The Role of Convection in PC Cooling

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada

The Role of Convection in PC Cooling

Post by MikeC » Tue Dec 20, 2005 10:25 am

It's true that heat rises, and when a thermal cooling design is carefully designed to take advantage of it, convection can be enough to move lots of air. I used to think this was very important in a PC. But then, I started playing around with the gear --- and have kept playing for 5 years. I've also studied the results of many experiments by SPCR forum members with chimney convection schemes for PC components. (Try a search for chimney).

Now, this is what I know to be true:

Any significant movement of air using convection alone requires high temperature differentials, a large amount of heat, and a real chimney. None of these conditions prevail inside any PC.

A single slow spinning fan moves enough air to easily overcome rising hot air due to convection in a PC case. Period.

In other words, for all practical intents and purpose, you can forget about air convection in a PC. It can be almost totally disregarded if you use a single fan at low speed in the right spot.
Mike Chin, SPCR Editor/Publisher
Support SPCR by buying your gear through this link: Amazon

Posts: 30
Joined: Fri Oct 07, 2005 11:47 am

Post by dorion » Tue Dec 20, 2005 11:35 am

So how would this play in the case of an external radiator for a water cooling system. If I have the fins of the radiator vertical, would the temperature difference of the outside air have a considerable effect?

Site Admin
Posts: 12283
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada

Post by MikeC » Tue Dec 20, 2005 12:30 pm

Not so much the temp difference of the outside air, but rather just the temp. Of course this has an effect -- the outside temp becomes the ambient operating temp of the system, and w/the same temp rise, the final temp of the CPU will be that much lower.

You'll note that my original post refers to the role of convection inside a PC case; I should add another condition: inside a conventional PC case.

When you wick the heat outside to large radiating surfaces -- such as the huge HS found on Zalman TNN cases and other passive (fanless) systems -- then convection obviously does matter and actually works.

Still, a couple of 120mm fans running super slow might achieve the same cooling result in a more conventional case for much less money and complexity, and the acoustic difference might be nonexistent because of HDDs -- if they are louder than the fans, then they set the quiet limit, not the fans.

Going back to my original post, I was referring specifically to the idea that blowing a fan against convection is highly inefficient for cooling -- it's not really so, because the amount of air movement caused by convection in a typical PC system/case is so miniscule, other factors weigh in much more heavily.

For example, take a horizontal MB setup like our HS test platform. Take a dozen "conventional" HS, where the fan is atop the HS & can be set to blow down or up. Now monitor the CPU temp under stress for all 10 HS w/ fan blowing down AND blowing up -- at low speed.

If convection (heat rises) was a serious factor, then the fan blowing up should always show an improvement. But it will not; the result will most likely be evenly split, and the difference between up/down will be very small.

It's not because convection does not play any role; it's because the fan produces so much more airflow than convection that it doesn't create any real resistance to the downward airflow nor any real assistance to the upflow.
Mike Chin, SPCR Editor/Publisher
Support SPCR by buying your gear through this link: Amazon

Posts: 36
Joined: Fri Feb 25, 2005 10:36 pm

Post by ToasterIQ2000 » Tue Dec 20, 2005 1:43 pm

I have the old radiator out of my 88 toyota ( roughly one foot by two ) on the side of my case mounted the tall way about an inch and a half away, with room for air to enter the gap from the bottom.

It is collecting heat from an athalon xp 3000+, nForce 2 motherboard, Seagate Barracuda V 120gb, and a Quadro FX 3000 ( ~ nv FX 5950 ): roughly 55 - 140 ( idle - load ) watts make it into the radiator out of 160 - 270 watts out of the wall socket. It keeps the cpu under 45 C, the mb under 30, and the hdd and gfx under 40.

With no fans on here is enough hot air rising to move a fine blonde hair, a thread, & a heavy jeans thread ( all about six inches long ) suspended over the chimney.

A single quiet 80mm (Pabst 8???NGL; I'd have to dissasemble to check today ...) fan placed kind of sloppily without ducting or shouds in the (mostly gone) lower wall of the case to pull air out of the case and across the inch and a half gap towards the radiator lowers the water temp in the loop by about ten to twelve degrees.

The convection & the chimney effect are noticable. Aside from the most obvious rules for using a passive radiator ( Put it outside the case; make it big; make sure that air is free to apprach from the bottom and rise around it; don't let anyone dump laundry on it ... ) I haven't found any trick for accentuating the passive convection that comes at all close to the effect of a single slow queit small fan mounted kind of close & blowing in the general direction of the radiator.

Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Tue Dec 20, 2005 2:32 pm

In other words, for all practical intents and purpose, you can forget about air convection in a PC.
It's important to make the distinction between passive or "natural" convection and forced convection (which is the major heat transport mechanism used in conventional ATX-format PC's). Generally passive air convection is only practical when evacuating small amounts of heat from well-ventilated cases, and as heat output rises forced convection (whether using water or air as the heat-transport medium) becomes the primary and inevitable choice.

Perhaps an exceptionally well-designed heatsink with a gigantic surface area (such as in the Hush PC's, Mappit A4F and Zalman TNN cases) would allow natural convection to compete with forced convection alternatives, but there is no question which is the most cost-effective and elegant solution for the vast majority of users.

The exploration of passive convection solutions can shed new light on how to make forced convection setups more effective/efficient, and so in that sense they are the fundamental interest of any silence-oriented PC fanatic.

Posts: 24
Joined: Sun Jul 11, 2004 4:36 pm

Post by bchung » Tue Dec 27, 2005 11:43 am

While we are on the subject of silent cooling practises that are against what is normally done...

I have a heavy medium tower: 80 lbs after insulation. 3/8" foam on 1/4" neoprene on 1/64" aluminum on 1/8" "chickenwire" neoprene on a 0.7mm steel case.

I initially had the airflow as is normally recommended: from bottom front HDs to out the PSU in the rear. This was ok, but during the summer the computer constantly overheated in marathon gaming sessions. Plus, it was too noisy for my tastes.

While I did make changes to the rest of the system, the most important change was that I reversed the direction of airflow in the psu and flipped the psu around. The psu fan is now a squirrel cage (radioshack, potentiometer controlled) that pulls air into the case and blows down onto the ram.

Before swapping psu fans/direction, my greatest source of sound was from the psu+fan (swapped in a panaflo @7V). Now, no/little sound escapes from the psu and my system temps have only gone up 5-6C. Added benefit is that my VGA card is 9C lower than it was before.

Sometimes it is better to turn the psu around to direct airflow into the case to minimize sound escaping.

Post Reply