DIY : Soldering Leds to heatsink ,without MCPCB.Doable ?

I'm starting this thread ,so to analyse together with those interested ,
if it is possible to avoid the use of a Metal Core Printed Circuit Board .

In fact my idea is based on "transforming " the heatsink itshelf into a " pcb " .....

Print the circuit ,directly onto the heatsink ,which is made of Aluminium ....

The advantages of succeeding to do it ,are :

-Superior Thermal management => Higher efficiency .

-Versatily of leds that can be used (mixed brands ) and positioning of them .

-More compact and simple design .Less trouble and tools .
No screws ,rivets,pads,paste,glue ,cables ,star pcb and pcbs at general ...

-Can even implement a DIY CC driver on same heatsink with leds .(Advanced electronic skills )

---------------------------------

Summary of idea :

First Step : Prepare Aluminium Surface of Heatsink and apply ( using spray ) a thin epoxy layer as electrical insulation layer .

Second Step : Attach copper sheet(plate ) ,over the insulating layer of epoxy .

Third Step : Design Circuit with image-processing / CAD software and print it
at appropriate size,on normal paper ,using a Laser Printer .

Fourth Step :Thermal Transfer laser-printed circuit to copper plate and with Iron(III) chloride* etch the circuit to the copper plate.

Fifth step :Clean the laser printer toner mask ,from the etched circuit .

{{{{Additional Step : Mask contact points with i.e . selotape pcs.
Apply final protective epoxy coat .
Unmask contacts from selotape ...}}}}}

Next Step: Solder the leds .....
...

* Iron(III) chloride
....Another important application of iron(III) chloride is etching copper in two-step redox reaction to copper(I) chloride and then to copper(II) chloride in the production of printed circuit boards ....

FeCl[SUB]3[/SUB] + Cu → FeCl[SUB]2[/SUB] + CuCl
FeCl[SUB]3[/SUB] + CuCl → FeCl[SUB]2[/SUB] + CuCl[SUB]2[/SUB]


https://en.wikipedia.org/wiki/Iron(III)_chloride

Click to expand...

....
Now ....
The detailed version to follow ....

Before continuing ,it should be noted that I'm planning on using
A certain brand of Epoxy .

West Systems .
http://www.westsystem.com/ss/


Now ....

Surface of heatsink should be flat ( ! ) and without cracks , holes ,ect ...

To bond with liquid two -part epoxy ,this preparation kit is better used :

http://www.westsystem.com/ss/surface-prep

But this only after the surface is lightly "cleaned " with steel-wool (to make it bond better with epoxy ) and de-fatted with acetone or other organic solvent ...

Now ...
First hard part .....

How to apply the epoxy ?

Thinking of using compressed air ....
Like with thin layers of paint ....
...
Two ways there ....
A+B epoxy parts ,sprayed together (pre-mixed )...
Or
....first spray the A part and then the B part .(in-situ mix ..Will it work ? A test will show ....)..

Thickness of this layer should not be more than 40-60 microns ( 50 µm ,average) ..

(Another ,second layer of same thickness ,
will be used -after first layer has fully hardened/cured- to bond copper sheet/plate with heatsink ...
So a total thickness of 80-120 microns of epoxy layer is desirable ...
No less -no more...
(less will not ensure enough electrical insulation ,more will act as thermal insulation .)

step one is not necessary, most Heat-slug are isolated or have a ceramic base.

http://www.led-tech.de/en/High-Power-LEDs-Cree/CREE-XM-Serie/CREE-XM-L-T5-on-Copper-PCB-LT-1810_120_170.html

Yep ,but copper lines and contacts of rest of circuit has to be insulated from heatsink ...
Otherwise there's no circuit at all .....

Heatsink surface has to be totally insulated ,so for the circuit to "sit" to the heatsik ..

And yes many (most by now ? ) type of leds have isolated thermal slugs ....
Still you cannot solder them in bare epoxy(or aluminium ) ...
You need a copper contact (not part of rest of circuit of course ... ),for the thermal slug to be soldered ..
Otherwise thermal conductive glue/paste or pad , have to be used, between bare epoxy (or hole in epoxy layer ,direct to heatsink ,let's say ... ) ...
Trying to leave an non epoxy-covered (masked ) area for the led thermal slug to sit directly on heatsink,
can be way tricky with i.e Oslons ...
3.1 x 3.1 mm led ...
Tiny little " sun " ....
With 3 contacts underneath ,so close together ....
More easy to design & etch a circuit for all of them ,than try to make insulated contacts for the two ,
while leaving a bare gap , at the heatsink surface for the middle one to sit on ..
(.Hope you understand what I'm trying to say .)
So ....


Now ...Take a look into some parts there ...
All ideas ,are very welcome ....

Hardeners :
http://www.westsystem.com/ss/hardener-selection-guide/

? fillers ? : http://www.westsystem.com/ss/fillers

Additives : http://www.westsystem.com/ss/additives

FAST Forward ...

Part 1 : (this is an Fiberglass pcb ... Same approx method will be followed to print the circuit on copper plate/epoxy layer / heatsink
Applying "laser printer toner" mask of circuit ...
All you need is the print of the circuit from a laser printer ..(toner is polymer -plastic )
[video=youtube;TNdJC7evW70]http://www.youtube.com/watch?v=TNdJC7evW70&feature=youtu.be[/video]

Part two
Etching with Iron(III) chloride (not particularly a hazardous or dangerous substance except that it stains clothing forever ....)
: [video=youtube;8WRP9WiSnYo]http://www.youtube.com/watch?v=8WRP9WiSnYo&feature=youtu.be[/video]

The whole idea is to keep the "layers " of a MCPCB that are needed ( Protective coat/finish -circuit/contacts - dielectric layer ) and
"disgard " the "middle-men layers" ...( alum plate of pcb -core- and thermal conductive layer between pcb and heatsink ....

The new "core" is the heatsink itshelf ...
The Pcb is at the same time the heatsink ....

The whole circuit is bonded with a thin epoxy layer with the heatsink ....

Various custom designs can be made this way ....

Thermal managment is superior than if a PCB/star PCB was used firstplace ..

(Path : Die =>Slug =>solder=>copper =>epoxy=>heatsink(alum)=>air instead of ..

Die =>Slug =>solder=>copper =>epoxy=>alum (pcb core plate )***=>***pad/paste/glue ***=>***heatsink(alum)=>air )...

****The weakest link of the thermal chain,takes a hike ....

This is a good idea which I have actually used a few times. There are a few things to consider. If you are using different pad layouts for different LEDs (say Osram and Rebels) you have to predetermine their positions and they are not easily changed after the first layout. Goud is correct.

Skip this first step and have a stencil that covers your thermal slug before you lay your first layer of epoxy. Lay your epoxy and have bare metal connection to the slug (if your thermal slugs are electrically isolated).
Lay your trace route for circuitry.
Create a stencil that covers all LED, Driver, and thermal slug connection points and soldier mask.
Mount LEDs.

Using stencils for all of this takes care of the majority of your problems you noted above. Here is where your real issues will lay. Soldiering SMD type chips requires you to reflow the whole heat sink. It is nearly impossible to ramp up and ramp down the whole of the heat sink in the allowable window for SMD LED soldiering. Using the cheap Chinese P2/K2 packaged LEDs would actually be a better choice for this mounting system (this is also why large panel companies still use old tech). They are much easier to mount than SMD diodes. I will disagree with Goud on this point. MOST LEDS are not electrically isolated to the thermal slug. Most of the higher quality diodes are, but the cheaper Chinese versions are not.

Now the technical part of your choices to use west and the one thing that you linked directly to. West systems is a great epoxy company. However, very few if any, of their epoxies are dielectric. Their epoxies are used extensively in the aircraft and boat manufacturing process. The cleaner that you linked is a aluminum detergent to clean and a (alodine/irrodite) chemical conversion coating. This is used to place a thin layer of corrosion over the metal to prevent future corrosion and help in the adhesion of a primer coating. This is not something you would want to do to a heat sink. It will actually decrease the thermal efficiency by a great deal. Secondly, the epoxies cannot be sprayed on Part A then Part B. The mix ratio is almost never 50/50 (usually 20/1 or 33/1) and requires the mixing to start the chemical reaction of the cure process. It is also highly unlikely you would be able to get a layer less than 5-10 mm in thickness with any of the West products. Glypatal is a much better choice for this application IMHO.

Hosebomber said:

This is a good idea which I have actually used a few times. There are a few things to consider. If you are using different pad layouts for different LEDs (say Osram and Rebels) you have to predetermine their positions and they are not easily changed after the first layout. Goud is correct.

Skip this first step and have a stencil that covers your thermal slug before you lay your first layer of epoxy. Lay your epoxy and have bare metal connection to the slug (if your thermal slugs are electrically isolated).
Lay your trace route for circuitry.
Create a stencil that covers all LED, Driver, and thermal slug connection points and soldier mask.
Mount LEDs.

Using stencils for all of this takes care of the majority of your problems you noted above. Here is where your real issues will lay. Soldiering SMD type chips requires you to reflow the whole heat sink. It is nearly impossible to ramp up and ramp down the whole of the heat sink in the allowable window for SMD LED soldiering. Using the cheap Chinese P2/K2 packaged LEDs would actually be a better choice for this mounting system (this is also why large panel companies still use old tech). They are much easier to mount than SMD diodes. I will disagree with Goud on this point. MOST LEDS are not electrically isolated to the thermal slug. Most of the higher quality diodes are, but the cheaper Chinese versions are not.

Click to expand...

Ok ....Still I prefer leds to be attached,in a more "solid " manner ,to heatsink by -also-soldered thermal contacts ....

Now the technical part of your choices to use west and the one thing that you linked directly to. West systems is a great epoxy company. However, very few if any, of their epoxies are dielectric.

Click to expand...

??? I've used it many times ....
I can ensure you that it doesn't conduct electricity .....
In fact it is quite strong electric insulant

Their epoxies are used extensively in the aircraft and boat manufacturing process. The cleaner that you linked is a aluminum detergent to clean and a (alodine/irrodite) chemical conversion coating. This is used to place a thin layer of corrosion over the metal to prevent future corrosion and help in the adhesion of a primer coating. This is not something you would want to do to a heat sink. It will actually decrease the thermal efficiency by a great deal.

Click to expand...

You're right on that ....Didn't thought about it ...

Secondly, the epoxies cannot be sprayed on Part A then Part B. The mix ratio is almost never 50/50 (usually 20/1 or 33/1)

Click to expand...

..more like 5:1 for the west systems .....

and requires the mixing to start the chemical reaction of the cure process.

Click to expand...

I'll test that ....Thing is I'm not making bulk quantity of epoxy .....
Just two very thin layers ....

It is also highly unlikely you would be able to get a layer less than 5-10 mm in thickness with any of the West products. Glypatal is a much better choice for this application IMHO.

Click to expand...

Trust me ,I've layered that epoxy ,(with a normal brush ) in layers under 1 mm ,while making some
bike-stuff with carbon fiber ....

This epoxy is rock-hard when cured ,transparent and very easy to work with ....(into very thin layers ....)

But thanx for the Glypatal info ...I'll have a look into that ,also ....

Hosebomber said:

This is a good idea which I have actually used a few times. There are a few things to consider. If you are using different pad layouts for different LEDs (say Osram and Rebels) you have to predetermine their positions and they are not easily changed after the first layout. Goud is correct.

Skip this first step and have a stencil that covers your thermal slug before you lay your first layer of epoxy. Lay your epoxy and have bare metal connection to the slug (if your thermal slugs are electrically isolated).
Lay your trace route for circuitry.
Create a stencil that covers all LED, Driver, and thermal slug connection points and soldier mask.
Mount LEDs.

Click to expand...

Ok....

Let's say I do so ....

Say I use my favourite & beloved Oslon Warm Whites ....

Having a "contact pad format " of :


Say I use the heatsink already in use :

200 x 160 x 50 mm ...

I ll try to lay 5 rows of 7 leds ( " 35 Watt " ) ...

Oslons are pretty tiny ... Just 3.1 x 3.1 mm !!!

Problem #1 : ... ( Not in scale )




100 microns of epoxy plus -another 100 microns of copper ?
0.2 mm distance between led thermal slug and heatsink ....
A good gap there ....
Way too much for any thermal conductive glue/grease/pad to "fill" .....
Solder can not be used ...(Alum keeps denying to bond with solder ...)

While the other way .... :



???

.............
While if I'll use the Seoul/Edison type of led (asian / cheapos ) ...
Which have a pad of :


Then I could bend leds " contacts " ,so for the slug to touch the heatsink ..
Thing is that in this type of leds ,slug is not isolated ....
It can not be done,neither with this type of leds .....

So for now ...
Problem #1 of first step :

Bondage/adhesion of epoxy with Aluminium surface of heatsink .


If special primers for epoxy/aluminium ,do form a transparent layer of Alumina
(oxidized form of Aluminium ) then yes ,thermal / cooling efficiency is decreased .

Pic of a MCPCB : .....


I've easily achieved thinner or of same thickness constructions using 3-4 layers of Carbon Fiber cloth and more than 10 coatings of West Systems epoxy ....

Here a detail from an old Carbon Fiber home made part (actually is an experimental design as a" tanto style" hunting knife's hand-guard.. ) :
=>here can be noticed the 3-4 layers of Car. fiber cloth....
Thickness is approx the same with pcb .... (~1 mm ) ...
More than 10 coating -with a normal brush - where applied ....
Then baked in oven (for bubbles to go out ..) ...


Thing is that I'd used Aluminium molds without PVA pre-coating ..
Epoxy didn't adhere to aluminium ....

So ....That is problem ....

Primers can not be used ......

Solution #1 :

Drilling small-tiny- shallow holes ( pits ) diagonally on heatsink surface ...
These holes will fill with epoxy ,securing the whole epoxy layer to heatsink,when epoxy will harden / cure ....
Kinda like roots ....
Hmm...


Too much work though, if to be made manually...

Steel wool on aluminum is a no no. Pieces of steel will get attached to the aluminum and cause corrosion.
Bead blast.
To get the epoxy to stick to aluminum, the surface needs to be rough, or better yet grooved.
Carbon fiber conducts electricity. (electrolysis)
The expansion coefficient of aluminum is considerably different than epoxy fiberglass CF.

A far better solution would be to use thin flexible circuit board material bonded to the HS after all the diodes are soldered to it.

Solution #2 :thin flexible circuit board material bonded to the HS

Click to expand...

as bad as kapton was

check here
http://www.doosan.com/dse/en/main.do

Great link ,there Guod !

http://www2.dupont.com/Pyralux/en_US/products/laminate/index.html
They give small samples with company name.

There are tons and tons of suppliers for this sort of thing.

Bumping Spheda said:

http://www2.dupont.com/Pyralux/en_US/products/laminate/index.html
They give small samples with company name.

There are tons and tons of suppliers for this sort of thing.

Click to expand...

That is GREAT !
Exactly what I was looking for ....

( I hope it will be better than Capton .....)

http://www.doosan.com/dse/en/products/heat.page?

http://org-www.doosan.com/dse/attach_files/pdf/Heat_Materials.pdf


Dielectric layer : 95 -100 um (microns ) of epoxy with ceramic filler ....


So ...I still like to stick to the epoxy insulating layer (dielectric ) ...

...

Thought you said Capcom, was gonna hafta open a can.

The Kapton and Copper clad thickness are variable. I think the thinnest Kapton they provide is like 20 micron, but that's completely off memory so don't quote me. Still, it's damn thin. I'm not exactly sure about power handling capabilities at that point, but LED's aren't very hungry (why we love them ) so I don't really foresee an issue here.