Diecast boxes heat dissipation

[Kevin Barstow]

Is there any rule of thumb regarding heat dissipation of waterproof
diecast boxes (with no venting) when used as a transistor heatsink?

Kevin Barstow

 

[Phil Allison]

"Kevin Barstow"

Is there any rule of thumb regarding heat dissipation of waterproof
diecast boxes (with no venting) when used as a transistor heatsink?

** What parameter/s are you after ??

And obviously - size matters.


..... Phil

 

[Adrian Tuddenham]

Is there any rule of thumb regarding heat dissipation of waterproof
diecast boxes (with no venting) when used as a transistor heatsink?

Temporarily fix a contact-cooled resistor (or two) inside the box and
run it at various wattages from a bench power supply. To measure the
temperature you can use a thermocouple thermometer or (slightly less
accurately) a liquid-in-glass thermometer with the bulb wrapped in
cooking foil under a small patch of expanded polystyrene.

All your temperatures need to be measured with respect to ambient in
still air and you may need to wait for some time for each reading as the
temperature stabilises.

Large die cast boxes can exhibit considerable temperature variation
between different parts, so it might help to distribute your
power-dissipating devices as widely as possible.


As a guide:

I have built a stereo 30 + 30 watt amplifier in a die cast box approx.
270 x 170 x 70 mm. It runs from 12v supplies and generates an internal
55v rail to run the TDA7295 output stages. Although it is far from
efficient and dumps nearly 50 watts into the box at full power, it only
feels warm on the corners after hours of running flat out into a pair of
dummy loads. You can see the two output stages well separated in
opposite corners:

<http://www.poppyrecords.co.uk/other/images/30W30Wamplifier.jpg>


A mono 60w design for 100v line from 12v supplies, complete with
6-channel microphone mixer is housed in a pair of diecast boxes bolted
together.

<http://www.poppyrecords.co.uk/other/images/60wPAAmplifier.jpg>

The main amplifier box is approx. 190 x 120 x 80 mm and the output
transistors are well distributed on the shorter sides. The output is
transformer push-pull Class-B with two triples in parallel on each side.
At prolonged full power into the correct load, the dissipation is around
30 watts and the sides of the box reach about 50 C above ambient.

On one occasion I was faced with an emergency situation which involved
running over 100 watts-worth of loading at full power for three hours.
On that occasion the sides of the box became too hot to touch, which
made operating the mixing controls rather inconvenient. Eventually I
enlisted the help of another P.A. engineer and we took it in turns to
fan the amplifier vigorously with a large piece of cardboard for the
rest of the evening.

 

[ehsjr]

Temporarily fix a contact-cooled resistor (or two) inside the box and
run it at various wattages from a bench power supply.

Yes! And run it for a loooong time, at least an hour per test at the
selected wattage. The heat just keeps on building and the temperature
climbing until thermal stability where the box dissipates to the
surroundings at the same rate as it absorbs from the R.

<snipped the rest - all good stuff which includes a mention of time>

You should have seen the smoke I got from one such test. At 10
minutes, all was fine in ~ 30 degree F ambient. At 1 hour, not
so much. Melting insulation makes a lot of smoke.

Ed

 

[Cydrome Leader]

Temporarily fix a contact-cooled resistor (or two) inside the box and
run it at various wattages from a bench power supply. To measure the
temperature you can use a thermocouple thermometer or (slightly less
accurately) a liquid-in-glass thermometer with the bulb wrapped in
cooking foil under a small patch of expanded polystyrene.

All your temperatures need to be measured with respect to ambient in
still air and you may need to wait for some time for each reading as the
temperature stabilises.

Large die cast boxes can exhibit considerable temperature variation
between different parts, so it might help to distribute your
power-dissipating devices as widely as possible.


As a guide:

I have built a stereo 30 + 30 watt amplifier in a die cast box approx.
270 x 170 x 70 mm. It runs from 12v supplies and generates an internal
55v rail to run the TDA7295 output stages. Although it is far from
efficient and dumps nearly 50 watts into the box at full power, it only
feels warm on the corners after hours of running flat out into a pair of
dummy loads. You can see the two output stages well separated in
opposite corners:

<http://www.poppyrecords.co.uk/other/images/30W30Wamplifier.jpg>

I see you laced some of the cables- I haven't seen this in quite some
time.

 

[Adrian Tuddenham]

I see you laced some of the cables- I haven't seen this in quite some
time.

It's a lost art
...it's also a sure way to discover you need one more cable in
the loom -just after you have tied the final knot.

 

[Spehro Pefhany]

I see you laced some of the cables- I haven't seen this in quite some
time.

It's a lost art
...it's also a sure way to discover you need one more cable in
the loom -just after you have tied the final knot.[/QUOTE]

Not lost, just misplaced:-

http://www.dairiki.org/hammond/cable-lacing-howto/


Best regards,
Spehro Pefhany

 

[Adrian Tuddenham]

Is there any rule of thumb regarding heat dissipation of waterproof
diecast boxes (with no venting) when used as a transistor heatsink?

Some further thoughts:

The dissipation will depend on which way up you use the box and what it
is standing on. If it is hemmed-in by other equipment or might be
accidentally covered with a blanket, you need to alow for much-reduced
dissipation. If there is any sort of risk like that, incorporate a
thermal swich.

The colour of the box makes a difference, but not in the way generally
supposed. Painting it black will make almost no difference to the heat
dissipation because it is not going to be very much hotter than its
surroundings, so the net radiant heat loss will be trivial. However, if
you are going to use it outdoors in sunshine, painting it black will
make it hotter as it will gain heat from the sun (which is at a much
higher temperature).

Leaving it with a natural 'silvery' finish will be a slight improvement,
but this may not be very reflective to infra-red, so it will still heat
up (it gets worse as it becomes old and grey). Some sort of sunshade is
the simplest answer.

 

[TheGlimmerMan]

Some further thoughts:

The dissipation will depend on which way up you use the box and what it
is standing on. If it is hemmed-in by other equipment or might be
accidentally covered with a blanket, you need to alow for much-reduced
dissipation. If there is any sort of risk like that, incorporate a
thermal swich.

The colour of the box makes a difference, but not in the way generally
supposed. Painting it black will make almost no difference to the heat
dissipation because it is not going to be very much hotter than its
surroundings, so the net radiant heat loss will be trivial. However, if
you are going to use it outdoors in sunshine, painting it black will
make it hotter as it will gain heat from the sun (which is at a much
higher temperature).

Leaving it with a natural 'silvery' finish will be a slight improvement,
but this may not be very reflective to infra-red, so it will still heat
up (it gets worse as it becomes old and grey). Some sort of sunshade is
the simplest answer.

Jeez guys. This is simple. You make an Aluminum block INSIDE the box
that is BIGGER than the tab on the transistor (like double or triple
wide). You use thermal epoxy to attach that "heat spreader" to the box
permanently.
You attach the heat source (transistor) to it with ideal means as well.
Then, you have to put a heat SINK on the outside of the box to grab the
heat that the heat spreader has passed through the box's thin wall now
over a greater area due to the heat spreader inside the box. This is the
key to success here. The heat spreader. Thin wall boxes suck at
delivering heat conduction flows well. If it is custom, you can have the
spreader cast in, removing yet one more lossy interface from the system.

The external heat sink can be a finned device meant to remove heat by
way of air currents (convection). OR it can be a conduction cooled
paradigm, where you only need to insure that a good thermal conductor is
against the side of the box that has the heat source. Of course, the
whole box will end up with a higher runtime temp, but the key here is
conduction of the main heat source to the outside of the box.

The footprint of the transistor tab alone is not enough because there
are multiple interfaces to lose conduction through, and the tab just
isn't enough. The heat spreader makes GOOD conductive (thermal)coupling
with the transistor tab, and SPREADS that heat out, so that the
attachment to the can is greater, and therefore the efficiency of any
external heat sink is greater.

No math needed here. It is simple mechanics. Big solder tips heat
faster than small solder tips. It is about thermal mass, and how you
present that to where you want the heat to go.

If you want to remove heat from a source inside the thin walled can,
you need to spread that heat out INSIDE the can so that you can more
effectively couple it to the outside of the can, so that you can then
carry the heat away by convective sink attachment or conductive cooling
sink attachment mass which you mount it against..

 

[Adrian Tuddenham]

Jeez guys. This is simple. You make an Aluminum block INSIDE the box
that is BIGGER than the tab on the transistor (like double or triple
wide). You use thermal epoxy to attach that "heat spreader" to the box
permanently.
You attach the heat source (transistor) to it with ideal means as well.
Then, you have to put a heat SINK on the outside of the box to grab the
heat that the heat spreader has passed through the box's thin wall now
over a greater area due to the heat spreader inside the box. This is the
key to success here. The heat spreader. Thin wall boxes suck at
delivering heat conduction flows well. If it is custom, you can have the
spreader cast in, removing yet one more lossy interface from the system.

The external heat sink can be a finned device meant to remove heat by
way of air currents (convection). OR it can be a conduction cooled
paradigm, where you only need to insure that a good thermal conductor is
against the side of the box that has the heat source. Of course, the
whole box will end up with a higher runtime temp, but the key here is
conduction of the main heat source to the outside of the box.

The footprint of the transistor tab alone is not enough because there
are multiple interfaces to lose conduction through, and the tab just
isn't enough. The heat spreader makes GOOD conductive (thermal)coupling
with the transistor tab, and SPREADS that heat out, so that the
attachment to the can is greater, and therefore the efficiency of any
external heat sink is greater.

No math needed here. It is simple mechanics. Big solder tips heat
faster than small solder tips. It is about thermal mass, and how you
present that to where you want the heat to go.

If you want to remove heat from a source inside the thin walled can,
you need to spread that heat out INSIDE the can so that you can more
effectively couple it to the outside of the can, so that you can then
carry the heat away by convective sink attachment or conductive cooling
sink attachment mass which you mount it against..

I had assumed the OP was interested in getting rid of the heat once it
was in the box walls. With portable equipment which has to be chucked
in with other kit, finned heat sinks are a bit of a liability, so they
are best avoided if the heat problem is not too great. They can be
protected by a perforated cover, but now the equipment is no longer a
simple box and may be inconveniently unwieldy.

You will notice that my 60W amplifier had multiple output transistors.
That was not just to cope with the high currents, which could have been
handled by fewer, bigger devices; it was to spread out the heat
generation and to have more parallel paths in contact with the box
walls. (It also made the output transformer design easier, for reasons
which are irrelevant to the present thread.)

I did mention that the devices need to be spread about, so as to
transfer the heat to as large an area of the box as possible without
their hot areas overlapping. That is another way of heat spreading
without having to make a custom-built spreader or a special box.

I did use your heat-spreader idea on an earlier design of 30W amplifier
with just a pair of big output devices; in fact I put heavy aluminium
bridge pieces across the corners so as to transfer the heat to two walls
instead of one. It worked quite well, but the 60W design without a
spreader was perfectly adequate and I doubt if a spreader would have
helped it to get rid of enough extra heat to allow me to uprate it
significantly.

 

[Don Klipstein]

In <1juh181.1yjmk1p1g9jznkN%[email protected]>,
Adrian Tuddenham wrote (edited for space):

<SNIP to the point on the color of the box)

The colour of the box makes a difference, but not in the way generally
supposed. Painting it black will make almost no difference to the heat
dissipation because it is not going to be very much hotter than its
surroundings, so the net radiant heat loss will be trivial. However, if
you are going to use it outdoors in sunshine, painting it black will
make it hotter as it will gain heat from the sun (which is at a much
higher temperature).

Leaving it with a natural 'silvery' finish will be a slight improvement,
but this may not be very reflective to infra-red, so it will still heat
up (it gets worse as it becomes old and grey). Some sort of sunshade is
the simplest answer.

One factor for heatsinks, especially without good convecting fins or
forced air cooling, is how well they radiate thermal infrared.

How good a thermal radiator of a particular surface is, of a particular
wavelength, is same as how well that surface absorbs the same wavelength.
Bare metal tends to reflect thermal IR well, so bare metal is a poor
emitter of thermal IR. This is why aluminum heatsinks are often anodized,
especially smaller ones that have low anticipation of good air flow.

Oxidized metal tends to radiate thermal IR better than bare metal does.
For that matter, heatsinks without fins or forced air cooling tend to get
rid of heat better if they are painted or covered by a layer of masking
tape - despite the thermal resistance of such a coating.

One more thing - non contact thermometers don't read bare metal surfaces
well. Adding a masking tape covering tor a layer of spray paint to a bare
metal heatsink tends to result in it reading hotter than before, while it
is actually cooler than before. You can try treating a small area of a
heatsink to get a good reading. After that, try treating the whole
heatsink to see the effect on the accurately-reading area.

 

[TheGlimmerMan]

I had assumed the OP was interested in getting rid of the heat once it
was in the box walls.

That doesn't matter either. The thickness (thinness) of those boxes
means that they are not going to pass the heat correctly along the area
needed to sink the heat in a timely manner.

Even if one gets the whole box temp down to say 60C, the source
transistor inside will probably be frying due to the poor interface to
the actual heat sink mass.

Since the box wall is then, the heat needs to be spread INTERNALLY
first, then the thermal current flow through the box wall will be fast
enough to cool the device well enough to keep it from running away.

Trying to spread the heat of a 1cm square source OUTSIDE the box will
not work. Doing so inside the box allow the sinking done outside the box
to have a high thermal conduction efficacy.

He IS trying to get rid of it there. But getting the heat there
(outside) efficiently is also important.

 

[Adrian Tuddenham]

In <1juh181.1yjmk1p1g9jznkN%[email protected]>,
Adrian Tuddenham wrote (edited for space):

<SNIP to the point on the color of the box)


One factor for heatsinks, especially without good convecting fins or
forced air cooling, is how well they radiate thermal infrared.

How good a thermal radiator of a particular surface is, of a particular
wavelength, is same as how well that surface absorbs the same wavelength.

That is the critical factor. If the'radiator' is not particularly hot
or some of the surrounding surfaces are even hotter, the net result
could easily be a heat gain rather than a loss. Black heatsinks may be
all right in some circumstances, but outdoors on a sunny day (which is
where you might be using your waterproof die-cast box) they could be a
real liability.

 

[Don Klipstein]

That is the critical factor. If the'radiator' is not particularly hot
or some of the surrounding surfaces are even hotter, the net result
could easily be a heat gain rather than a loss. Black heatsinks may be
all right in some circumstances, but outdoors on a sunny day (which is
where you might be using your waterproof die-cast box) they could be a
real liability.

In that case, use white spray paint. That absorbs little of the solar
radiation that makes it through Earth's atmosphere, but it is a good
thermal IR emitter.

 

[Adrian Tuddenham]

Kapton tape is almost black in the thermal IR. So is black whiteboard
marker.

....but for outdoor use, you don't want it IR black, you want it IR
reflective, otherwise it will heat up.

 

[Don Klipstein]

...but for outdoor use, you don't want it IR black, you want it IR
reflective, otherwise it will heat up.

Actually, one would want this to be reflective of solar radiation
(nearly all getting through the atmosphere is .3 to 3 micrometers).

And preferably "black" in "low temperature thermal IR" - mostly
wavelengths 4 to 40 micrometers.

White spray paint, anyone?

 

[TheGlimmerMan]

...but for outdoor use, you don't want it IR black, you want it IR
reflective, otherwise it will heat up.

It will also then REFLECT the heat from within BACK in.

 

[TheGlimmerMan]

Actually, one would want this to be reflective of solar radiation
(nearly all getting through the atmosphere is .3 to 3 micrometers).

And preferably "black" in "low temperature thermal IR" - mostly
wavelengths 4 to 40 micrometers.

White spray paint, anyone?

ANY COATING of paint will REDUCE the emissivity of the overall device.

Even if the paint is highly emissive on its surface after drying, the
paint SLOWS the migration of heat from the box though its thickness ever
so slightly.

The best emissivity you can get is to grit blast the box exterior so
that it gains a matte finish. This is more emissive that the smooth
finish, and does not act like a blanket, like paint does.

Better still is to ATTACH the heat source side of the box to a THICK,
"cold plate", which is a heat sink that relies on mass not fins.

 

[TheGlimmerMan]

Paint on metal generally increases emissivity, often by a lot.

Absolutely not. It is 100% dependent on the paint.

For you to make that blanket remark proves that you only have a cursory
grasp of it, if even that. By a lot.

 

[TheGlimmerMan]

Paint on metal generally increases emissivity, often by a lot.


"Slightly" is correct. The thermal resistance of a coat of paint is
miniscule compared to the thermal resistance of the first mm of air.

A blanket is a blanket. A layer is a layer. The heat will migrate OUT
of the box SLOWER if it is painted.

The only difference is if the right IR paint is put onto it AT the
right coating thickness. Then, and ONLY then, will the painted box
dissipate its source faster than the unpainted box.

And as has been stated, that is true only if the box is not sitting out
in direct sunlight, which MAY be able to keep it from giving up its heat
as fast due to infusing more additional heat into the box than the box is
able to radiate out.

So, there are SEVERAL factors that determine whether or not painting
will slow or speed heat dispersion from a surface, and it has nothing to
do with "the first mm of air".