CXA3590 Build Plan

[Doer]

I want to start with something @SupraSPL was saying:
https://www.rollitup.org/t/diy-led-cree-cxa3070.789575/page-12 (post 230)
For comparison if we ran (1) CXA3590 at full blast (1.8A) on that same heatsink it would dissipate 150W at 28.4% efficient = 43 watts of photons.

If we ran (3) CXA3590 at 700mA on the same heatsink dissipation would be 153W at 42% efficient = 64 watts of photons
------------------------
After almost 200 pages of reading these last few days, I want to thank @tightpockt for driving me INSANE!!

I thought we were still at 10w emitters. But, noooooo...

So, this is a good time to being to plan. I move into my new house at the end of this month. And I was a bit at a loss. I was looking at @Gaius stuff, and how he has one LED per heatsink and fan. But, now I am thinking the way to go is to create these 150w modules as my basic building block.

So, first thing I am confused.@??

700mA on a 3590 is driving soft, so how can 3 dissipate 50w? He said driving one at 1.8A (heat sink equal) was 150w. I thought we are shooting for 25 watts per 3590. I have to be missing something.

IAC, here is my idea for a basic LED bloom module, no mounting rails, etc

3 - CXA3590
3 - 700mA AC-DC constant current drivers
1 - 150w active heat sink

~$250

Am I thinking about this correctly, yet?

 

[tightpockt]

Sounds like youre on the right track..i just got confirmation from digikey that my cxb3070s are on the way...at 1.4a they are like 44% efficient. The 3590's if i remember correctly have a fv of 77 or something like that. If you ran them at .7a youd be pulling the same wattage and efficiency with a higher upfront cost because of the more expensive driver.
Im just nitpicking though.
If i could do it over and i werent so cheap id go with a large passive heatsink and still use a small fan.

 

[tightpockt]

what I meant about the heat sink was I would do large enough that if the fan failed it wouldnt kill the led but not so large that it would be cost prohibitive

 

[nogod_]

Each 77v cxa3590 is going to consume 50w @ 700ma and actually have vf of ~72.

If its the 36v version, 25w @ 700ma and vf35

If youre shooting for everything on 1 heatsink. 24-30" x 5.886" should do the trick. Might as well throw them all on the same driver (hlg-185h-c700) which has 286v max. There are no beefy drivers out there for 3x 3590s @ 700ma that satisfy my OCD.

I want to start with something @SupraSPL was saying:
https://www.rollitup.org/t/diy-led-cree-cxa3070.789575/page-12 (post 230)
For comparison if we ran (1) CXA3590 at full blast (1.8A) on that same heatsink it would dissipate 150W at 28.4% efficient = 43 watts of photons.

If we ran (3) CXA3590 at 700mA on the same heatsink dissipation would be 153W at 42% efficient = 64 watts of photons
------------------------
After almost 200 pages of reading these last few days, I want to thank @tightpockt for driving me INSANE!!

I thought we were still at 10w emitters. But, noooooo...

So, this is a good time to being to plan. I move into my new house at the end of this month. And I was a bit at a loss. I was looking at @Gaius stuff, and how he has one LED per heatsink and fan. But, now I am thinking the way to go is to create these 150w modules as my basic building block.

So, first thing I am confused.@??

700mA on a 3590 is driving soft, so how can 3 dissipate 50w? He said driving one at 1.8A (heat sink equal) was 150w. I thought we are shooting for 25 watts per 3590. I have to be missing something.

IAC, here is my idea for a basic LED bloom module, no mounting rails, etc

3 - CXA3590
3 - 700mA AC-DC constant current drivers
1 - 150w active heat sink

~$250

Am I thinking about this correctly, yet?

 

[tightpockt]

Also, im of the opinion that, with the same wattage more smaller led's (2590's) give better coverage than fewer more powefull (3070, 3590) led's but its way too expensive

 

[Doer]

Sounds like youre on the right track..i just got confirmation from digikey that my cxb3070s are on the way...at 1.4a they are like 44% efficient. The 3590's if i remember correctly have a fv of 77 or something like that. If you ran them at .7a youd be pulling the same wattage and efficiency with a higher upfront cost because of the more expensive driver.
Im just nitpicking though.
If i could do it over and i werent so cheap id go with a large passive heatsink and still use a small fan.

Well, I think the consensus is to use the 36 vF variant, they have now. Cheaper up front. And large heat sinks are getting expensive and fans are cheap.

I don't want to be dissipating into the grow space, so I want to build modules, and put 3 modules, with their heat sinks in an air tube of some TBD config. That is 9 emitters running soft, and use an additional fan at the end of the tube to exhaust the air out of the space.

I guess I am trying to design a high power LED Cool Tube. I figure that one of these L-Cool Tubes, worth is 1000w HPS.

 

[Doer]

Each 77v cxa3590 is going to consume 50w @ 700ma and actually have vf of ~72.

If its the 36v version, 25w @ 700ma and vf35

If youre shooting for everything on 1 heatsink. 24-30" x 5.886" should do the trick. Might as well throw them all on the same driver (hlg-185h-c700) which has 286v max. There are no beefy drivers out there for 3x 3590s @ 700ma that satisfy my OCD.

Ah, thanks. The discussions range wide and far. Hard to keep up.

Since I am not worried about long term illumination decay, I might go with the 72 vF. I think I will be replacing these COB with better ones in a few years.

 

[Doer]

"Well, surface area is what you're trying to compute. To compute it, multiply outer perimeter by length.
1 in = 2.54 cm
1 in^2 = 6.4516 cm^2"

Thank you @alesh
https://www.rollitup.org/t/heatsinks-for-diy-led-lamps.851907/page-7 post #23

I was stumped, trying to figuring our what they mean by perimeter, on heatsinkusa.com

 

[Doer]

Well, I may as well know what I am doing or I am in for a big shock.

0.4°C/W - thermal resistance of gap filler (Rb)
90C - max case temp (Tc)
N/A - max junction temp (Tj) (Just need Tc for Cree COBs)

So, beginning with practical assumptions, and available materials, etc I used an on-line calculator to begin to swag in some design ideas. First the heat sink, based on RIU pointers and such, I get this.

Extruded Aluminum Heatsink Alloy: 6063-T6 Width: 5.886" Fin Height: 1"Base: .270"C/W/3": approximately 1.75 Weight per Inch: .25 48" = Price: $83.52

Outer perimeter = 39.594
C/W/3 = 1.75

That length is arbitrary for now and just represents the spacing of 3 plants in a row. The idea is to put 6 x 3590s @100w each.

Next swag is use the 177 cfm axial fan I happen to have, to exhaust the tube.

So, I have to swag a tube size. To make is easy for rough in, I figure a 1 foot square tube 48" long.

Now I have an enclosure I can calculate volume air exchange at 177 cfm.

 

[Doer]

The Ducting calculator I used gives:

Complete Air Replacement = every 1.44 seconds
Cubic Feet per hour = 10,000
Fan required = 166 cfm

 

[Doer]

Outer Perimeter 39.594 x 48 = 1900.512 sq inches of radiator surface.

Convert to metric
1900.512 x 6.4516 = 12261.3432 sq cm.

12261.3432 / 110 = 111.47w passive (??)

Well, that is comforting, I think. So the next effort is to figure out the factor for air exchange. If I can get 6 x this rail good for 600w. (or is the side of the wet paper bag I'm trying to punch out?)

 

[Doer]

I personally think you've overthinking this with the cooltube. You already have a long bar with airflow channels between each fin. If you covered the top with sheet metal, you'd already have a duct made out of fin channels.

One problem you might bump into regardless is that one side of your heat sink will be slightly cooler than the other since it's so long because the air going across the fins on one side is significantly cooler.

Do you live in a desert? Why even bother with cool tubes anyway? Maybe you have Stockholm syndrome? Putting a CPU fan in the middle of each long bar has been sufficient for most.

No I have Stockton syndrome. It gets hot out here.

So a 600w rail will add the heat to my grow room, I want to take it out. And I would like to explore tunnel efficiency and the Bernoulli principle of reducing the incoming air air temp thru venturi effect.

I like your idea. It is indeed a small tunnel, but without air volume overhead, you end end up conducting heat thru contact onto the sheet metal.

And I only began with 1 foot square tube and 4 foot sink. Also, I think it will look cool. Form over function.
The drivers can be in the airflow, as well.

What I am nattering about is pulling air down the length of the sink as you say, and that has to be more efficient than down firing fans into individual heat sinks.

A venturi to drop the temperature and a right sized tunnel may be able to take 1000 watts. I don't know yet.

 

[Doer]

For example, let's go small fan on the tube. We want the least watts for the fan, IAC:

Say, 48 cfm for a 92 mm case fan. And let's go with @churchhaze idea and reduce the tube height to 6 inches, over 5.8 inch fin height.

You are pulling 2400 cubic feet per hour. So, I like it.

 

[Doer]

But, as I said, it needs to get to linear feet per minute for heat sink.

Fortunately I found some conversion factors for common sized case fans after loss corrections.
http://www.crydom.com/en/Tech/HS_WP_FA.pdf

A 120 mm fan that is rated for around 100 cfm will pull 600 linear feet per actually.

This gives a correction for passive radiation to active, of .309.

So the subject heat sink above, has a rating of 1.75.

1.75 x 0.309 = .541 That is quite an efficiency gain.

This means I can multiply the effective size of heatsink by, 3.2363

12261.3432 sq cm. x 3.2363 = 39680.7223

39680.7223 / 110 = 360w @ 600 feet per minute

I think we are scaling linearly now. So, I think for this 1st Concept, 600w can be handled on this heat sink with 1200 feet per minute air flow.

 

[Doer]

In fact when we see this chart we can understand why I am interesting tunnel efficiency. The faster you move the air, the better your correction factor. Simple wind chill physics.

.For the doc I linked previously:
Once the fan’s LFM value is calculated, the improvement that it has on the heat sink’s thermal
impedance rating is calculated by multiplying a correction factor per Table 3 below times the
heat sink’s free air convection thermal resistance rating.
LFM
600 ----- 0.309
1000 ---- 0.239

Just look at the change between 600 feet and 1000 ft per min.

 

[nogod_]

Stockton syndrome....

No I have Stockton syndrome. It gets hot out here.

So a 600w rail will add the heat to my grow room, I want to take it out. And I would like to explore tunnel efficiency and the Bernoulli principle of reducing the incoming air air temp thru venturi effect.

I like your idea. It is indeed a small tunnel, but without air volume overhead, you end end up conducting heat thru contact onto the sheet metal.

And I only began with 1 foot square tube and 4 foot sink. Also, I think it will look cool. Form over function.
The drivers can be in the airflow, as well.

What I am nattering about is pulling air down the length of the sink as you say, and that has to be more efficient than down firing fans into individual heat sinks.

A venturi to drop the temperature and a right sized tunnel may be able to take 1000 watts. I don't know yet.

 

[Doer]

But, the correction doesn't scale lineally, I see.

When you get to 1000 feet per minute, you find 0.239 as the correction.

1.75 x . 239 = .481

The multiplier now is 4.184 instead of 3.2363

 

[Doer]

I admit that a lot of that material is way over my head (heat sink engineering is not trivial by any means), but the majority of designs here recently have been trying to reduce air speed/pressure considerably. The efficiency of a heat sink is sort of irrelevant, and in many ways misleading in this case.

Consider a red hot heat sink, for example. The radiant heat emitted at that temperature combined with the high temperature difference between the metal and the air would mean very high cooling efficiency, but your cobs would be running at red hot temperatures!

Now consider a completely passive setup. The efficiency of the cooling system does not matter in this case, because the input power to the cooling system is the waste heat from the cobs. It seems like a red herring to consider the efficiency of the cooling setup, because the cooler you run the heat sinks, the less efficient they are, but the more effective they are at being heat sinks. (which ultimately makes the cobs more efficient). There is a point of diminishing returns as you approach ambient temperature.

I'll admit I didn't follow that last part very well. But, I know I am not trying to cool the sink to ambient. I want it to run at the best temp. whatever that is.

And I do think the efficiency of the sink is proportional to how much watts you can hang on it. So, what is easy to throttle? Fan Speed. What is easy to sense it's absence? Fan Speed.

The idea I thought was to drive as much watts as you can in the smallest foot print. And Forced Air creates efficiency in heat sinks.

So, all I want to do is drive 6 of these at 100 watts each on as economical total system as I can. And that means transporting heat out of the grow. I did that with a water cooled 1000 hps in a hall bathroom.

Consider the duct is insulated, I can take all that heat out and it doesn't matter what the heat sink temperature is. I can adjust that with fan speed. All I have to do is keep if from damaging the COB.
And I can run the heat sink at the best temp for the COB.

 

[Doer]

It totally sounds like Stockholm syndrome to me if you're considering insulating your heat sinks! All that precious surface area lowering thermal resistance and case temperatures... lost! lol.

The lower you can keep the case temperatures of the cobs, the greater the efficiency of the cobs will be. It's not just a matter of sucking all the heat out. Requirement #1 for the heat sinks is lowering case temperature. Evacuating all that hot air with the same fan would be a neat bonus! Consider you may be jumping over a 100 dollar bill to pick up a penny here or there.

Again, I glazed over with a lot of the airflow stuff you were talking about earlier, so I could be totally missing something, but it also sounds like you are focusing on a secondary requirement while sacrificing the main requirement (get Tc as close to ambient temperature as possible before hitting diminishing returns)

Insulate the duct, not the heat sinks. I am isolating the sinks, in their own air flow, not insulating them.

So, which is it?

large, heavy passive heat sinks

or

Smaller forced air heat sinks?

That is the only trade. Close to ambient is a balance with the ambient of the room which is a variable in several ways. The more heat you dissipate the higher the ambient or more load on your cooling.

Forced air on a fail safe seems a win, to me. 600w in 48", on one cooling fan which also exhausts from the grow space is a win to me.

So, you get the penny before it rolls in the drain and pick up the $100 bill, you have your foot on.

 

[Doer]

It totally sounds like Stockholm syndrome to me if you're considering insulating your heat sinks! All that precious surface area lowering thermal resistance and case temperatures... lost! lol.

The lower you can keep the case temperatures of the cobs, the greater the efficiency of the cobs will be. It's not just a matter of sucking all the heat out. Requirement #1 for the heat sinks is lowering case temperature. Evacuating all that hot air with the same fan would be a neat bonus! Consider you may be jumping over a 100 dollar bill to pick up a penny here or there.

Again, I glazed over with a lot of the airflow stuff you were talking about earlier, so I could be totally missing something, but it also sounds like you are focusing on a secondary requirement while sacrificing the main requirement (get Tc as close to ambient temperature as possible before hitting diminishing returns)

Stockholm syndrome is where you empathize with those that hold you hostage.

So, it doesn't really compute for me...sorry.

Sucking all the heat out is lowering the case temperature.