### Network

J

#### John Larkin

Jan 1, 1970
0
There's all types of glass. Finding one that conducts the most heat
would be a interesting.
D from BC

You can use a thermally conductive epoxy that has very small fill
particles, and add just a little dab of glass beads to enforce spacing
until the epoxy sets. The thermal properties of the beads doesn't
matter.

John

D

#### D from BC

Jan 1, 1970
0
On Tue, 06 Mar 2007 16:09:05 -0800, John Larkin

[snip]
You can use a thermally conductive epoxy that has very small fill
particles, and add just a little dab of glass beads to enforce spacing
until the epoxy sets. The thermal properties of the beads doesn't
matter.

John

I think thermal epoxies have better thermal conduction than glass. I
might shoot for minimal glass beads.
Also, too much glass and the bond strength probably gets worse.
Too few glass beads and the part shorts out if it's glued on crooked.
I'll guess...maybe 10 parts glass beads to 1 part epoxy....
Somewhat of a probability problem.

Dabbing sounds better than math..
D from BC

D

#### D from BC

Jan 1, 1970
0
On Tue, 06 Mar 2007 16:09:05 -0800, John Larkin

[snip]
You can use a thermally conductive epoxy that has very small fill
particles, and add just a little dab of glass beads to enforce spacing
until the epoxy sets. The thermal properties of the beads doesn't
matter.

John

I think thermal epoxies have better thermal conduction than glass. I
might shoot for minimal glass beads.
Also, too much glass and the bond strength probably gets worse.
Too few glass beads and the part shorts out if it's glued on crooked.
I'll guess...maybe 10 parts glass beads to 1 part epoxy....
Somewhat of a probability problem.

Dabbing sounds better than math..
D from BC

oops ..Correction..
Not 10 parts glass beads to 1 part epoxy..
Should be
10 parts epoxy per glass bead...
D from BC

J

#### John Larkin

Jan 1, 1970
0
On Tue, 06 Mar 2007 16:09:05 -0800, John Larkin

[snip]
You can use a thermally conductive epoxy that has very small fill
particles, and add just a little dab of glass beads to enforce spacing
until the epoxy sets. The thermal properties of the beads doesn't
matter.

John

I think thermal epoxies have better thermal conduction than glass. I
might shoot for minimal glass beads.
Also, too much glass and the bond strength probably gets worse.
Too few glass beads and the part shorts out if it's glued on crooked.
I'll guess...maybe 10 parts glass beads to 1 part epoxy....
Somewhat of a probability problem.

Dabbing sounds better than math..
D from BC

oops ..Correction..
Not 10 parts glass beads to 1 part epoxy..
Should be
10 parts epoxy per glass bead...
D from BC

No, 100 parts epoxy to 1 part beads.

John

D

#### D from BC

Jan 1, 1970
0
On Tue, 06 Mar 2007 16:09:05 -0800, John Larkin

[snip]

You can use a thermally conductive epoxy that has very small fill
particles, and add just a little dab of glass beads to enforce spacing
until the epoxy sets. The thermal properties of the beads doesn't
matter.

John

I think thermal epoxies have better thermal conduction than glass. I
might shoot for minimal glass beads.
Also, too much glass and the bond strength probably gets worse.
Too few glass beads and the part shorts out if it's glued on crooked.
I'll guess...maybe 10 parts glass beads to 1 part epoxy....
Somewhat of a probability problem.

Dabbing sounds better than math..
D from BC

oops ..Correction..
Not 10 parts glass beads to 1 part epoxy..
Should be
10 parts epoxy per glass bead...
D from BC

No, 100 parts epoxy to 1 part beads.

John

Let's say.. the hot part has a 5mm x 5mm contact area with the heat
sink.
Assume the glass beads are 0.1mm in diameter.
There are 50 x 50 places for glass beads.
Using a 100:1 epoxy to glass bead ratio yields 25 glass beads over a
5mm x 5mm contact area.
Or 1 bead per sq mm..

Kinda like holding up a table with 25 legs...
100:1 looks good.
I just guessed at the 10:1 ratio..
Thanks
D from BC

G

#### GregS

Jan 1, 1970
0
There's all types of glass. Finding one that conducts the most heat
would be a interesting.
D from BC

Diamonds.

I have used very small dust just for insurance. The micron sized
particles a a bit small, but clumps of the diamonds work for me.
Diamond is something like 20 times the conductance of copper.
Cheaper than you might think. I got three vials for $50, probably at least 20-25 teaspoons worth.15-25u 100 cts. greg G #### GregS Jan 1, 1970 0 Diamonds. I have used very small dust just for insurance. The micron sized particles a a bit small, but clumps of the diamonds work for me. Diamond is something like 20 times the conductance of copper. Cheaper than you might think. I got three vials for$50, probably at
least 20-25 teaspoons worth.15-25u 100 cts.

That was just one vial, I think I ordered 250 cts.

greg

J

#### John Larkin

Jan 1, 1970
0
Diamonds.

I have used very small dust just for insurance. The micron sized
particles a a bit small, but clumps of the diamonds work for me.
Diamond is something like 20 times the conductance of copper.
Cheaper than you might think. I got three vials for $50, probably at least 20-25 teaspoons worth.15-25u 100 cts. greg Diamond grit is not super expensive, but I doubt it's worth it as a thermal filler. The thermal conductivity of a filled grease or epoxy is dominated not by the filler but by the epoxy or grease itself. There's not very good contact between the filler particles, so even if they were thermal superconductors the net result wouldn't be much better than some standard ceramic. Diamond particles as epoxy spacers are really no better than glass beads, since the density is low. The most important thing about spacer beads is that the gap be accurately controlled, and most grits or sandy things don't do that well. Cataphote does sell "spacer grade" beads, sorted for diameter, but they're very expensive. The key to good thermal conductivity is to have very flat, smooth surfaces and, if you need insulation, the absolute minimal insulating gap. A thin wafer of BeO or AlN, with silicone grease, under moderate pressure, is about as good as you can reasonably do. No spacered filled epoxy will be that good. Hard anodize and grease is very good, too. John R #### Rich Grise Jan 1, 1970 0 No, 100 parts epoxy to 1 part beads. Just out of curiosity, where do you get these things, and how big should they be? I googled for "glass beads" and got 30,000 pages on how to make a necklace. ;-) Thanks, Rich J #### John Larkin Jan 1, 1970 0 Just out of curiosity, where do you get these things, and how big should they be? I googled for "glass beads" and got 30,000 pages on how to make a necklace. ;-) Thanks, Rich Cataphote. Or your nearest beach, maybe. The Pacific stuff is kind of coarse and non-uniform; I'd guess that Pensacola has the best domestic sand. The Cataphote stuff is sold mostly as retroreflectors, but they do sell "spacer grade" beads. John G #### GregS Jan 1, 1970 0 Diamond grit is not super expensive, but I doubt it's worth it as a thermal filler. The thermal conductivity of a filled grease or epoxy is dominated not by the filler but by the epoxy or grease itself. There's not very good contact between the filler particles, so even if they were thermal superconductors the net result wouldn't be much better than some standard ceramic. Diamond particles as epoxy spacers are really no better than glass beads, since the density is low. The most important thing about spacer beads is that the gap be accurately controlled, and most grits or sandy things don't do that well. Cataphote does sell "spacer grade" beads, sorted for diameter, but they're very expensive. The key to good thermal conductivity is to have very flat, smooth surfaces and, if you need insulation, the absolute minimal insulating gap. A thin wafer of BeO or AlN, with silicone grease, under moderate pressure, is about as good as you can reasonably do. No spacered filled epoxy will be that good. Hard anodize and grease is very good, too. They do sell flat polished diamond specifically for his purpose. Some new semiconductors use diamond technology. greg greg D #### D from BC Jan 1, 1970 0 On Wed, 07 Mar 2007 13:48:37 GMT, [email protected] (GregS) wrote: [snip] Diamonds. I have used very small dust just for insurance. The micron sized particles a a bit small, but clumps of the diamonds work for me. Diamond is something like 20 times the conductance of copper. Cheaper than you might think. I got three vials for$50, probably at
least 20-25 teaspoons worth.15-25u 100 cts.
greg

Probably ok for low breakdown voltage but the glass beads can set a
precise predictable bond thickness.
The more bond thickness, the higher the breakdown voltage.
D from BC

D

#### D from BC

Jan 1, 1970
0
Diamond grit is not super expensive, but I doubt it's worth it as a
thermal filler. The thermal conductivity of a filled grease or epoxy
is dominated not by the filler but by the epoxy or grease itself.
There's not very good contact between the filler particles, so even if
they were thermal superconductors the net result wouldn't be much
better than some standard ceramic.

Diamond particles as epoxy spacers are really no better than glass
beads, since the density is low. The most important thing about spacer
beads is that the gap be accurately controlled, and most grits or
sandy things don't do that well. Cataphote does sell "spacer grade"
beads, sorted for diameter, but they're very expensive.

The key to good thermal conductivity is to have very flat, smooth
surfaces and, if you need insulation, the absolute minimal insulating
gap. A thin wafer of BeO or AlN, with silicone grease, under moderate
pressure, is about as good as you can reasonably do. No spacered
filled epoxy will be that good. Hard anodize and grease is very good,
too.

John

BeO or AlN wafers?? I haven't seen that yet.
Hopefully available at Digikey
I'll have to google that someday. ...
D from BC

G

#### GregS

Jan 1, 1970
0
The companies promote diamonds as in the best thermal coumpounds.
One company said they have a superior coumpound compaired to the diamond
using carbon nanotubes.

greg

G

#### g

Jan 1, 1970
0
I'm gluing a part to a heat sink with a non-conductivethermaladhesive.
After cure, testing shows no conduction between the part and the heat
sink. Good..

Problem: After power up, a short occurs between the part and the
aluminum heat sink. After power down, the part is still shorted to the
heatsink!! Bizarre..
There is no testing mistake.
(200VDC exists between part and heat sink during normal operation.)

What's going on?
Is it possible that an small arc develops in theadhesive. The arc
vaporizes some aluminum. Then when the power is cut off the aluminum
vapor cools and develops a conductive bridge (a short).
Possible?
conductive..

Anybody encounter something like this?
The shorting effect doesn't happen when I increase the bond line.
However, I'm trying to keep my bond line to a minimum.
D from BC

When I was epoxying LEDs, I set up a multimeter to measure ohms as I
to make sure there was no low ohm shorts. I don't know how the
Lumileds are spaced
when they are assembled to make stars, but it seems epoxy is the only
spacer and insulator.
greg

J

#### John Larkin

Jan 1, 1970
0
Let's say.. the hot part has a 5mm x 5mm contact area with the heat
sink.

That's tiny. If you need serious cooling, solder or thermal epoxy the
hot gadget to a copper heat spreader and insulate *that* from the
grounded heatsink. More footprint, less theta.

John

Replies
8
Views
2K
Replies
8
Views
1K
C
Replies
6
Views
2K
P
Replies
4
Views
2K
Z
M
Replies
9
Views
5K
M