Painted Heat Sinks

[Spehro Pefhany]

While playing with my new IR thermometer, I tried measuring the light
fixture above my drafting table. The polished brass part measured well
over 100C with a thermocouple but barely above room temp with the IR.

John

Perhaps the IR thermometer's assumption of e = 0.95 was a mite off.



Best regards,
Spehro Pefhany

 

[Kryten]

Have you had any experience with painted heat sinks? I have a customer

who would like a painted finish and I'm having a hard time finding the
thermal conductivity of paint so I can calculate what the thermal
effect would be.

Heatsinks are not painted because they don't need to be.
They are usually made from aluminium or copper and run hot, so it is not
going to suffer from damp-related corrosion. Aluminium ones are often
anodised and black.

High temperatures might make the paint flake off, you could see what they
use for radiators.

 

[Ignoramus20878]

I wonder if anyone has suggestions for infrared remote thermometers
with laser sights.

i

 

[Ken Taylor]

I wonder if anyone has suggestions for infrared remote thermometers
with laser sights.

i

Our sparky uses one when checking the power switchboards. Don't recall the
make off-hand but I think it's a Fluke.

Ken

 

[Ross Herbert]

Morning,

Have you had any experience with painted heat sinks? I have a customer
who would like a painted finish and I'm having a hard time finding the
thermal conductivity of paint so I can calculate what the thermal
effect would be.

Any suggestions?

Thanks,
Montie
[email protected]
Montie Design
www.montie.com

Rod Elliot's (ESP website) contains some information on this subject.
http://sound.westhost.com/heatsinks.htm

 

[John Larkin]

Perhaps the IR thermometer's assumption of e = 0.95 was a mite off.

Is it true that emissivity + reflectivity = 1?

That seems intuitively to be the case. The brass was shiny enough in
the IR that what I was measuring was the reflected image of the room,
not the temp of the brass itself.

So you can measure emissivity at thermal wavelengths by heating up the
thing in question and scoping it with a cheap IR thermometer. There
must be an equation somewhere.

John

 

[John Larkin]

So, how much power does it take to raise the 100 sq cm heatsink in your
first paragraph to 50C? How much of the total power is radiated?

The still-air convection cooling of a surface is messy: depends on
size, shape, orientation. Figure very roughly 150 k/w per square inch,
or 1000 k/w per square cm (although it's nonlinear as hell). Assuming
that, my 100 sq cm hunk has a convection coefficient of about 10 K/w,
not much less than the 12.5 k/w for the radiation case cited. But if
there was any air flow, or if the heatsink has fins, or if e < 1, the
balance is further tipped in favor of convection.

If anybody has better numbers, I'd like to hear them.

John

 

[John Larkin]

I wonder if anyone has suggestions for infrared remote thermometers
with laser sights.

i

I have a cheap ($70 roughly) Extech that seems pretty accurate, but
Fluke has a new one (for $99 I think) probably better quality and it
has a smaller field of view, which is good.

Fluke also has a full IR imager for a mere $9995, which is sorely
tempting.

John

 

[Ignoramus20878]

I have a cheap ($70 roughly) Extech that seems pretty accurate, but
Fluke has a new one (for $99 I think) probably better quality and it
has a smaller field of view, which is good.

That's interesting, do you have any model #s? I have a Extech
multimeter and it is super nice (military surplus).

i

 

[Dave]

If anybody has better numbers, I'd like to hear them.

36 - 28 - 36

 

[Jasen Betts]

Morning,

Have you had any experience with painted heat sinks? I have a customer
who would like a painted finish and I'm having a hard time finding the
thermal conductivity of paint so I can calculate what the thermal
effect would be.

Any suggestions?

It depends on the composition of the paint.

is an anodized finish unsuitable?

Bye.
Jasen

 

[Jasen Betts]

His example 2 is just silly. If you glop on tons of silicone grease
and apply even modest mounting pressure, it oozes down to under 100
microinches.

It depends on the heatsink compound. some formulations have copper or
silver particles in them and the only ooze down to the diameter of the
pparticles.


Bye.
Jasen

 

[Terry Given]

Powder coat is probably thicker than standard paint, which may hinder things.

greg

I've done the calcs for 0.15mm thick powder coat, and its a piss-poor
thermal conductor - IIRC a few thousand times higher than aluminium.
which is why we anodise that particular heatsink. And why we castigated
the twit who decided powder coating was a good idea.

Cheers
Terry

 

[Terry Given]

His example 2 is just silly. If you glop on tons of silicone grease
and apply even modest mounting pressure, it oozes down to under 100
microinches.

John

and what about the +/-0.15% change in oscillator frequency - thats
hilarious. esp. cf the initial accuracy of the chip itself, which is
+/-8%. the datasheet actually gives a 10% temp. stability spec....

I used an LM2660 a while back, its oscillator is specd at 40-80kHz!

Cheers
Terry

 

[John Larkin]

It depends on the heatsink compound. some formulations have copper or
silver particles in them and the only ooze down to the diameter of the
pparticles.

Sure, but I meant the regular white thermal grease. Just hand pressure
will squeeze it below 100 microinches, which was my measurement
resolution.

One fun thing to do is to add a dash of Cataphote glass beads (they're
used as retroflectors, usually) into thermal epoxy, to force a small
insulating gap. I guess some fine sand would work, too.

Here in California, they stencil big STOP letters in the street at an
intersection, in addition to the regular STOP sign. They put down a
stencil, spray white paint, remove stencil, then toss handfulls of the
Cataphote stuff onto the wet paint. The wind blows it around, and the
sidewalks sparkle for some weeks until it washes away.

It also comes in "spacer grade", with uniform glob sizes, but that's
much more expensive.


John

 

[John Larkin]

36 - 28 - 36

Fat! 34 - 22 - 34 will have better longterm reliability.

John

 

[John - KD5YI]

The still-air convection cooling of a surface is messy: depends on
size, shape, orientation. Figure very roughly 150 k/w per square inch,
or 1000 k/w per square cm (although it's nonlinear as hell). Assuming
that, my 100 sq cm hunk has a convection coefficient of about 10 K/w,
not much less than the 12.5 k/w for the radiation case cited. But if
there was any air flow, or if the heatsink has fins, or if e < 1, the
balance is further tipped in favor of convection.

If anybody has better numbers, I'd like to hear them.

John

Are you saying (for this simplistic example) that the power dissipated
through radiation is about 4 Watts and the power dissipated through
convection is about 5 Watts? If so, then the radiation is a very significant
portion of the total.

John, too.

 

[John Larkin]

Are you saying (for this simplistic example) that the power dissipated
through radiation is about 4 Watts and the power dissipated through
convection is about 5 Watts? If so, then the radiation is a very significant
portion of the total.

John, too.

Yes, but the numbers are very rough and the assumptions - zero air
flow, e=1, 50C rise, no heatsink fins, no reflections from enclosure
walls - unreasonably favor radiation.

If we use the somewhere-cited convection equation

Theta = 50/(sqrt A) A in sq cm,

Then the convection theta here is 5 K/w, further diluting the share
assignable to radiation.

But the only way to really predict a heatsink's behavior is to test
it.

John

 

[John Larkin]

That's interesting, do you have any model #s? I have a Extech
multimeter and it is super nice (military surplus).

i

The Extech is a 42520. Seems pretty good.

John

 

[John - KD5YI]

Yes, but the numbers are very rough and the assumptions - zero air
flow, e=1, 50C rise, no heatsink fins, no reflections from enclosure
walls - unreasonably favor radiation.

If we use the somewhere-cited convection equation

Theta = 50/(sqrt A) A in sq cm,

Then the convection theta here is 5 K/w, further diluting the share
assignable to radiation.

But the only way to really predict a heatsink's behavior is to test
it.

John

Then that would mean 10 Watts convected and 4 Watts radiated (about 28%).

In fact, I usually guess at about 30 to 35% radiation under conditions that
would be ideal for radiation. Still, that is a rather significant amount and
should not be ignored.

You are correct, of course, that the only way to know for sure is to test.
However, I find it handy at times to estimate such things before having the
object available for tests.

Returning to the subject of painting the heatsink, I remember a design
manual I had about 30 years ago mentioning that the radiation of a heatsink
could be improved by painting with oil-based paint. They said the color was
relatively unimportant. The booklet was produced by Motorola or National, I
think, and it concerned designing voltage regulator circuits. In fact, I
think the manual's title was something like Voltage Regulator Design
Handbook. But, don't hold me to that -- thirty years is a long time.

Cheers,
John, too.