How much airflow do we really need?

Joined
4 Aug 2007
Posts
22,395
Location
Wilds of suffolk
I know its an open ended question and all :) and its a bit of a random thread so random answers/thoughts only please ;)

So lets break it down, three main heat causing areas as far as I can really deduce (in order)
1) Graphics card if its high spec
2) Processor, could be highest depending on overclocking and what graphics card is in use
3) Power supply, but more efficient = lower heat (the heat is the inefficiency)

Rest of components of course generate some, but RAM, controllers etc are generally not many watts from what I can tell, maybe HDDs generate a fair bit if used heavily for prolonged periods but short bursts I would think would seep the heat over time so not have a dramatic effect on the background heat

So power supply is easy, they are semi self cooling as they exhaust hot air themselves although they do appear to radiate some heat its unclear if this escapes into the case really or is actually pulled back via airflow through the power supply.

Processor well these are fairly well understood, we know what sort of airflow is required to keep one cool enough. I am assuming they are radiating a lot of heat into the case even if you have them pointed towards an exhaust since airflow will be turbulent?

Graphics card, this is the one that generally I am not 100% sure about, its got a small fan (relative to processor) for the heat it produces, so either its very efficient at venting that heat or its going to be radiating a fair amount into the case?

So I am thinking are we waaay over the top on overall airflow being pumped through cases to cover weakspots in the overall design?

Im looking at the expedit pc thread and wondering if a single large fan (such as the 220mm one on here) could be used to pump enough airflow into the "case" so that the overall airflow would be enough to eliminate any hotspots etc and effectively give all components ambient air. It flows 90cfm of air so thats a damn lot surely? I am starting to think that pc cases are generally a pretty poor design for cooling but that they are accepted in form and hence cant be really changed.

So if you were designing from scratch what would yours look like?

Im thinking something like an 18" diamater tube, large 18" fan at each end in push pull config and everything else inside with most heat critical components closer to in and less critical closer to out.

So where do I get that 18" tube? :D
 
I'm thinking that a tube would still give you dead/hot spots if the airflow is in one direction (with the MB parallel to the airflow) which is what I envisage from your description. Perhaps intake at both ends, and exhaust around the components?

With the MB perpendicular to the airflow you effectively have an Antec Skeleton, although the air disperses with a Skeleton, as opposed to being channelled, as per a tube.
 
Last edited:
As for the question of air - cooler components probably don't need much airflow to keep the temps under control. Don't have any fans in my pc (link in sig) and the motherboard mosfets, hard-drive, RAM, even graphics card VRMs run okay heavilty overclocked. VRMs run a bit toasty though. Plenty of people with bench tables have no fans over components or eg when sticking LN pots onto the CPU and GPU. They're all open systems with passive airflow though - would be a different story with a sealed box and no airflow.
 
I use a large midi tower case (LL PC-S80B) with just a single fan as 7v intake.
Only two holes in the case (all others blocked)
Fan intake - rad outlet (psu does not 'vent' to the outside)
gpu, cpu and NB are watercooled (but no fans on rad)
PSU is fanless (but has external heatsink fins that stick out the back of case)
Mosfet / HHD / ram have finned heatinks
Air flow is very low (xbitlab measured the fan at 18cfm at 7v)

E8200es oc'ed and undervolted, (xspc edge)
3780 (ek full block)
P45 (ek/silentpipe hybrid)
D5 with EK top setting 3

Temps are as stock intel/nvidia cooler running stock speeds (bit more in summer, less in winter)

Zalman TNN / mcubed cases also only use one fan (psu)
As do most passive watercooling builds (cape corra/reserator/konvekt-o-matic)

Hotter the hardware higher the minimum airflow.
 
Last edited:
Silverstone Fortress FT02B modded to fit 3 fans across the top too. Fans to move air + natural thermodynamics = win. Though to be honest I am not sure how much trouble it would be to add three fans on the top.

Could always go for the tube you mentioned with little 120mm fans to pull air away from the hottest spots or direct air to dead spots.
 
With air cooling you are never going to be able to get sub ambient temps, the best you can hope for is inside your case / tube what have you is to have exactly the same inside as out.

In respect to how much airflow that entirely depends on the amount of heat being generated and where from, there are some very good 230mm fans and indeed 120mm fans on the market that will shift 70+ cfm at sub 15dba which in the real world is more than enough.
 
I know its an open ended question and all :) and its a bit of a random thread so random answers/thoughts only please ;)

So lets break it down, three main heat causing areas as far as I can really deduce (in order)
1) Graphics card if its high spec
2) Processor, could be highest depending on overclocking and what graphics card is in use
3) Power supply, but more efficient = lower heat (the heat is the inefficiency)

Rest of components of course generate some, but RAM, controllers etc are generally not many watts from what I can tell, maybe HDDs generate a fair bit if used heavily for prolonged periods but short bursts I would think would seep the heat over time so not have a dramatic effect on the background heat

So power supply is easy, they are semi self cooling as they exhaust hot air themselves although they do appear to radiate some heat its unclear if this escapes into the case really or is actually pulled back via airflow through the power supply.

Processor well these are fairly well understood, we know what sort of airflow is required to keep one cool enough. I am assuming they are radiating a lot of heat into the case even if you have them pointed towards an exhaust since airflow will be turbulent?

Graphics card, this is the one that generally I am not 100% sure about, its got a small fan (relative to processor) for the heat it produces, so either its very efficient at venting that heat or its going to be radiating a fair amount into the case?

So I am thinking are we waaay over the top on overall airflow being pumped through cases to cover weakspots in the overall design?

Im looking at the expedit pc thread and wondering if a single large fan (such as the 220mm one on here) could be used to pump enough airflow into the "case" so that the overall airflow would be enough to eliminate any hotspots etc and effectively give all components ambient air. It flows 90cfm of air so thats a damn lot surely? I am starting to think that pc cases are generally a pretty poor design for cooling but that they are accepted in form and hence cant be really changed.

So if you were designing from scratch what would yours look like?

Im thinking something like an 18" diamater tube, large 18" fan at each end in push pull config and everything else inside with most heat critical components closer to in and less critical closer to out.

So where do I get that 18" tube? :D

hate to mention ****** but their quite good at that
 
As for the question of air - cooler components probably don't need much airflow to keep the temps under control. Don't have any fans in my pc (link in sig) and the motherboard mosfets, hard-drive, RAM, even graphics card VRMs run okay heavilty overclocked. VRMs run a bit toasty though. Plenty of people with bench tables have no fans over components or eg when sticking LN pots onto the CPU and GPU. They're all open systems with passive airflow though - would be a different story with a sealed box and no airflow.

i dont get your point...dont you think thats a bit out of context
 
Back
Top Bottom