DiY LED - Cree CXA3070

Awesome graphs thank you SDS! Better than the Cree datasheets, makes it so much easier to visualize temp vF and efficiency! Here is how I think of it in my head:

BT those are looking awesome!!! Heatsink USA is usually very fast, maybe they had a hang up.

SDS, one thing I have been confused about is the relationship of Tc and Tj. I assume the relationship is not linear? In other words, does the thermal resistance of the COB package change depending on the drive current or is it a constant? In the Cree data sheet they seem to imply that a Tc of 55C = Tj 85C when running at 1925mA.

SupraSPL said:

SDS, one thing I have been confused about is the relationship of Tc and Tj. I assume the relationship is not linear? In other words, does the thermal resistance of the COB package change depending on the drive current or is it a constant? In the Cree data sheet they seem to imply that a Tc of 55C = Tj 85C when running at 1925mA.

Click to expand...

I remember reading the datasheets and they state that the Tj is 30*C higher than Tc. I think that is just how it is.

SupraSPL said:

Awesome graphs thank you SDS! Better than the Cree datasheets, makes it so much easier to visualize temp vF and efficiency! Here is how I think of it in my head:
View attachment 3166914

Click to expand...

Yes...Good and pretty accurate ..'mind visualisation " ...
!!! ..
But ...
Check this out,also ...
Some other thoughts on the issue ...

SupraSPL said:

SDS, one thing I have been confused about is the relationship of Tc and Tj. I assume the relationship is not linear? In other words, does the thermal resistance of the COB package change depending on the drive current or is it a constant? In the Cree data sheet they seem to imply that a Tc of 55C = Tj 85C when running at 1925mA.

Click to expand...

CannaBare said:

I remember reading the datasheets and they state that the Tj is 30*C higher than Tc. I think that is just how it is.

Click to expand...

No it is not linear ....

T junction = T case + Rθ * P heat

( Rθ = thermal resistance between semiconductor junction and bottom of array case ...
Expressed in C/Watt .Multiplied with dissipated heat power = ΔT between junction and case .
Tj - Tc = Rθ * Power heat ...)

If Tc is of known value ..(say 45C ,for example... )

Rθ of CXA3070 is of constant value .
0.8 C/W (awesome figure ..because thickness of case is 1mm and surface is 27.35 ^2 mm . )

P heat = elec. Power - Radiant Power

elec .Power = If *Vf

Radiant power = elpower * efficiency

effic. = L/W / LER


so ...

T junction = T case + { 0,8 * ( ( If*Vf ) -( If*Vf *L/W /LER ) ) }
or
T junction = T case + { 0,8 * ( ( 1 -(L/W /LER ) ) *If*Vf ) }

For Tc =45 and If= 1.4 A ...
Tj = 45 + { 0.8 * ( ( 1-0.375 ) * ( 1.4 * 37.18 ) ) } =>
Tj = 45 + 32,53 = 77,53 C .....


The worst operational case for the CXA ..
TC=85 C ,If =2.8 A....


Vf= 41.4
efficiency = 81.6 / 325 = 0.251
Tj = 85 + { 0.8 * ( ( 1-0.251 ) * ( 2.8 * 40.94 ) ) } =>
85 +68.68 = 153.68 C

Really tough led array ...
But operating with Tj=~ 154C ,service life* is decreasing rapidly ..
A rough guess would be at ~ 8-10K hours ...
Or maybe even less ....

*(>70% of initial output )

Hey SDS what does LER stand for?

SNEAKYp said:

Hey SDS what does LER stand for?

Click to expand...

LER stands for Luminous Efficacy of Radiation.

The theoritical maximum of lumens per watt is the 683 lumens per radiant Watt .
But that figure belongs to strictly monochromatic light of exactly 555 nm .

Any other light source has lower LER than that figure ....
Every light source of white light has a certain LER value .
( Units of LER :Luminous flux [lm] / Optical power [W] )

The CXA3070 3000K 80 CRI has a LER of 325 lm/w .
It means that for every radiant watt of light , the cxa will have a luminosity of 325 lumens ..

Using CREE PCT (Product Characterization Tool ) ,
http://pct.cree.com/dt/index.html
the Luminous Efficacy value of a led light source ,
can be obtained for every given Tsp (TC ) in relation with If .
Expressed in Luminous flux [lm] / Electrical power [W]

So ,dividing he Luminous Efficacy value of a led light source wiuth it's LER ,the
radiant efficiency value can be obtained .

( Luminous flux [lm] / Electrical power [W] ) / ( Luminous flux [lm] / Optical power [W] ) =>
Optical power [W] / Electrical power [W] =radiant efficiency ....


http://apps1.eere.energy.gov/buildings/publications/pdfs/ssl/ohno_color_raleigh2010.pdf

One important notice about Cree PCT and CXA series ...

Tj is not applicable in PCT spreadsheet regarding CXA series .
Either you have Tj or Tsp ,is actually the Tc value ,regarding CXA series ...

For example :



From the Cree's datasheet ,at 1300mA CXA3070 ,no matter the bin ,will have a VF of approx 37 Volts ..
At T case =25C .....


Going to Cree PCT ...


So ..No matter the bin or if Tj or Tsp ,are selected ,at 25 C (which is T case for CXA series ...),Vf is indeed
37.09 Volts ..

So no matter if Tj or Tsp selected in Cree's PCT ,for the CXA is actually the Tc figure ....

Tj of CXA3070 , has to be calculated by the
" Tj=Tc + ( 0.8 * Heat power dissipated ) " equation ...

And here's an example ,that underlines the crucial importance of good cooling of CXA3070 ....

3 different (hypothetical ) led fixtures using all of them ,the same type of CXA3070 AB bin ...

Fixture A )
CXA is driven at 2 A .
Cooling is exceptional ,as water/liquid cooling (forced conduction) is used and
Tc is maintained at 33C .
Efficiency is 117.6 /325 = 0.36 .
Thus irradiance is 0.36 * 78.368 =28.35 Watts.

Fixture B )
Cxa is driven at 1.9A .
Cooling is good ,as a fan/heatsink (forced convection) system is used and
Tc is maintained at 45C .
Efficiency is 117.1 /325 = 0.36 .
Thus irradiance is 0.36 * 73.611 =26.52 Watts.
Fixture C )
Cxa is driven at 1.45 A .
Cooling is below average ,as a naturally convecting heatsink (passive ) is used and
Tc is maintained at 85C .
Efficiency is 117.8 /325 = 0.36 .
Thus irradiance is 0.36 * 53.47 =19.38 Watts.



but there are also some more pros &* cons for each case ...

-Fixtures A & B use additional energy /power for their active cooling systems ...
-Service lifetime of CXA arrayy of Fixture C is going to be way less/shorter than
those of Fixtures A & B ...
..To name a couple of them ,for example ...

Supra I would like a big favor from you .
I used a LER of 325 lm/Wrad for the CXA3070 3000K 80CRI...

(As the 328 one ,did not seem accurate to me ...
How it is possible for 5000K having a LER of 324(which is also not accurate-has to be more )
and the 4000K with a LER value of 323 (also not accurate-has to be more than that.. ) ,
while the 3000K has a LER of 328 ?
It can't be ..
Both 5000K & 4000K have more power 'concentrated ' closely to 555nm ..
So ,normally ,they would have a bigger LER value of 3000K ..
Way bigger than 328 ...Or 325 ...)

So ,if the accurate LER values ,for 3000K is 328 lm/ radiant Watt and for
2700K ( 80CRI ) has aLER of 321 lm/radiant Watt

-which theoritically stands as true ,as the 2700K has less power
towards 555nm than the 3000K and also more power at >700nm ,
which those two facts alone ,decrease the LER value of the 2700K -

I used the 325 value for the spreadsheet calculations...
Anyway ...
The favor ....

Can you gather and post all the known (as accurate ) LER values about any array
of the CXA 3070 series ?

If 3000K 80 CRI has a LER =325
and 2700K 80CRI a LER=321 ...

Then the 90 CRI ones (3000 & 2700 ) will have lower LER values,respectively ,than their 80CRI counterparts...

As also the 4000 & 5000 K (80CRI ) will have higher LER values than 325 ...

(My rough guess would be around 328-330 for the 4000K 80CRI and about
326-328 for the 5000K 80CRI -less power of green wls than the 4000K ,but more blue ..

A tiny -big- favor .....
Hope you're able to do it ...

Thanx .
Cheers.

Wow great info thnx SDS! I think I understand how to estimate Tj much better now. My goal is to try and stay near Tj 50C and I am running the ABs at 900mA so I just need to measure Tc and add 13.5. I have a thermocouple and a very accurate infrared thermometer so I will try both ways.

Our first source of LER data was KNNAs estimation back in the Golden Dragon/ Cree XRE days, 280 for cool white, 310 warm white.

Then we got Mr Flux calculus in the CXA analysis thread, 328 for 3000K, 323 for 4000K and 324 for 5000K. Those numbers are based on the curves in the CREE datasheet which are not very specific about their color temp, guidelines only. Mr Flux explained that the LER for each curve was very similar as long as you don't go high CRI, very warm or very cool. The Vero was more specific about its color temps and he came up with 320 for 3000K, 325 for 4000K and 337 for 5000K. The Vero 3000K curve is similar to the CXA 2700K curve with 8% blue, so it makes sense that they have similar LER.

And finally I got word directly from Cree on the CXA 2700K (321) and CXA 3000K (325), which closely agreed with Mr Flux data

SupraSPL said:

Wow great info thnx SDS! I think I understand how to estimate Tj much better now. My goal is to try and stay near Tj 50C and I am running the ABs at 900mA so I just need to measure Tc and add 13.5. I have a thermocouple and a very accurate infrared thermometer so I will try both ways.

Our first source of LER data was KNNAs estimation back in the Golden Dragon/ Cree XRE days, 280 for cool white, 310 warm white.

Then we got Mr Flux calculus in the CXA analysis thread, 328 for 3000K, 323 for 4000K and 324 for 5000K. Those numbers are based on the curves in the CREE datasheet which are not very specific about their color temp, guidelines only. Mr Flux explained that the LER for each curve was very similar as long as you don't go high CRI, very warm or very cool. The Vero was more specific about its color temps and he came up with 320 for 3000K, 325 for 4000K and 337 for 5000K. The Vero 3000K curve is similar to the CXA 2700K curve with 8% blue, so it makes sense that they have similar LER.

And finally I got word directly from Cree on the CXA 2700K (321) and CXA 3000K (325), which closely agreed with Mr Flux data

Click to expand...

Tj of 50C ?

WoW !
That is a really low figure ....
...
Start from a Tc=25 C


Tj = 25 + ( 0.8 * ( (1- ( 151.5/325 ))*32.147) ) =>
Tj= 25 + (0.8 * (( 1- 0.466)*32.147 )) =>
Tj=25+ ( 0.8 * ( 0.533* 32.147 ) ) =>
Tj = 25 + ( 0.8 * 17.161 )=>
Tj= 25+ 13,729 =>
Tj = 38,729 ....


For Tc =30C ...


Tj = 30 + ( 0.8 * ((1- (150/325) )*32.094 )) =>
Tj = 30 + 13.8 = 43.8 ....
.......................................................................................
I think Tc has to be somewhere around 36 C ...
At 900mA ...
In order to have a Tj of 50C ..

Tj=36 + ( 0.8 * (( 1- (148.1/325 ) )*32.031 ))
Tj= 49.94 C

Well ...
At low ambient temperatures (20-25C ) and with good cooling ,
that is doable enough ..

Tj =49.4 C = Tc +(0.8 *17.43 )

Tc= 36 C = Ta + ( RTIM+HEATSINK *17.43 )
For Ta =25 C =>
RTIM+HEATSINK = (Tc-Ta ) / 17.43 =>

RTIM+HEATSINK = 0.631 C /W

Rtim= Thickness / (T.conductivity *Surface )

If Rtim = 0.00005 / ( 10.2 * 0.000564 ) = 0.00869 = 0.0087C/W

( Prolimatech PK2 : 10.2W/mK
Thickness = 0.05 mm = 0.00005 m
Surface = 23.75 mm x 23.75 mm = 0.000564 m^2 )

So Rheatsink has to be lower than 0.631 - 0.0087 = 0.622 C/W

Guod suggested the Fischer Elektronik SK 47 150
With Rth=0.53 C/W and 150 x 200 x 40 mm size ( L x W x H ),20 fins .
http://www.conrad.com/ce/en/product/188824/Fischer-Elektronik-SK-47-150-SA-Heat-Sink



http://www.fischerelektronik.de/web_fischer/en_GB/heatsinks/A01/Standard extruded heatsinks/PR/SK47_/$productCard/dimensionParameters/index.xhtml;jsessionid=B7583AE09E5459349B15E1EB7A36646C


You can achieve your goal with one CXA per heatsink ,without active cooling ...
Passive cooled at 25C ambient temp.

You can still achieve your goal with two CXAs per heatsink and at higher Ta ,
but utilising forced convection ..(Fans ) ...

( That is a very good heatsink ,I must admit !!! )

Cheers.

Thanks for the math! During the hottest part of summer the ambient temp in the tent will get as high as 26C but most of the year it is 22C.

I am using this heatsink profile in a 6" length. It claims C/W/3": approximately .80. Normally I would only install 50W on this heats for passive cooling or up to 120W with active cooling. But since the AB is so efficient at 900mA I plan on running 64W on the passive heatsinks and see what Tj I end up with.

SupraSPL said:

Thanks for the math! During the hottest part of summer the ambient temp in the tent will get as high as 26C but most of the year it is 22C.

I am using this heatsink profile in a 6" length. It claims C/W/3": approximately .80. Normally I would only install 50W on this heats for passive cooling or up to 120W with active cooling. But since the AB is so efficient at 900mA I plan on running 64W on the passive heatsinks and see what Tj I end up with.
View attachment 3167344

Click to expand...

This is a heatsink made specifically for forced convection.
You can see the differences from the SK 47 ...

Fins of that heatsink are way 'taller ' and thinner than the fins of the SK 47 150 ..
And more closely spaced ..They have almost no 'triangular' profile ,but rather parallel fin sides ...

Those fins are made to co-operate with a fan or fans .
Fan(s) should be blowing cool air ,firstly at fins ..
Naturally convecting ,this heatsink will suffer .
But if actively cooled ,can actually be better than an active cooled SK 47* ..

*(Which is designed as ' naturally convecting heatsink ' ...
With short ,thick ,triangular profiled fins ,placed quite spaced apart ..)

Because of the large surface area 110cm²/W, thick base plate, large space between fins (10mm) and slight assistance from the circulation fan, they do surprisingly well passively cooled. The hottest part of the heatsink, in between the fins, never gets above 30C and most modules are only 25C. But yes they perform even better actively cooled, you can install quite alot of power, with heatsink temps just a few degrees above ambient.

I designed the RWB modules with these 10.08" heatsinks before COBs were available. So now the problem is that COBs are putting out so much light, when I actively cool them and add lots of power, the light is too concentrated because of the tall fins giving the heatsink a small footprint. So for the most part these units are best used passively cooled and more of them, to better spread that light over the canopy. Another problem is that the 140mm fan doesnt reach all the fins and is not being used very efficiently because of the short contact time of the forced air with the fins.

Now with COBs in mind, I have been recommending this profile (6" Heatsink USA) for active cooling, similar to the SK 47. 140mm fan fits and blows air a long way on all of the fins so the fan power is getting used very efficiently. The large footprint allows the power to be spread out across the canopy. I am installing 208W on each of these 24" heatsinks and fan is dissipating 1.6W, up to 2.2W if necessary. Will report back with ambient, heatsink and Tc temps.

You can tie the plants down to get more room also. And you will get the added bomus of getting more light to those lower bud sites.

I am surprised that there are not more people using liquid cooling systems. They are quieter and you can remove much more heat.

SDS always coming through with the data. Thanks for the explanation on LER!

Mohican said:

I am surprised that there are not more people using liquid cooling systems. They are quieter and you can remove much more heat.

Click to expand...

Wouldnt that be more for a much mich larger grow? I once seen a liquid cooled sealed room, one thing i noticed is the dehumidifier kept having problems from all that humidity. But like u said mo, it does the job! I jist think its too much energy and water, wouldnt b ideal for a smaller set up imo?