. Utilize 1" angle (I can get it from work for free, albeit not aluminum, it will work fine) and build a setup. I think for the price, these cannot be beat.

 

[klx]

For my first build, I went with heatsinks from HUSA. Next build, I am thinking about trying something different. Found these on ebay and looks like they will work fine. Frostytech did a tech on CPU which rated them @ 46c on a 125w heat load. Should be plenty considering I will only have a 22w heat load (roughly).

GlacialTech Igloo 7223 (look it up on ebay).

They will cost $44 for 12 of them shipped. Utilize 1" angle (I can get it from work for free, albeit not aluminum, it will work fine) and build a setup. I think for the price, these cannot be beat.

Hard to argue with that cost.

 

[SomeGuy]

For my first build, I went with heatsinks from HUSA. Next build, I am thinking about trying something different. Found these on ebay and looks like they will work fine. Frostytech did a tech on CPU which rated them @ 46c on a 125w heat load. Should be plenty considering I will only have a 22w heat load (roughly).

GlacialTech Igloo 7223 (look it up on ebay).

They will cost $44 for 12 of them shipped. Utilize 1" angle (I can get it from work for free, albeit not aluminum, it will work fine) and build a setup. I think for the price, these cannot be beat.

Thanks for.posting. picked up some myself

 

[Led Newbie Oz]

http://www.fastron.com.au/powere_1e.php

Aussie heat sink suppier.

Unlike Conrad this supplier can cut sections of up to 3m

 

[threetwobravo]

not trying to be a dick bro but try running the numbers your self using the formula's we used a couple of pages back, its pretty easy. Come back with what you get and I'm sure the crew will tell ya if its not right.

 

[klx]

http://www.fastron.com.au/powere_1e.php

Aussie heat sink suppier.

I'm putting together 2 panels for a 3'x4' Bloom tent

Am thinking of running (x4) cxb3590 @ 50w each along a 750mm (actively cooled) H4 heatsink, per panel
Do you guy think this heat sink will be capable of handling the heat.
If not the H42 is more like a 5.9' profile but costs 3 times more.

Price for H4/750R $34.80 each, 1-2 days


Price for H4/750M $47.10 each, 1-2 days


Price for H42/750R $116.05 each, 1-2 days


Price for H42/750M $134.50 each, 1-2 days

unlike Conrad this supplier can cut sections of up to 3m/9.8'

Hey Bro,

What does the 750R and 750M stand for? Are those prices for 750mm lengths?

If you are actively cooling then it should be ok but I cant recall the active cooling calculation. For passive you want 120cm^2 / heat watt which works out costing a bomb for a bloom light.

A third option I have been reconsidering for my first coulple of lights is the CPU cooler option. I found these on ebay which are virtually identical specs to the GlacialTech Igloo 7223 that doz posted about, with free shipping.

http://www.ebay.com.au/itm/New-4200-RPM-Heatsink-Fan-Cocket-754-939-940-for-AMD-CPU-Athlon-64-HK-Shipping-/231676284129?hash=item35f0fb8ce1:g:HxEAAOSwMmBV6WgR

 

[dionysus4]

not trying to be a dick bro but try running the numbers your self using the formula's we used a couple of pages back, its pretty easy. Come back with what you get and I'm sure the crew will tell ya if its not right.

yeah fully

did everyone miss the xl file I uploaded to calculate this or is it not working?

 

[Led Newbie Oz]

OK point taken, will do the math now.
Klx, yes the quote was for 750mm section.
R stand for raw and M stands for Milled.

 

[Led Newbie Oz]

Math done,
(profile width X 2) + (fin height X 2 X # of fins) + (base height * 2) = Perimeter

H4 (98x2) + (25 x 2 x 13) + (8 x 2) = 862mm
Perimeter = 86.2cm
86.2 x 75 =

Assuming 120cm²/heat W, that heat sink/length would passively cool 53 Heat W
Assuming 40cm²/heat W (Active), that heat sink/length would actively cool 160 Heat W.

With (x4) top Bin cxb3590 Cd run (using HLG-185H C1400B) Softly , i should get 56% efficiency.
So 200w = 88 Heat W
A 750mm section is too small to passively cool 88 heat W, but will easily actively cool 88 heat W.

Should i keep the 750mm length for good spread?

 

[sanjuan]

HeatsinkUSA has 10% off for orders over $100 until the end of this month.
Same code as before: GIVEME10

 

[doz]

Go with the 10" profile. You can save a few bucks and scale it down to 24" if you wanted and still be OK on passive. 32"x10" will be fine though. Or grab 2 16x10 if you want a bit more customization of the light placement.

 

[nevergoodenuf]

If anyone could recommend a cooler that can handle a 320 watt COB passively, post up. The only one I know will work is the stadium light heat sink. I doubt I will run it that hard but want it to run cool.

 

[ttystikk]

If anyone could recommend a cooler that can handle a 320 watt COB passively, post up. The only one I know will work is the stadium light heat sink. I doubt I will run it that hard but want it to run cool.

Those are some fat chips! I am just peeking in, I don't know what they are, cost or came from? How hard do you plan to run them?

 

[nevergoodenuf]

These are from Flip Chip Opto, the Luna 300 in 3000K and 4000K 80cri. They are $70 each and can handle over 600 watts Max. They are also a short drive from me.

 

[tsharp513]

For my first build, I went with heatsinks from HUSA. Next build, I am thinking about trying something different. Found these on ebay and looks like they will work fine. Frostytech did a tech on CPU which rated them @ 46c on a 125w heat load. Should be plenty considering I will only have a 22w heat load (roughly).

GlacialTech Igloo 7223 (look it up on ebay).

They will cost $44 for 12 of them shipped. Utilize 1" angle (I can get it from work for free, albeit not aluminum, it will work fine) and build a setup. I think for the price, these cannot be beat.

Just ordered mine, switch shipping to USPS expedited to save yourself some money. $22 for 5 sinks!
Get em' while they're hot!

 

[organic-fanatic]

Grabbed a bunch a thermometers and settled on this one

http://www.amazon.com/Digital-Candy-Deep-Fry-Thermometer/dp/B0075M5ESY/ref=sr_1_10?s=kitchen&srs=2530581011&ie=UTF8&qid=1446179683&sr=1-10&keywords=taylor thermometer

It gives readings that agree with the mercury, but much faster and easier (10-20 seconds) so I have a lot more confidence in the measurements/ I have abandoned the IR thermometer which has proven to be precise and repeatable, but very inaccurate for measuring aluminum.

These heatsink temps were taken directly behind the COB which was the hottest place on the heatsink that I could find, including the front next to the COB which was a few degrees cooler. Ambient in the tent was ~75F(24C) and there was decent airflow from the circulation fan hitting all heatsinks with a slight breeze.

Passive cooled in flower tent, all driver turned up to max
====================================
CXA3070 3000K AB bin @ 25.5W ea:
8.46"X12", 96F (35.5C) 188cm²/heat W

CXB3070 3000K AD bin @ 24.5W ea:
8.46"X12", 96F (35.5C) 222cm²/heat W
4.9"X14", 96F (35.5C) 153cm²/ heat W

CXB3590 3500K CD bin @ 24W ea:
8.46"X12", 93F (34C) 277cm²/heat W (good demonstration of the law of diminishing returns)
4.9"X14", 93F (34C) 190cm²/heat W (this heatsink was tilted unfavorably for cooling, did not matter)
4.9"X14", 93F (34C) 190cm²/heat W
10.08"X6" 90F (32C) 318cm²/heat W (overkill surface area, but a good data point)

CXB3590 3000K CB bin @ 26.5W ea
10.08"X6" 84F (29C) 247cm²/heat W (these were on the edges, must be getting the best airflow)
=============================================

Passive cooled in veg tent, driver set at maximum
single CXA3590 5K DB 24.5W:
5.88"X6" with slight airflow hitting heatsink 90F (32C), 179cm²/heat W
5.88"X6" with almost no airflow 108F (42C), 179cm²/heat W

According to the PCT, the difference in output between these 2 temps is about 2%.

=================================================

Active cooled veg fixture:
pair of CXA3590s 5K DB @ 85W ea:
3.5"X36" with 2.5W of 80mm fan, 108F (42C), 52.6 cm²/ heat W (measured 4% temp droop)
===================================================

Still digesting this data