Sin22
Master Member
- Joined
- Nov 8, 2001
- Messages
- 4,229
- Reaction score
- 11
Aircooling is going to be the main source of thermal cooling for many items in the world today and in the future. That is the fact of the matter even though there are many strong pro-ponents of alternate forms of cooling like water & phase-change. Its just too hard to get away from the ease at which aircooling is capable of keeping a hot item at a respectable temperature level.
Now what must be understood about aircooling with respect to computers and CPUs is that heatsinks can come in many different shapes and sizes but physical realities must be imposed on them and thus optimised to obtain the greatest amount of heat transfer from a particular shape or design.
It is already common knowledge that copper has a thermal conductivity rating which is significantly above that of aluminum (400W/mk vs. 220W/mk). Now take that into account when the design of baseplates for a heatsink is considered. For a copper heatsink, the rise in temperature of the heatsource can be calculated to be = 2.5degC x sqrt(baseplate thickness in mm) and for a aluminum heatsink it changes to 4degC. That means that the thicker the base of the heatsink, the higher the CPU core temp will increase with respect to each mm increase in thickness. However, there is a trade-off, the thicker you have the base, the more heat load that can be spread over an area and due to this larger area, more fins can be attached. This is where the real cooling part comes into play for a heatsink...the fins.
Taking what was said once in a thread on radiator & heatercore design, lets look at things from a simplified thermodynamics point of view. The heat is conducted up through the base and then to the fins and then taken away by air. But if its just passive air, not much conduction is going to be happening correct ? Each air molecule has to hit the metal of the heatsink, get heat transfered to it, and then moved out of the way so the next air molecule can hit it again and absorb more energy in the form of heat. Remember that the air molecule has its own starting form of energy and as such is only receiving energy in the form of heat transfer from the molecules in the fins to the molecules in the air, and as such you will never go below ambient temperatures.
Now combining what is known already about baseplates and fins, it would mean that we would want a thick baseplate with a relatively large planar surface area connected to a large amount of fins that rise up to about 6~8cm above the baseplate which would create a large surface area for heat transfer to occur.
Now, we need to look at the fans themselves, to blow in or suck out.
Lets look at how 90% of fans are designed. Most fans use a pressure differential to create movement of air. However, on the intake side, it sucks in the air from all directions, it doesnt bother where the air is, it just sucks it in. Then the air hits the obstruction of the hub and the stakes holding the hub together and then enters the fan blades, gets spun around and shot out at the other end. However, there is a dead zone. There is an area directly beneath the hub where no air moves, its a stagnation zone really where there is no moving air. The further away from this area, the more air gets thrown into that deadzone "path".
Put the two together, the fan and the heatsink. If you were to put the fan on so that its sucking air through the heatsink, people may be thinking, well its sucking air in from all over the place. Correct, but this air isnt moving very fast and there isnt a very large volume of it, it will cool the heatsink down mainly because it covers a large area of metal. But it wont cool it down that much because the volume and velocity of the air just isnt there. Next, you turn the fan around and blow on the heatsink. Yes there is the deadzone and its a very visible deadzone. Take out any heatsink which is dusty and dirty and you'll see a clean clear zone on the heatsink directly under the fan hub. However, it cools much better because of the higher volume of air being pushed through the heatsink. More air is getting through and conducting the heat away from the metal.
Now it is known that the deadzone gets reduced to about 60% of its original size if the measurement is taken at a distance 3x the hub diameter and is all but gone by 5x the hub diammeter. This is one reason why windtunnels exhibit such good performance and are used in OEM systems.
So now, try and combine the two, we require a heatsink that has the above mentioned attributes, and a fan that can be placed at a location far enough from the heatsink fins so that most of the air impacts all the fins. Unfortunately, this rarely happens. With heatsink designs like the SLK800 & earlier MCX462+ series, a high CFM & high pressure fan was required to force air through the heatsink and impact all the fins. Remember, those fins are restrictions and will cause a pressure buildup on the outlet of the fan, as such a stronger fan is required. A low CFM low pressure fan doesnt work as well becasue it doesnt have the required amount of "muscle" to force the air through the fins.
So now we have another issue. Designing a heatsink that is both excellant at high CFM and high pressure fans and also with low cfm low pressure fans. Thats where heatsinks like the MCX462-V come along with enough surface area for heat distribution and spacing between each "fin" to allow relatively easy air passage.
Some of the concepts and information in here were obtained from a post made by Cathar on OCAU, others were obtained from what I know of heat transfer in heatsinks and the lot. It wont be 100% correct, but it does give a much better idea of how heatsinks are designed and aircooling works. Bear all of this in mind the next time you look & buy a heatsink
To add, take a look at this great website that shows the actual flow visualisation of a axially rotating fan and the resulting movement of the air due to the fan.
Very impressive movies and helps even further in explaining what I was talking about above.
Flow visualisation
Now what must be understood about aircooling with respect to computers and CPUs is that heatsinks can come in many different shapes and sizes but physical realities must be imposed on them and thus optimised to obtain the greatest amount of heat transfer from a particular shape or design.
It is already common knowledge that copper has a thermal conductivity rating which is significantly above that of aluminum (400W/mk vs. 220W/mk). Now take that into account when the design of baseplates for a heatsink is considered. For a copper heatsink, the rise in temperature of the heatsource can be calculated to be = 2.5degC x sqrt(baseplate thickness in mm) and for a aluminum heatsink it changes to 4degC. That means that the thicker the base of the heatsink, the higher the CPU core temp will increase with respect to each mm increase in thickness. However, there is a trade-off, the thicker you have the base, the more heat load that can be spread over an area and due to this larger area, more fins can be attached. This is where the real cooling part comes into play for a heatsink...the fins.
Taking what was said once in a thread on radiator & heatercore design, lets look at things from a simplified thermodynamics point of view. The heat is conducted up through the base and then to the fins and then taken away by air. But if its just passive air, not much conduction is going to be happening correct ? Each air molecule has to hit the metal of the heatsink, get heat transfered to it, and then moved out of the way so the next air molecule can hit it again and absorb more energy in the form of heat. Remember that the air molecule has its own starting form of energy and as such is only receiving energy in the form of heat transfer from the molecules in the fins to the molecules in the air, and as such you will never go below ambient temperatures.
Now combining what is known already about baseplates and fins, it would mean that we would want a thick baseplate with a relatively large planar surface area connected to a large amount of fins that rise up to about 6~8cm above the baseplate which would create a large surface area for heat transfer to occur.
Now, we need to look at the fans themselves, to blow in or suck out.
Lets look at how 90% of fans are designed. Most fans use a pressure differential to create movement of air. However, on the intake side, it sucks in the air from all directions, it doesnt bother where the air is, it just sucks it in. Then the air hits the obstruction of the hub and the stakes holding the hub together and then enters the fan blades, gets spun around and shot out at the other end. However, there is a dead zone. There is an area directly beneath the hub where no air moves, its a stagnation zone really where there is no moving air. The further away from this area, the more air gets thrown into that deadzone "path".
Put the two together, the fan and the heatsink. If you were to put the fan on so that its sucking air through the heatsink, people may be thinking, well its sucking air in from all over the place. Correct, but this air isnt moving very fast and there isnt a very large volume of it, it will cool the heatsink down mainly because it covers a large area of metal. But it wont cool it down that much because the volume and velocity of the air just isnt there. Next, you turn the fan around and blow on the heatsink. Yes there is the deadzone and its a very visible deadzone. Take out any heatsink which is dusty and dirty and you'll see a clean clear zone on the heatsink directly under the fan hub. However, it cools much better because of the higher volume of air being pushed through the heatsink. More air is getting through and conducting the heat away from the metal.
Now it is known that the deadzone gets reduced to about 60% of its original size if the measurement is taken at a distance 3x the hub diameter and is all but gone by 5x the hub diammeter. This is one reason why windtunnels exhibit such good performance and are used in OEM systems.
So now, try and combine the two, we require a heatsink that has the above mentioned attributes, and a fan that can be placed at a location far enough from the heatsink fins so that most of the air impacts all the fins. Unfortunately, this rarely happens. With heatsink designs like the SLK800 & earlier MCX462+ series, a high CFM & high pressure fan was required to force air through the heatsink and impact all the fins. Remember, those fins are restrictions and will cause a pressure buildup on the outlet of the fan, as such a stronger fan is required. A low CFM low pressure fan doesnt work as well becasue it doesnt have the required amount of "muscle" to force the air through the fins.
So now we have another issue. Designing a heatsink that is both excellant at high CFM and high pressure fans and also with low cfm low pressure fans. Thats where heatsinks like the MCX462-V come along with enough surface area for heat distribution and spacing between each "fin" to allow relatively easy air passage.
Some of the concepts and information in here were obtained from a post made by Cathar on OCAU, others were obtained from what I know of heat transfer in heatsinks and the lot. It wont be 100% correct, but it does give a much better idea of how heatsinks are designed and aircooling works. Bear all of this in mind the next time you look & buy a heatsink
To add, take a look at this great website that shows the actual flow visualisation of a axially rotating fan and the resulting movement of the air due to the fan.
Very impressive movies and helps even further in explaining what I was talking about above.
Flow visualisation
Last edited:
