Posted: Sat Sep 22, 2007 3:09 am
Felger Carbon: "Definitely not better undervoltability. Reducing the cooling fin area is never a cooling win."
One should never say never. Whoopsie, I said it twice.
Consider a scenario when you have an unmodded Freezer 7 running at 7V (lower than can be recommended for most CPUs): airflow would probably be so low it's impossible to feel it due to airflow resistance of the heatsink. Also, with air moving slowly the exhaust air is pretty much the same temperature os the heatsink itself. This is because air can store very limited amount of heat per volume (because it's so thin).
If you use a heatsink that will have exhaust air at the same temperature as heatsink itself, the amount of transfered heat directly related to air passing through. I reality this is impossible as exhaust will be more or less cooler than fins. If we think of Freezer 7 at 7V we are relatively close to this assumption though because at 7V the airflow just drops too low to make the HSF useful.
If you were to increase surface area even further by creating a very big heatsink, say at least 10 to 20 cm long airflow path and keeping fin spacing the same as Freezer 7, the exhaust air will still be roughly the same (maybe increase just by few degrees), but the airflow resistance just got doubled, meaning airflow with same fan speed got lowered (=> reduced cooling) so more voltage is needed to be fed to the fan to compensate. After compensation (=> increased noise) you'd again have the same amount of airflow and roughly same exhaust temperature => same amount of total heat transfered... with the cost of extra noise.
Reducing cooling fin area is potential win when running HSF undervolted or passive. You can do this be increasing fin spacing or making fins shorter in dimension that's longitudal to airflow direction (reducing other dimensions would be bad).
I have strong belief based on SPCR's undervolting measurements that reducing cooling area would be beneficial for all but 12V operation (which is what this cooler most likely was optimized for). What I'm mostly worried about is whether cutting the fins using metal saw (since I don't own a dremel and it's bigger cousins are probably overkill) will damage the solder between heatpipes and fins. It may vibrate quite a bit when I do the cutting. I would cut the fins just a millimeter or two behind heatpipes and would sand the cut edge smooth so that there'd be no extra turbulence when air leaves the HSF.
Other mods I would consider for this HSF are
- blocking the sides of the heatsink so that centrifugal "force" doesn't blow air out the shortest route (this tendency is lowered by shortening the fins so it might not be necessary)
- creating a frame or even a duct for the fan to see if it increases turbulent noise and whether it increases useful airflow through the HSF. These frameless fans work OK when there's no backpressure, but considering how placing a Noctua fan (with frame) on floor with exhaust side blocked can create a noticeable amount of radial airflow from the intake side, it'll probably happen quite easily when adding backpressure to frameless fans.
Why did they make the fan push air into HS instead of sucking it? If it was sucking air out, even centrifugal airflow would contribute to airflow that passes through the HS. For sucking, frameless fans are no doubt better but for pushing air against backpressure... I don't get it.
One should never say never. Whoopsie, I said it twice.
Consider a scenario when you have an unmodded Freezer 7 running at 7V (lower than can be recommended for most CPUs): airflow would probably be so low it's impossible to feel it due to airflow resistance of the heatsink. Also, with air moving slowly the exhaust air is pretty much the same temperature os the heatsink itself. This is because air can store very limited amount of heat per volume (because it's so thin).
If you use a heatsink that will have exhaust air at the same temperature as heatsink itself, the amount of transfered heat directly related to air passing through. I reality this is impossible as exhaust will be more or less cooler than fins. If we think of Freezer 7 at 7V we are relatively close to this assumption though because at 7V the airflow just drops too low to make the HSF useful.
If you were to increase surface area even further by creating a very big heatsink, say at least 10 to 20 cm long airflow path and keeping fin spacing the same as Freezer 7, the exhaust air will still be roughly the same (maybe increase just by few degrees), but the airflow resistance just got doubled, meaning airflow with same fan speed got lowered (=> reduced cooling) so more voltage is needed to be fed to the fan to compensate. After compensation (=> increased noise) you'd again have the same amount of airflow and roughly same exhaust temperature => same amount of total heat transfered... with the cost of extra noise.
Reducing cooling fin area is potential win when running HSF undervolted or passive. You can do this be increasing fin spacing or making fins shorter in dimension that's longitudal to airflow direction (reducing other dimensions would be bad).
I have strong belief based on SPCR's undervolting measurements that reducing cooling area would be beneficial for all but 12V operation (which is what this cooler most likely was optimized for). What I'm mostly worried about is whether cutting the fins using metal saw (since I don't own a dremel and it's bigger cousins are probably overkill) will damage the solder between heatpipes and fins. It may vibrate quite a bit when I do the cutting. I would cut the fins just a millimeter or two behind heatpipes and would sand the cut edge smooth so that there'd be no extra turbulence when air leaves the HSF.
Other mods I would consider for this HSF are
- blocking the sides of the heatsink so that centrifugal "force" doesn't blow air out the shortest route (this tendency is lowered by shortening the fins so it might not be necessary)
- creating a frame or even a duct for the fan to see if it increases turbulent noise and whether it increases useful airflow through the HSF. These frameless fans work OK when there's no backpressure, but considering how placing a Noctua fan (with frame) on floor with exhaust side blocked can create a noticeable amount of radial airflow from the intake side, it'll probably happen quite easily when adding backpressure to frameless fans.
Why did they make the fan push air into HS instead of sucking it? If it was sucking air out, even centrifugal airflow would contribute to airflow that passes through the HS. For sucking, frameless fans are no doubt better but for pushing air against backpressure... I don't get it.