General:
Ok, we all know that the heatsink conducts heat from the cpu die to the fins/pins.
Conduction increases with:
1. increasing conduction coefficient
2. increasing cross-sectional area,
3. decreasing material thickness, and
4. increasing temperature differential.
This last factor is why peltiers are used, they increase the temperature differential, essentially creating a steeper "energy slope"- moving energy from the die to the heatsink faster. This last factor also comes into play in regard to airflow directions and how suck and blow are not equivalent.
Scenario 1: Fan-->Heatsink-->CPU.
There is a temperature differential across your heatsink's fins with the point furthest from the cpu being the coolest. As cool air comes in contact with the warmer fin, it takes up heat energy, and becomes slightly warmer.
This slightly warmer air now moves closer to the cpu and encounters a warmer point on the heatsink. Since there is still a temperature differential between the two surfaces, heat moves from the warmer fin to the relatively cooler air.
So, the air is gradually warmed over the course of it's travel over the fins, removing heat all along its path, increasing the temperature differential between the cpu and the heatsink, and improving heat flow from cpu to sink as well as removing heat from the heatsink.
Cool air--------------warm air--------warmer air---->
Warm fin-------------warmer fin---------hot fin-------CPU
Scenario 2: <--Fan<--Heatsink<--CPU
The cool air is now being pulled past the cpu, since there is a larger temperature differential heat will move more quickly from the heatsink base into the air rushing past, however, since the air is encountering the hottest surface first and warms up quickly, it is no longer as efficient at picking up heat from the fin as it travels down it length. The hottest air is coming in contact with the coolest part of the sink. Overall the efficiency of fin-to-air convection is reduced by the lessened temperature differential.
<--warm air-----------warm air----------cool air------
---Warm fin-------------warmer fin---------hot fin-------CPU
On the upside, you are creating a larger temperature differential between the cpu and the heatsink at the hottest part, it's base, thereby moving heat quickly from the base to the heat sink, but then not moving heat from the heatsink to the air as quickly.
Conclusion:
In the end, it's sort of a wash either way, but at steady states, the first scenario should prove to be slightly more efficient overall, especially in applications with large flat fins. Not surprisingly, the blood vessels in our extremities are set up in the same manner as scenario so as to preserve (rather than dissipate) our core temps in cold environs.
One exception would be heatsinks such as the Alpha. These heatsinks are very good at getting heat from the CPU to the heatsink, but not as effective as moving the heat up the small pin-style radiators, so they benefit from reversed intake. The company uses a shroud to make sure air is pulled past the hottest part of the sink first to increase this effect.
So, in summary:
Flat, blade-style radiator fins = Blow
round or square pin-style radiator fins = Suck
Not exactly earthshattering, but it does serve to illustrate basic concepts of thermodynamics and that which way a fan points does make a difference in the physical execution of conduction and convection.
Edited by Mjolnir - 10/7/08 at 9:59pm
Ok, we all know that the heatsink conducts heat from the cpu die to the fins/pins.
Conduction increases with:
1. increasing conduction coefficient
2. increasing cross-sectional area,
3. decreasing material thickness, and
4. increasing temperature differential.
This last factor is why peltiers are used, they increase the temperature differential, essentially creating a steeper "energy slope"- moving energy from the die to the heatsink faster. This last factor also comes into play in regard to airflow directions and how suck and blow are not equivalent.
Scenario 1: Fan-->Heatsink-->CPU.
There is a temperature differential across your heatsink's fins with the point furthest from the cpu being the coolest. As cool air comes in contact with the warmer fin, it takes up heat energy, and becomes slightly warmer.
This slightly warmer air now moves closer to the cpu and encounters a warmer point on the heatsink. Since there is still a temperature differential between the two surfaces, heat moves from the warmer fin to the relatively cooler air.
So, the air is gradually warmed over the course of it's travel over the fins, removing heat all along its path, increasing the temperature differential between the cpu and the heatsink, and improving heat flow from cpu to sink as well as removing heat from the heatsink.
Cool air--------------warm air--------warmer air---->
Warm fin-------------warmer fin---------hot fin-------CPU
Scenario 2: <--Fan<--Heatsink<--CPU
The cool air is now being pulled past the cpu, since there is a larger temperature differential heat will move more quickly from the heatsink base into the air rushing past, however, since the air is encountering the hottest surface first and warms up quickly, it is no longer as efficient at picking up heat from the fin as it travels down it length. The hottest air is coming in contact with the coolest part of the sink. Overall the efficiency of fin-to-air convection is reduced by the lessened temperature differential.
<--warm air-----------warm air----------cool air------
---Warm fin-------------warmer fin---------hot fin-------CPU
On the upside, you are creating a larger temperature differential between the cpu and the heatsink at the hottest part, it's base, thereby moving heat quickly from the base to the heat sink, but then not moving heat from the heatsink to the air as quickly.
Conclusion:
In the end, it's sort of a wash either way, but at steady states, the first scenario should prove to be slightly more efficient overall, especially in applications with large flat fins. Not surprisingly, the blood vessels in our extremities are set up in the same manner as scenario so as to preserve (rather than dissipate) our core temps in cold environs.
One exception would be heatsinks such as the Alpha. These heatsinks are very good at getting heat from the CPU to the heatsink, but not as effective as moving the heat up the small pin-style radiators, so they benefit from reversed intake. The company uses a shroud to make sure air is pulled past the hottest part of the sink first to increase this effect.
So, in summary:
Flat, blade-style radiator fins = Blow
round or square pin-style radiator fins = Suck
Not exactly earthshattering, but it does serve to illustrate basic concepts of thermodynamics and that which way a fan points does make a difference in the physical execution of conduction and convection.
Edited by Mjolnir - 10/7/08 at 9:59pm
2 Thumbs up!!
