Which is the most important factor in cooling?

[AmigaMance]

 Is it the airflow which the fan provides or is it the heat-sink?
 Yesterday, the fans which i have attach in the heatsinks of my BPPC and BVision, stopped spinning for 10-15 minutes, due to a defective wire. Both card were starting to crash (specially the BVision which after i become aware of the problem, i touched it and it was the hottest) and it wasn't a hot day at all. About 22c.
 So, is the airflow the most important factor? Oh, and was i lucky that nothing got fried? :-D

[Oliver]

Hi,

You really need both.  Aluminium is used to absorb heat from the chip, ususally with a thermal transfer paste between the two, just to help connect the two thermal bodies.  The heat sink transfers energy into the surrounding environment via both radiation and conduction.

If there is no forced movement in the surrounding air (or water, etc), then before too long the air gets warm.  Quasi equilibrium is reached when there is a stable thermal gradiant in the air surrounding the heat sink.  Heat is then primarily transfered via diffusion/convection through the air, as well as the afore mentioned radiation.  Note, that although the system is in a reasonably stable thermal state, it is an open system, and there is still net heat transfer through the air.  Obviously, if the air is only being moved by the thermally generated diffusion/convection processes, it trasnfers much less heat than if cool air is forced over the heat sink.  If you move air with a fan, the heat is dispersed throughout your study room, instead of being so localised around your chip.

Heat sinks are designed to have good thermal conductivity, large surface areas, and often surface treated to improve radiation in the peak bandwidth for the expected temperatures.

Without either the heat sink or the forced air flow, there is much less heat transfer away from the chip.

Hope that helps.

[AmigaMance]

Hope that helps.

It does. Thanks.

[tonyvdb]

You can never have enough fans to move the air though the case. Like was said above if you do not remove the warm/hot air away from the devices then this will cause dammage.

I have two towers in my Amiga system one for the A4000 and the other for my SCSI drives and TBC's. I have 4 fans in the A4000 tower and the other has 6 and both after running for an hour both are removing noticably warm air.

[Cymric]

If there is no forced movement in the surrounding air (or water, etc), then before too long the air gets warm.  Quasi equilibrium is reached when there is a stable thermal gradiant in the air surrounding the heat sink.  Heat is then primarily transfered via diffusion/convection through the air, as well as the afore mentioned radiation.

Not quite, although I doubt you left the following out because you didn't know about it :-). If there is no forced convection (i.e., a fan blowing air across the fins of a cooling element) then as the air heats up, you get natural convection because warm air is lighter than cold air and thus begins to move upwards. This process only works in the presence of a gravitational field; in other words, natural convection doesn't arise in space. You get very weird effects, as illustrated by this picture of a burning candle in a low-g environment:


Natural convection is a very inefficient way of generating an air flow, and thus is of little practical importance in cooling of earth-based applicances. At room temperature, a rule of thumb says that the magnitude of heat dissipation by natural convection is approximately equal to that of radiation, but as the latter is proportional to the temperature of the hot object to the fourth power (!), radiation quickly becomes the more important mechanism.

Neither can hold a (figurative) candle to forced convection, however, unless the temperature is really high, but by that time you no longer have working computer.

[Oliver]

@Cymric

Sounds right.  I haven't considered fanless heat transfer for some time.  I underestimated the relative effect of the radiation versus conduction in a fanless heat sink.  I generally just pipe heat away.  Like you said, radiation is only a minor factor at moderate temperatures, compared to what can be achieved with conduction into a forced air stream.

Thanks also for the photo.  Can you tell us what gives rise to the apparent corona effect?  Is that the sparser/hotter air convecting outwards as a front?  Is the air on that corona still undergoing combustion at that point?

[Doppie1200]

I allways thought it could be simplyfied too surface to air resistance taking the difference in ambient and surface temprature into account.

Airflow help keeping the ambient temprature lower. Otherwise the air directly around your heatsink will warm up too much reducing the heatflow. Then there is resistance, hence larger surfaces provide less surface to air resistance. Here the material of your heatsink is important too. And added to the surface to air resistance is the object to heatsink resistance. Here the thermal paste comes in.
You want to lead the heat away from the object to somewhere else, usually the air. The heat pumping can be accelerated by means of liquid. But as a true airhead I don't like that.
Like all Beetle drivers say: 'Wasser ist um eier kochen' ;-)