• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Downward blowing fans & heatsinks

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Posted by BadRobot:
1) Good enough for me, so basically the air coming of the heat sync is still plenty capable of cooling other components.

2)And the redirection of the airflow caused by the fan blowing directly at a component is negated if your case is pressurized (more air in than out)

On item 2, I'm not sure what you mean. We build up pressure inside the case with the intake fans. The CPU-Fan pulls this pressurized air into the duct box, and the exhaust fan pulls it out of the duct-box and out of the case. This latter fan will also pull additional pressurized air through the seams of the duct -- hence it is a good ideal to cut the duct so that it fits closely to the motherboard. You can even use components -- like the VGA card or RAM modules as "part of" the duct paneling, with the idea that narrow apertures between the duct-panels and these components will allow additional air to be drawn past those components.

Using the Ultra-120-Extreme, I've devised an improvement in ducting over my Prescott system for my Core-2-Duo build. The 120-Extreme doesn't much allow you to push air down on the motherboard, so I have a 140mm intake fan sitting behind a ducted hot-swap drive box pulling air from the front of the case, and additional ducting directs this air directly at the broad face of the CPU cooler. A 120mm exhaust fan assists in pulling this air immediately from the case. Another 120mm exhaust fan is ducted to a flat Lexan panel that sits below the 120-Extreme duct-tubes, and very close to the motherboard, so that it draws more pressurized air from the case, past the motherboard components, and out the fan exhaust and out of the case.

Any surplus pressurized air that doesn't leak into the duct box for expulsion is going to force its way through the PSU fan and vents. Everything else in the case is sealed off. Ultimately, as enough pressure builds up inside the cool part of the case volume, the intake fans become less effective, but all of this case air is going to go out of the exhaust through three paths: past the CPU cooler, past the motherboard components, and through the PSU.

On the issue of top-side fans -- true -- heat rises. But there won't be any hot air escaping into the unducted case volume area. So while I tried top-side blowholes in earlier builds before I began working with ducts, I don't do it anymore. The objective is to reduce the number of fans necessary, and make them more effective by isolating heat-radiating components within the path to immediate exhaust from the case.

This raises the issue of noise mitigation. Ideally, it is better to have fans sitting behind the hard disks and in the center of the case of their intake comes from the case-front with no leakage of air already in the case to the fan's intake side. It is also better to have fans in the bottom of the case drawing air off the floor, but the case must be suspended (on wheels, for instance) to provide airflow.

This is why I prefer full-tower cases, because there is a better chance of deploying fans like this, and it makes building and installing the ducts a bit easier. But otherwise, without these features, one might prefer a midtower case for the same reason people cite as a problem with (unducted) pressurization of cases.
 
I misunderstood your post. I knew that you use the 120 ultra extreme on your system, but when you said...

quote:

--------------------------------------------------------------------------------
Originally posted by: BonzaiDuck

There seems to be a prevalent myth that pressurizing cases is not an effective cooling method, or that it leads to stale air-pockets which increase in temperature within the case.

--------------------------------------------------------------------------------

I thought you were saying this with reference to a fan that blew perpendicular to the mobo and that if it were pressurized correctly then the airflow would still move as well as one that blew parallel.

After I re-read your post though, I don't know how I came to that conclusion. I read it at work and I was busy =)
 
Heh! You know what?

This is an interesting thread on all sorts of levels...

Heatsinks (and heatpipes) do the majority of cooling in an air-cooled setup!

All fans do is remove hot air! Air, after all, is a poor conductor of heat, as we can all agree...

To my way of thinking, fans should be used for respiration, e.g. ventilation, not cooling!

Put another way, cool air in, hot air out...

This air movement should NOT be used for cooling purposes, per sey, but rather evacuation of hot air from inside the case!

It's like 'fart fans' in bathrooms. The idea isn't to make sh!t smell better, but to get rid of the stink! 😀

I think that's the Achilles tendon of face-down coolers...

Ppl assume air cools, which it doesn't. Sinks and heatpipes cool, and fans take away the heat.

If every device surrounding the face-down fan cooler on your mobo had a sink or heatpipe installed - MOSFETs, RAM, Caps, NB, et cetera - these face-down fans could and would make a difference, but...

All these face-down fan coolers do is disturb air flow and recirculate hot air over the same devices over n' over again!

It's like aiming a fan at your butt, and expecting the smell to go away! All it does is spread to other rooms...

That's why face-down coolers don't work worth jack...
 
Ah! But they do! I DID sink my Mosfets and PLL chips, and replace the chipset heatsink with ThermalRight's. Even so, the stock ASUS chipset heatsink did pretty well with the ducting.

Without the ducting -- you're right -- you wind up mixing warmed air with cooler air, and the entire case heats up. Air DOES have a much lower ability to conduct heat than water, but it is fairly plentiful, so we run a lot of CFMs through our cases to compensate. With the ducting, there isn't any "recirculation" of hot air -- the air coming through the CPU heatsink fins picks up more thermal energy from the motherboard components and then Whoosh! -- out the exhaust fans.

But heat-pipes are a hybrid animal. They combine "air-cooling" with water-cooling, since the coolant inside the heat-pipes is water. And since we use radiators to remove heat from our water-cooled rigs, water-cooling is not divorced from air-cooling.

The problem with water-cooling is this: a simpler, manageable water-cooling rig may have a water-block on the CPU and a water-Block on (one or more) VGA card(s). We add complexity by adding a water-block to the Northbridge chipset, but the complexity is only additional hoses, water-block, etc. But even here, without fans moving air to cool other components, water-cooling misses the Mosfets and other warm components. A ducted motherboard with thoughtful fan deployment would help a water-cooled rig.

You can build an air-cooling duct arrangement so that every PCI card, every VGA card -- has dense air forced through a low-volume space over those components, and even if you're using the same air serially, enough CFM will make insignificant any loss of cooling potential for the cards and components at the end of the air's journey.

Ducting is a lot of tedious work, but it doesn't cost anything. Foam board is cheap. $40 worth of clear Lexan will provide enough material to build an "over-hill-and-dale" set of duct-boxes for the entire motherboard and all expansion cards. But you spend hours measuring twice, cutting once, fiddling with Poly-Zap glue, and fitting the pieces.

It becomes a clever challenge to design the ducting pieces so they can be removed with the least trouble to service motherboard components, replace RAM modules and so on.

Didn't several models of Porsche's use air-cooled engines, like the VW beetle?
 
Side-blowing fans would be very hard to implement. The side-blowing CPU heatsinks work because the heatsink protrudes ~17 cm out of the motherboard, thus providing a location for mounting a fan. You can only mount fans where there are protrusions. Northbridge coolers (such as my Noctua NC-U6) are now transitioning to tower-style heatsinks. However, you can only mount fans where there are large protrusions, otherwise the performance improvement is simply too small to justify adding the extra cost.

If you want a fan blowing across the motherboard, I don't see how that's going to work. It's not a issue of principle, it's an issue of execution.
 
I thought I described the design I'm implementing over the next few weeks.

Separate air-throughput for a tower-cooler and the motherboard -- by pulling air from under a flat motherboard ducting panel with a second exhaust fan . . .
 
A lot of good information for me posted here. Thanks for all the responses. And if you get some updates on your ducts post'em up bonzai =)
 
My weakness in life has been a difficulty in keeping things "under my hat" until I can reap my own benefits while "sharing."

I've kept a photographic record of my case-mod. The basic case was a freebie, SECC sheet-steel, manufactured around 1994 with an inner case-panel/motherboard-pan with proprietary mobo standoffs, but EISA slots (8 of 'em) derive from the old ISA and AT standard, and conform to the evolutionary branch of the PCI standard. In that respect, everything fits!

And -- what a marvelous, marvelous tool one finds in a tap-and-die kit -- especially, 6-32 screw-taps!

You've heard of the "four-cheese-pizza?" This is a five-case-case. The spring-latched quick-release PSU cage comes from a 1996 ATX midtower case marketed by a very . . . true-blue, establishment manufacturer. An auxiliary hard-disk cage with quick-release latches comes from a 1999 ATX midtower originally made by, ah, a WELL-known OEM (duh!) Two modifications to the hot-swap SCSI-drive cage incorporate some aluminum expander products from a popular Taiwanese manufacturer -- I'm not lyin'! The rear-corner motherboard brass-in-a-slot standoff comes from an old 486 system built by some . . . dairy farmer in South Dakota. The case itself, bigger than most mainstream full-towers, is . . . not very compact, but professional and reliable -- promising good service.

I'll leave the rest for what I post here, but before I post the grand saga of the "Great Case and Ducting Mod," I want to submit it to a contest. After what I put into this project, well, I also buy lottery tickets . . .

I'm not that proud of my "MOJO" for appearances sake -- the ducted Prescott I've shown here. I wanted to use a FrozenCPU Case-window grommet on this old 1995 Gateway full-tower, but the ThermalRight VGA cooler heat-pipes poked against the case-panel, and to make a quick "kloodge" so I could get my beloved MOJO up and running, I made a quick-and-dirty grommet of Akasa PaxMate (two layers), and just stuck the Lexan panel on the outside.

But the "Great Case and Ducting Mod of 2007" -- I'm getting anal-retentive about the craftsmanship, repairing my little "mistakes," attention to paint and other features. For a "freebie," in comparison to other retail cases, this was going to cost me less than $100 (additional sheet-steel and aluminum). But the paint alone has now increased the outlay by some $250.

Maybe it's like buying a lot of lottery tickets. But at least I get to keep the computer . . .

In the meantime, you can ask me the names of three "good family men" for whom each in succession my great-grand-uncle chauffeured, and I can regale you with a real "Road to Perdition" story.
 
Originally posted by: BonzaiDuck

But the "Great Case and Ducting Mod of 2007" -- I'm getting anal-retentive about the craftsmanship, repairing my little "mistakes," attention to paint and other features. For a "freebie," in comparison to other retail cases, this was going to cost me less than $100 (additional sheet-steel and aluminum). But the paint alone has now increased the outlay by some $250.

You might want to ask around at some auto body shops.
I got that tip from a friend and sure enough I asked around and found one that painted a pc case for me a nice fire engine red with the metal flakes for sparkle 🙂

It cost me 20.00 . The guy just told me I had to prep the case how I wanted it and then he let me know when he had a job to do with the color I wanted. I brought the case and he sprayed it with the excess paint he had from the car job he was doing.
He even dried it under the lights to bake the paint.
 
You're probably right, and you must be right -- per your $20 cost.

My problem was the choice of colors, and the decision to DIY.

Scroll down the price list at this link:

Killer Cans

I had attempted to use some Dupli-Color paint from the local auto-parts store -- a metallic paint.

Then I discovered that putting a clear-coat enamel on it destroys the desired effect. Even worse, you can let it cure for more than the requisite 7 days, and when you run your finger over it, the fine metallic powder just comes off and deposits on the fingertip. Those paint cans were about $5 each.

A case-mod paint guru pointed me toward Alsa Corp's line, and I was astounded at the $40-per-can price. I just decided to go through with it. I can dispense with the base-coat and use a $5 can of Rust-o-leum. I still have to use Alsa's "clear" after the color coat, but I'm hoping that successive coats of clear enamel can be made with Rust-o-leum's "Crystal Clear."

Mixing different brands of paint products is not wise as a rule-of-thumb, and I'll probably do a test to assure that the additional clear-coats with the cheaper paint will still give the desired effect -- toward applying a 3M abrasive polishing compound and then a glaze.

If it weren't for the fact that I would try and get this into a modding contest, I would've been happy with the original powder-primer-coat on the case interior and about $20-worth of some simple colors to expunge the well-known -- and hated -- IBM Beige.

I wish that Chamillionaire had come to my front door offering the GAS team's assistance in painting my case -- for free . . . .

Pimp My Ride

Spending this kind of money to "keep up appearances" ain't my style, bro' . . . .
 
Originally posted by: VinDSL
Heh! You know what?

This is an interesting thread on all sorts of levels...

Heatsinks (and heatpipes) do the majority of cooling in an air-cooled setup!

All fans do is remove hot air! Air, after all, is a poor conductor of heat, as we can all agree...

To my way of thinking, fans should be used for respiration, e.g. ventilation, not cooling!

Put another way, cool air in, hot air out...

This air movement should NOT be used for cooling purposes, per sey, but rather evacuation of hot air from inside the case!

It's like 'fart fans' in bathrooms. The idea isn't to make sh!t smell better, but to get rid of the stink! 😀

I think that's the Achilles tendon of face-down coolers...

Ppl assume air cools, which it doesn't. Sinks and heatpipes cool, and fans take away the heat.

If every device surrounding the face-down fan cooler on your mobo had a sink or heatpipe installed - MOSFETs, RAM, Caps, NB, et cetera - these face-down fans could and would make a difference, but...

All these face-down fan coolers do is disturb air flow and recirculate hot air over the same devices over n' over again!

It's like aiming a fan at your butt, and expecting the smell to go away! All it does is spread to other rooms...

That's why face-down coolers don't work worth jack...


Vin, are you trolling are or you really that 'confused'?


 
Originally posted by: etech
Vin, are you trolling are or you really that 'confused'?
Neither! Not trolling... and not confused! 😉

I use Free Association when thinking about things!

There I was, sitting in my house, thinking of a reply, and it dawned on me that computer cases and dwellings are a lot alike.

Take storage for instance...

Do ppl cool the 'storage room' in their homes? I would *guess* not... and neither do they cool their 'data storage' - their HDs - other than to half-heartedly blow a little air over them... which does next to nothing without heatsinks or heatpipes! That's why they burn out like popcorn.

Now, a question for you...

Are you trolling or are you really that 'confused'? 😀
 
Originally posted by: etech
Ok, you're just trolling along.

I just wanted to make sure before taking the time to try and educate you.
Yes, fine, educate me... don't scold me! 😀

Seriously, when it comes to this kinda stuff, I really *feel* that computers, et al, are kinda like 'doll houses' for boys! I've even seen guys put plastic frogs and stuff inside them... or Bawls bottles, et cetera... whatever!

SUP with the LED fans, cold cathodes, and lasers, yada, yada?!?!?

A certain segment hangs onto the belief that face-down coolers have some sort of mystical and magical qualities that mortals like myself cannot see...

So, yes, please educate us all with your pearls of two-line wisdom, Mr. Almighty Lifer! 🙂

Against all common sense, why are face-down coolers better than heatpipe towers?
 
I tend to agree with VinDSL about the need to get rid of the heat.

Perpendicular to the MB airflow (blowing "down" orb/radial sinks) was fine when the CPU was the hottest thing in the PC. [and it wasn't even "that" hot] It gave you a bonus of theoretically cooling the rest of the MB components.

Parallel to the MB airflow ("side" blowing towers i.e. TR-120, TuniqTower) is designed to move air and take heat with it.

A case that pulls in cooler ambient temp(25C) air; swirls it around to cool 50C CPU, NB, GPU heatsinks, and then tries to exhaust that warmed air out the back and PSU is still imparting heat into the caseframe and other components. The convoluted layout of the case just adds more surfaces for the air to hit and transfer heat. This raises the ambient temp of the case interior, which is always bad.
Heatsink cooling is a function of temperature differential and the air velocity. A case that isolated and perfectly exhausted the heat from the CPU, PSU, and GPU would have a close to ambient temp interior that would provide enough temperature differential to keep the MB components comfortable even without much airflow. Since PCs are far from perfect in too many ways, the next best thing is to make the hottest parts exhaust their heat as fast as possible in the shortest and straightest path so they don't raise the ambient case temp. By doing so, you only have to deal with the convection heat coming off of the MB heatsinks. Letting the hot air coming off of the CPU and GPU using (orb-type perpendicular flow) swirl around just heats up the interior and means you have to deal with radiant heat coming from all the surfaces that would have been close to ambient temp if that heat had been taken away at the start.

So yes, parallel is better than perpendicular in our ATX layouts.

A better solution would be to have one main airflow path and all the components and their radiators aligned to use that path. But that isn't likely to happen as long as graphic cards face down and away from the cpu. Maybe the AMD/ATI hybrid CPU/GPU will solve things when/if it comes out.

If you are really worried about MB cooling just get a case with a 250mm fan in the side panel and that can fit a tower CPU cooler. That should be enough to make the circuits happy.
 
(I'm pretty sure I'll regret bothering, but I just can't help myself...)

Example 1.
=====================================================
For the purposes of discussion, assume the following...

- your case has a removeable motherboard tray and the opening (clearance above the mb, in order to slide the tray out) is 150mm tall.
- Side-blower HSF A is 160mm tall.
- Top-blower HSF B is 143mm tall (including fan)
- Top-blower HSF C is 160mm tall (including fan)

Scenerio 1:
-----------------
All other factors being equal (weight, price, fan speed/noise etc.), also assume that:

- HSF A out-performs HSF B by 4C at an extreme overclock
- HSF A out-performs HSF C by 1-2C at an extreme overclock
- both HSF B&C still allow that same extreme overclock, but the cpu is a few C hotter

Question:

Which is the 'better' heatsink? (hint, it's a trick question)


Scenerio 2:
-----------------
All other factors being equal (weight, price, fan speed/noise etc.), also assume that:

- HSF A out-performs HSF B by 8C at an extreme overclock
- HSF A out-performs HSF C by 4C at an extreme overclock
- neither HSF B or C still allow that same extreme overclock, but do allow one just shy of there.

Question:

Which is the 'better' heatsink? (hint, it's a trick question)


Scenerio 3:
-----------------
All other factors being equal (weight, price, fan speed/noise etc.), also assume that:

- HSF A out-performs HSF B by 8C at an extreme overclock
- HSF A out-performs HSF C by 4C at an extreme overclock
- neither HSF B or C still allow that same extreme overclock, but do allow one just shy of there.
- you don't plan to overclock

Question:

Which is the 'better' heatsink? (hint, it's a trick question)



Example 2:
=====================================================

For the purposes of discussion, assume the following...

- your case has a fan on the side, blowing right down onto the HSF

Scenerio 1:
-----------------
All other factors being equal (weight, price, fan speed/noise etc.), also assume that:

- Side-blower HSF A cools your cpu to 50C under load, at your current overclock.
- Top-blower HSF B cools your cpu to 45C under load, at your current overclock.

Question:

Which is the 'better' heatsink? (hint, it's a trick question)


Scenerio 2:
-----------------
All other factors being equal (weight, price, fan speed/noise etc.), also assume that:

- Side-blower HSF A cools your cpu to 45C under load, at your current overclock.
- Top-blower HSF B cools your cpu to 50C under load, at your current overclock.

Question:

Which is the 'better' heatsink? (hint, it's a trick question)



Example 3:
=====================================================

For the purposes of discussion, assume the following...

- you have $50.

Scenerio 1:
-----------------
All other factors being equal (weight, performance, fan speed/noise etc.), also assume that:

- Side-blower HSF A costs $30 (including shipping).
- Top-blower HSF B costs $50 (including shipping).

Question:

Which is the 'better' heatsink? (hint, it's a trick question)


Scenerio 2:
-----------------
...nevermind - I'm bored now 🙂.
 
Ok.. so I have too much time on my hands 🙂. Don't bother replying to the above, unless it's something intellegent (or at least witty).

The point I was/am trying to make is that there are any number of valid reasons why someone might consider a top-blower heatsink to be the 'best' solution for thier setup.
 
No need to apologize, Spanki. I could also talk about "time on my hands," but I need to take a break between stripping paint off case parts.

I'll keep this brief, and it's not a "well-tested" observation, but it seemed fairly certain when I made it. [I still have to work through Spank's examples and scenarios, but I'll look it over after my next paint-stripping episode . . ]

In the process of jerry-rigging my own "ducted mobo" mod -- and actually before I started on it -- I had a 120mm side-panel blow hole with the retail black-vinyl duct tube cut to fit within an eighth inch of the [XP-120 at that time] CPU fan.

I also filtered the air with one of those two-part frozen-CPU fine-mesh wire screenies. And I tried the filter in two ways: filter on the side-panel, and filter on the CPU fan. You'd think with the former option, that the CPU-fan would pull additional pressurized air from the case interior through the gap between the vinyl duct and the fan, but no help there to my recollection.

My best results for ducting caused me to close off the blow-hole like an abandoned mine-shaft. The best results came from multiple intake fans -- even with the same screen filters and an otherwise closed case with no blowhole with air-movement only through the intake and exhaust fans -- and the CPU fan pulling air between them. I think the author of the OverClocker's article on ducting made the same observation.
 
Originally posted by: gorobei
I tend to agree with VinDSL about the need to get rid of the heat.

Everyone agrees that the goal is to get rid of the heat.


Perpendicular to the MB airflow (blowing "down" orb/radial sinks) was fine when the CPU was the hottest thing in the PC. [and it wasn't even "that" hot] It gave you a bonus of theoretically cooling the rest of the MB components.

Why did you say "theoretically"?


Parallel to the MB airflow ("side" blowing towers i.e. TR-120, TuniqTower) is designed to move air and take heat with it.

Downward blowing heatsinks were designed to move air and take heat with it.



A case that pulls in cooler ambient temp(25C) air; swirls it around to cool 50C CPU, NB, GPU heatsinks, and then tries to exhaust that warmed air out the back and PSU is still imparting heat into the caseframe and other components. The convoluted layout of the case just adds more surfaces for the air to hit and transfer heat. This raises the ambient temp of the case interior, which is always bad.
Heatsink cooling is a function of temperature differential and the air velocity. A case that isolated and perfectly exhausted the heat from the CPU, PSU, and GPU would have a close to ambient temp interior that would provide enough temperature differential to keep the MB components comfortable even without much airflow. Since PCs are far from perfect in too many ways, the next best thing is to make the hottest parts exhaust their heat as fast as possible in the shortest and straightest path so they don't raise the ambient case temp. By doing so, you only have to deal with the convection heat coming off of the MB heatsinks. Letting the hot air coming off of the CPU and GPU using (orb-type perpendicular flow) swirl around just heats up the interior and means you have to deal with radiant heat coming from all the surfaces that would have been close to ambient temp if that heat had been taken away at the start.

So yes, parallel is better than perpendicular in our ATX layouts.

Not really proven but for the sake of argument are you postulating that any tower cooler will beat any downward blowing heat sink?



A better solution would be to have one main airflow path and all the components and their radiators aligned to use that path. But that isn't likely to happen as long as graphic cards face down and away from the cpu. Maybe the AMD/ATI hybrid CPU/GPU will solve things when/if it comes out.

If you are really worried about MB cooling just get a case with a 250mm fan in the side panel and that can fit a tower CPU cooler. That should be enough to make the circuits happy.


We work with what we have and that includes the necessity of cooling not only the processor but the other motherboard components.

"Heatsink cooling is a function of temperature differential and the air velocity.

I agree so I'd rather just use a down blowing heatsink in a case such as the antek 900 with it's 200mm fan pulling the heat out of the case. You get the velocity of the close fan on the hot components.

http://images.tweaktown.com/im.../news_gigaasusttst.pdf
 
Back
Top