High power LEDs on heatsinks

[Dirk Bruere at NeoPax]

How are batwing high power LEDs attached to heatsinks?

 

[Raveninghorde]

How are batwing high power LEDs attached to heatsinks?

Solder to aluminium clad pcb and heatsink the metal side.

This type of pcb has a very thin insulator layer between the copper
and aluminium so has good thermal conductivity.

 

[Dirk Bruere at NeoPax]

they generally get soldered to aluminum substrate PCBs,which then are
bolted to larger heatsinks.

for examples;
http://www.theledlight.com/LuxeonLEDs.html

you can buy the aluminum substrates from Deal Extreme.(Hong Kong)

Are they soldering iron solderable or is it an oven job?

 

[Kevin McMurtrie]

they generally get soldered to aluminum substrate PCBs,which then are
bolted to larger heatsinks.

for examples;
http://www.theledlight.com/LuxeonLEDs.html

you can buy the aluminum substrates from Deal Extreme.(Hong Kong)

Does anyone sell un-etched aluminum-clad PCB to individuals?

I've had to resort to double-sided 0.01 inch PCB glued to a heatsink.
It works well thermally but bonding anything to copper is a PITA. I
have to use copper pipe adhesive sealed with anti-oxidation lacquer.
It's hobby-grade at best.

 

[Raveninghorde]

Is there something wrong with (standard) copper?

On standard board most of the thickness is fibreglass. On aluminium
board the insulator layer is thin and most of the thickness is the
aluminium. Vastly better heat transfer from front to back of the
board.

 

[Raveninghorde]

Are they soldering iron solderable or is it an oven job?

For prototypes it can be soldered by hand.

Until we got our new oven we had to have production boards made out of
house to make sure we kept within the soldering profile of the LEDs.

 

[Dirk Bruere at NeoPax]

On standard board most of the thickness is fibreglass. On aluminium
board the insulator layer is thin and most of the thickness is the
aluminium. Vastly better heat transfer from front to back of the
board.

How about sticky backed copper foil on a Al heatsink?

 

[Raveninghorde]

How about sticky backed copper foil on a Al heatsink?

It should be OK for a prototype.

How many LEDs in series? You might want to check the voltage that can
be applied to the insulator on the foil. My last design the LED string
had 70V across it.

 

[Don Klipstein]

I use thermal epoxy. Gets tricky IF the LED is not isolated. Easy
if it is. For insulation I used diamond dust.

My favorite prototyping epoxy for gluing heatsinkable LEDs to heatsinks
is the "4-minute" epoxy for metal that some auto parts stores have. IIRC,
twin-tube syringe-style dispenser, one part black, the other gray. In my
experience so far, dries quickly, gets pretty-much as hard as a rock,
sticks well to metal and almost anything else, and conducts heat better
than plain epoxy. After that, look for "plastic steel" epoxy.

My favorite electrical insulator between a semiconductor's heatsinkable
surface and a heatsink in prototypes: Plastic grocery bags, especially
ones that "crinkle" more loudly in the process of being squeezed into a
little ball. I find that epoxy sticks usably well to those, even though
most of these are made of HDPE and HDPE is a form of polyethylene, noted
often to be a "hard to glue" plastic. If you can get somke nice thin
plastic grocery bags with the /2\ recycling symbol (HDPE), that has better
heat conductivity than others. -
And a ~.02 mm thickness of plastic is less of a barrier to heat than .25
mm (estimate) of mica. Mica's heat conductivity is only twice that of
HDPE and around 2.5-4 times that of most other plastics, maybe 4.5 times
that of PETE (recycling symbol /1\) and 4.5-5 times that of a few others.
-

 

[Kevin McMurtrie]

Bonding via flow soldering?

I haven't figured out how to do that easily. Maybe the board could be
soldered to a plate with high temperature solder then the LEDs applied
to the board with low temperature solder. I don't know if high temp
solder remains strong enough for that to work without the PCB popping
off and scattering $150 of LEDs all over the place. Another option may
be creating a silicone slab that could press the hot PCB down without
touching the LEDs. It would require some testing to see how big those
holes need to be to allow for thermal distortion. A grid of screws
won't work because they will interfere with surface mount lenses.

I still think etching aluminum-clad PCB would be easier for personal
projects.

 

[Dirk Bruere at NeoPax]

I haven't figured out how to do that easily. Maybe the board could be
soldered to a plate with high temperature solder then the LEDs applied
to the board with low temperature solder. I don't know if high temp
solder remains strong enough for that to work without the PCB popping
off and scattering $150 of LEDs all over the place. Another option may
be creating a silicone slab that could press the hot PCB down without
touching the LEDs. It would require some testing to see how big those
holes need to be to allow for thermal distortion. A grid of screws
won't work because they will interfere with surface mount lenses.

I still think etching aluminum-clad PCB would be easier for personal
projects.

Alternatives:

a) Built in heatsink - anyone do a basic, cheap, range of deep red LEDs
3W+ with its own heatsink eg TO3

b) Heatsink compound with some kind of mechanical fixture to a standard
heatsink

 

[Kevin McMurtrie]

Alternatives:

a) Built in heatsink - anyone do a basic, cheap, range of deep red LEDs
3W+ with its own heatsink eg TO3

b) Heatsink compound with some kind of mechanical fixture to a standard
heatsink

Too bulky. I'm thinking of densely mounted LEDs for portable lighting.

This is a prototype that I photographed:
http://www.pixelmemory.us/Photos/Nerd/Alien Bike Light 5.0/

The dense mounting allows for momentary use at extremely high intensity
or continuous use at low power where the LEDs are most efficient. I'm
using the final product as a bicycling and hiking light. It normally
consumes under 2W but can temporarily operate at up to 60W for finding
landmarks in remote areas. It continuously adjusts from a faint glow to
brighter than car highbeams.

I'd like to build a more compact version using smaller lenses but I
don't have a good solution for transferring such a concentrated source
of heat. Building with aluminum rather than copper, pyrolytic graphite,
and carbon fiber would simplify the design a lot too.