Thermal Paste as TIM & Cree CXA30x0

stardustsailor

Well-Known Member
A great thing about COB technology ,is the ability of the array(s) to be directly installed on a heatsink / thermal pad( in case of liquid cooling ).

The elimination of 'Metal Core PrintedCircuitBoard " by itself is the total absence of three layers ..

1) Copper (solder pads/traces ) layer (50-100 micron / 0.05-0.1 mm micron )
2) Thermal conductive / electrical isolating layer (5-50 micron / 0.005-0.05 mm )
3) Aluminium substrate layer (0.5 -2 mm )

Add also
4) Solder layer ( 50-200 micron )
and
5) Thermal interface material between MCPCB and heatsink (50- 200 micron )

6) Consider also , the extremely low thermal resistance ( 0.8 C / W ) between
semiconductor's junction and bottom of the array case ..

Rθj-c CXA30x0 = ΔΤ { Tj - Tc (aka Tsp ) } / Heat Power = 0.8 C / W ....or K /W ....

5 layers of different "surface grit contact" level
and with different thermal
and mechanical (expansion ) characteristics ...

Are gone for good ...
( For the led DIYer ? That is a touch of Heaven! (: ... )

What stands between The CXA array and the heat sink...

Is a thin ( Thin to win ! ) layer of paste or pad ...

And while cooling becomes more efficient ...
And less demanding than of small format leds ...
Still it has to designed and done carefully ...


So,lets cut to the chase...
Let's refresh some Cree knowledge acquired ...
thds.JPG


More about Thermal resistance ...
thermal resistance.JPG


Tj from Tsp / Tc ...

tj vs tsp_tc 1.JPG

......

tj vs tsp_tc 2.JPG

And about TIM thermal resistance calculating ....
tim 2.JPG
 
Last edited:

stardustsailor

Well-Known Member
Thermal Paste is probably the best choice of TIM between a CXA array and an copper/ aluminium surface ..

From Cree Appl. sheet ...
Tim 23.JPG
tim most pref.JPG
......

So TIM = L / ( k * A )

where:
RθTIM is the thermal resistance of the TIM
L is the thickness of the TIM (m)
k is the thermal conductivity of the TIM (W/m K)
A is the contact area (m2)

...
Before moving into details about thermal paste ....

Let's have a look at the CXA30x0 array mechanical dimensions ...


From Cree Datasheet ...
spec ds  mechdim.JPG



And a 3D model ,on SketchUp ...

cxa TIM MechDim.jpg


cxa TIM surface.jpg


cxa TIM thickness.jpg


Notice the 0.0485 mm ( 48.5 microns ) dark layer ....
That is a desirable thickness of TIM (thermal paste ) -visually-compared
to the 1.15mm thickness of the CXA array ...

Aim For thin . ..
Cree suggests ( In a rather indirect manner, through test reports ) a thickness of 0.05 mm ,
for thermal paste(s).....

.........
ctr.JPG
 
Last edited:

Positivity

Well-Known Member
I especially like the point of metal being the best conductor of heat. We are just trying to fill the voids..not necessarily put a complete layer.

That said..the cool silver pad does seem to work pretty good since it is a extremely thin pad. A bit challenging to work with as it's very fragile..but workable with practice.

Also noticing that my cxa 3070 light runs hotter now that it is in an enclosed space.

So future designs of mine will include..ample heatsink space between cobs..not too close to edges of a heatsink as cooling will suffer...and possibly fan over each area with cobs.

My design for example could be improved with a bit wider heatsink and bit longer to fit the same amount of leds. Which would also have given me the extra room to put fans over each area.
 

CannaBare

Well-Known Member
Still love your sketch-up work! I cannot get anything like that. I am trying to design my own reflector for 100* viewing angle. Any tips for that parabolic shape?

Also, is sketch-up 3D printer compatible?

Thanks in advance!
 

stardustsailor

Well-Known Member
Still love your sketch-up work! I cannot get anything like that. I am trying to design my own reflector for 100* viewing angle. Any tips for that parabolic shape?

Also, is sketch-up 3D printer compatible?

Thanks in advance!

Parabolic ?

Something Like this ?
parabolic.jpg
Did you search the 3D WareHouse ?
Maybe there's already a similar(to the desirable ) model uploaded there ...
( Tab : File => 3D WareHouse )



Dunno really ,for the compability of SketchUp and 3D printers .
I think not directly ..
You probably have to do some file format conversion .
I.e . SketchUp file to somekind of Mesh or CAD file ,first ...
Or more complicated stuff ...
I do not know ,really ...
 
Last edited:

zangtumtum

Well-Known Member
for 3D printers the format file normaly to use it's .STL, you must convert to this format.
I have read that for SketchUp there is a plug-in open source to install.
Other way,I think you can use Blender (freeware) for open and convert in many format.
About the parabolic shape with controlled optical performance it's not easy to draw perfect,
and finaly to controI the quality of output file(STL) in tasselete process,
first you must start to work with stringent tollerance,
other problem it's material of 3D print you need and the quality of printer.....
But this is another story...
Sorry for OT
 
Last edited:

stardustsailor

Well-Known Member
The bottom surface of the CXA30x0 array is :

0.02735 m x 0.02735 m = 0.0007480225 square meters ... (m^2 )

A= 0.000748

Let us follow Cree 's recommendations
and TIM thickness should not be thicker than 50 microns ...
50 microns = 0.05 mm = 0.00005 m

L = 0.00005

So ,having those two as constants ...
The thermal resistance of TIM is :
0.00005 / ( 0.000748 * k )

tim 1.JPG
 

stardustsailor

Well-Known Member
Rθj-c CXA30x0 = ΔΤ { Tj - Tc (aka Tsp ) } / Heat Power = 0.8 C / W ....or K /W ....

Veros are so bad here....
see theVero 29
Data
View attachment 3167726
???
Bad ? !!!!

They have even lower Thermal resistance than the CXA's !!!
( Way lower ....:shock::shock::shock: ... )

image003.jpg

Physical dimensions of Vero 29 ?

LES of 29.2 mm ...
Surface Area = π * 14,6 mm ^2 = 669.66 square mm.

..............................................................................................

Qcond = -k * A* ( Δτ / Δx )

Fourier’s law of heat conduction.
where:
Qcond
is the amount of heat transferred through conduction (W)
k is the thermal conductivity of the material (W/m K)
A is the cross sectional area of the material through which the heat flows (m2)
ΔT is the temperature gradient across the material (°C)
Δx is the distance for the heat must travel (m)

1 / Qcond = Δx / ( -k * A* Δτ ) =>

Rθj-C = Δx / ( -k * A )


Plus there's the Thermal conductivity value k ...
(*Although my guess is , that both Bridgelux and Cree use similar ceramic materials,
for their COBS ..Most probably Ca doped Aluminium Nitride .Take it with a grain of salt.)

So ,the difference in thermal Conductivity between Vero29 and CXA3070 ,
partially can be explained by the larger LES of the VERO29 ....

( CXA LES surface : 415.475 square mm )
 
Last edited:

stardustsailor

Well-Known Member
Different Thermal pastes have different k values ....

An 'base limit ' k value should be 6.5 W/ mK .
( TIM= 0.01 °C/W =max TIM thermal resistance for CXA30x0 series . )

Thermal pastes with lower k values than 6.5 W/mK ,should be avoided .

tim cxa30x0.JPG
 
Last edited:

stardustsailor

Well-Known Member
Some Thermal Paste recommendations and info ....
( Thermal Paste Compounds ,presented are of the highest k scoring )

1) Gelid GC Extreme

k= 8.5 W/mK
~2 EUR / gram

2) JunPus D9000
k= 9.6 W/mK
(diamond nano-particles)
~5 EUR / gram
3) JunPus DX1
k= 16 W/mK
(diamond nano-particles)
~4.5 EUR / gram

4) Dimastech HTX EE
k= 8.6 W/mK
~1.7 EUR / gram

5) Prolimatech PK-1
k= 10.2 W/mK
~1.7 EUR / gram

6) Prolimatech PK-2
k= 10.2 W/mK
~2 EUR / gram

7) Prolimatech PK-3
k= 11.2 W/mK
~3 EUR / gram

8 ) Arctic Silver 5
(Silver-Ceramic particles)
k= 8.9 - 9 W/mK
~1.7 EUR / gram

9) Arctic Silver MX-4
k= 8.5 W/mK
~1 EUR / gram

10) Fischer Electronic WLPK
k= 10 W/mK
~1.4 EUR / gram​

11) Coollaboratory Liquid Pro
k=82 W/mk
!!! cannot be used with aluminium heatsinks !!!


12 ) Coollaboratory Liquid Ultra
k=38.4 W/mK

!!! cannot be used with aluminium heatsinks !!!

13 ) be quiet! DC 1
k= 7.5 W/mK
2.5-3 EUR/ gram

14 ) Cooler Master Extreme Fusion X1
k= 9.5 W/mK
~2.5 EUR / gram
 
Last edited:

Getgrowingson

Well-Known Member
good info here man thanks! how do you reccomend applying the past and maintaining such a thin layer with no air bubbles under it? About to get mounting my cobs and this seems to have many different ways but you seem to have a very good grasp of proper ways of doing things and would appreciate your input!
 

Rahz

Well-Known Member
Razor for sure. You don't have to guess at what the thickness will be and have a high degree of control over the application.
 

Getgrowingson

Well-Known Member
Razor for sure. You don't have to guess at what the thickness will be and have a high degree of control over the application.
How are you using the razor? I was thinking about using masking tape on my heatsink and using maybe a (razor?) as a bridge? between the two pieces of masking tape to ensure an even thin layer of TIM. Just need to figure out a good way to apply to the cob now.....
 

Rahz

Well-Known Member
I haven't tried using a stencil yet, though I think it's a great idea. I just place a line on one side and drag it with the razor, a couple passes to distribute the paste well and then a single pass or two to create the thin flat surface.
 
Top