It is currently Wed Mar 20, 2019 1:22 pm

All times are UTC - 8 hours




Post new topic Reply to topic  [ 31 posts ]  Go to page 1, 2  Next
Author Message
 Post subject: Ducting PSU exhaust to CPU HSF intake?
PostPosted: Sun Feb 27, 2005 5:23 pm 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
I'm serious.

Consider a single-HD system with low-power (maybe even on-board) graphics and reasonable powerful CPU. Assume we want to achieve the absolute lowest noise possible (who doesn't? :roll:).
Given a reasonable choice of components and judicious placement/muffling of the HD, wind noise will be dominating, meaning, the low-noise goal essentially translates to a low-flow goal. There should be a single (muffled) intake and a single outlet, the rest sealed and foamed and damped and....
To achieve lowest flow, that air needs to be heated up maximally without overheating the computer. The question then is, in which order to place components along the way the air traveles through our system. People have come up with systems running on a single (PSU) fan. But to my knowledge, the PSU has always been last in the chain.

My proposal to put the PSU before the CPU is based on the intuition that the component that produces the most heat, and that can tolerate the highest temperature should be last in that thermal chain.

A couple of observations:
  • If the PSU is very efficient (say 80%), and the total power draw is CPU dominated (our imaginary machine will be folding for SPCR, of course :wink:), the CPU produces 3-4 times more heat than the PSU.
  • The tolerance to high temperature can be well assessed for CPUs (it's stable or not), less so for PSU (it might work for a month at 65C, but that doesn't mean it won't die soon).
  • CPUs can handle higher temperatures than PSUs (is that correct, generally?)
  • Having the single remaining fan in our system as far from the exhaust as possible can only help with noise.
I guess it boils down to my feeling that a CPU that runs a little hotter would be preferrable to a PSU that runs a lot hotter.

What am I missing :?:
Or has it been done :?:

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Sun Feb 27, 2005 6:15 pm 
Offline
*Lifetime Patron*

Joined: Sat Jan 18, 2003 2:19 pm
Posts: 5316
Location: St Louis (county) Missouri USA
This system is basically a one-fan setup.....with the fan located between the CPU heatsink and the PSU(no fan in the PSU). It blows outward through the PSU. The two hottest parts of the system are last in line, with the heat then exiting the case.

Now if you wanted the CPU last in the airflow path, just how would you physically arrange the case? You'd want the airflow off your CPU the exit the case without heating anything else up. That would mean a tight duct to the outside.

That setup in the pictures could be modified to operate like that, by reversing the airflow direction of the fan, and extending the cpu shroud to blow out through the rear case opening. All the intake would then come in through the PSU.

I really don't see any benefit to that setup.....plus the rest of the case would have very little airflow. just where would you locate your PSU in your setup? :?


Top
 Profile  
 
 Post subject:
PostPosted: Mon Feb 28, 2005 12:49 am 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
I was thinking about that starting from the configuration I have in my second rig. I already have the CPU exhaust ducted to the outside in there. The PSU could flip down, so its intake is to the left of the memory (all directions relative to looking on the upright case from the front), just below the horizontal metal bar. Its exhaust would blow towards the rear, right where the CPU intake is. A second shroud surrounding the existing one would likely be needed to force some of the air over the northbridge and voltage regulators (see comment on that further down in the post). Shouldn't be hard to do at all. The optical drive would obviously go to one of the (now accessible) upper bays.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Mon Feb 28, 2005 3:24 am 
Offline

Joined: Mon Dec 06, 2004 8:05 am
Posts: 92
i guess you could have PSU -->CPU--> graphics (with exhaust cooler). ok, so graphics cards dont kick out much heat, but they have a higher heat tolerance than CPU!


Top
 Profile  
 
 Post subject:
PostPosted: Mon Feb 28, 2005 4:01 am 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
Yes, I thought about that, and I think you're right. Personally, my graphics is so low power that it wouldn't be worth the effort. Graphics cards are kind of hard to move around in a case, so such a setup would be a lot more involved. But it could work in a case where the M/B is upside-down.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Sun May 29, 2005 11:25 pm 
Offline

Joined: Sun Jan 16, 2005 2:59 am
Posts: 1066
Location: Santa Cruz
maybe its in the wording,but the PSU would be taking warm air from in the case,heating it more and blowing it at the CPU,which would need big rpms to get much heat transfer. You would need real high rpms also in the psu as it then re-sucks the same air and creates a heat cycle. You want a relatively quiet rear fan located where it pulls heat from the CPU in the quickest most direct way and gets that heat out. The PSU has itself to cool,and can pull some of the warm air out. Coolmax makes a PSU with an adjustable speed 140 mm fan,at under 1000 rpm it's quiet but has good airflow. Line up an exhaust fan to pull heat from the CPU-chipset area and you're good. The front can be a passive,filtered,intake-if it has enough area. A fan in front is just noise too near your ears.

If you let heat build up and circulate in the box,the CPU fan runs hot-loud...not good. A duct may help. You may want to reverse the CPU fan and have a duct catch that heat and suck it out quick...but the psu to cpu flow is not gonna do well at all


Top
 Profile  
 
 Post subject:
PostPosted: Sun May 29, 2005 11:28 pm 
Offline

Joined: Tue Mar 22, 2005 11:54 pm
Posts: 58
Get a fanless PSU (I'd recommend the Etasis, it barely gets warm with my system), an Athlon64 with a SonicTower or Titan Vanessa L Type and you'll easily cool the whole system with a single 120mm low speed fan in the back.

Ducting to the PSU usually gets the PSU fan to ramp up and you can't easily decouple those.

A passive Radeon 9200 is just fine and doesn't care for airflow much at all if you don't want to game with the rig.


Top
 Profile  
 
 Post subject:
PostPosted: Tue May 31, 2005 2:50 pm 
Offline

Joined: Fri Feb 06, 2004 11:21 am
Posts: 363
Location: Houston, TX, USA
Actually, this has been done. When ATX was in its infancy, some of the very first ATX power supplies drew air from the outside of the case and exhausted it into the case. In many cases, the CPU was located in the exhaust path of the PSU. Basically, this is the exact opposite of most ATX power supplies on the market today.

My sister had such a machine. It had a 200W ATX power supply in the "normal" location, with the "reverse" airflow. Directly below the power supply, in the exhaust path, sat a Pentium MMX 233 with a large passive heatsink. The case also had a 92mm case fan in the back, exhausting air from the case. This set up a fairly short loop for the CPU heat. However, when the 92mm fan died, the CPU did start having overheating problems. It was easy enough to fix with a new case fan, though.

I have since moved this power supply to another computer. Directly beneath the power supply's exhaust, this time, is a hard drive cage. The motherboard is a Baby AT (FIC VA-503A) with the CPU towards the front. The CPU is an AMD K6-III+ 400 MHz, running at 550 MHz @ 1.8V. The CPU has a dinky Coolermaster DP5-5E11 Socket 7 HSF. The graphics card is a Geforce4 MX 420. There is no case fan at all in the computer, and it runs just fine. :)

The question remains how well this can be adapted to higher power components. But yes, it has been done before. ;)


Top
 Profile  
 
 Post subject:
PostPosted: Tue May 31, 2005 10:01 pm 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
ronrem wrote:
maybe its in the wording,but the PSU would be taking warm air from in the case,heating it more and blowing it at the CPU,which would need big rpms to get much heat transfer. You would need real high rpms also in the psu as it then re-sucks the same air and creates a heat cycle.
Sorry, I wasn't clear on this but of course this would only work, if the air only goes one way, that is, if the system is a single channel in terms of airflow and air can only go from the PSU to the CPU and from there out through the back. Of course this would need some serious ducting.

mongobilly wrote:
Get a fanless PSU (I'd recommend the Etasis, it barely gets warm with my system), an Athlon64 with a SonicTower or Titan Vanessa L Type and you'll easily cool the whole system with a single 120mm low speed fan in the back.
The question was about how to best arrange standard components. Of course with a fanless PSU, the optimal arrangement will be different. But it so happens that I own a Seasonic Super Tornado, and I'm not going to buy a fanless PSU to replace it...

mongobilly wrote:
Ducting to the PSU usually gets the PSU fan to ramp up and you can't easily decouple those.
Huh? I'm planning to duct the output of the PSU to the CPU. So the air coming into the PSU has seen heat only from the harddrives, a low-power graphics card and the memory. On the contrary, this should help in not having the fan ramp up. This being said, I do plan to run the PSU fan off a fanmate, just because the ST I happen to own is a RevA2, and its erratic fan will probably need to be tamed, even in this setup. Yes this will void warranty, but it's easy enough to do.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Thu Jun 09, 2005 2:04 am 
Offline

Joined: Sun Jan 16, 2005 2:59 am
Posts: 1066
Location: Santa Cruz
I see no point in sendig air to the CPU that has already been preheated by the PSU. You want ,ideally,both PSU and CPU getting the coolest air possible,and you want to get the warmed air out asap. Heat=noise


Top
 Profile  
 
 Post subject:
PostPosted: Thu Jun 09, 2005 4:06 am 
Offline

Joined: Mon Aug 09, 2004 5:49 am
Posts: 26
Interesting idea (PSU > CPU airflow) and it would not difficult if your case allows you to put your PSU in the bottom and at the front. Personally I like redundancy so I would add another 12 cm midships, as exhaust or at the top.

For optimum results I think you should get rid of the PSU's original casing and guide the airflow as much as possible.

i.e.
intaketunnelover psu > fan > tunnel > fan > exhaust tunnel over cpu heatsink

I guess absolute heat dissipation of a CPU is not that different from the heat dissipation of a PSU (?).

If your PSU is 75% efficient and stressing the system, the PSU might


Top
 Profile  
 
 Post subject:
PostPosted: Thu Jun 09, 2005 4:07 am 
Offline

Joined: Mon Aug 09, 2004 5:49 am
Posts: 26
Interesting idea (PSU > CPU airflow) and it would not be very difficult if your case allows you to put your PSU in the bottom and at the front. Personally I like redundancy so I would add another 12 cm midships, as exhaust or at the top.

For optimum results I think you should get rid of the PSU's original casing and guide the airflow as much as possible.

i.e.
intaketunnelover psu > fan > tunnel > fan > exhaust tunnel over cpu heatsink

I guess absolute heat dissipation of a CPU is not that different from the heat dissipation of a PSU (?).


Top
 Profile  
 
 Post subject:
PostPosted: Thu Jun 09, 2005 4:20 am 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
I really don't understand why people here think this is a strange or novel idea. I'm in the middle of a major restructuring of several systems, and all three of the new/modified systems use this single fan strategy.

It's not hard to implement, you just flip the fan on the PSU and duct the PSU output toward the CPU. On my two mini-tower cases, this essentially means PSU->CPU->GPU. On my scratchbuilt case, this essentially means PSU->GPU->CPU.

One thing that radically increases my flexibility is using notebook drives. These drives are so thermally tolerant that it really doesn't matter where I put them in the airflow chain. They don't even need to be in any airflow path at all!

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Fri Jun 10, 2005 1:15 am 
Offline

Joined: Sun Jan 16, 2005 2:59 am
Posts: 1066
Location: Santa Cruz
I'd totally understand using the PSU fan to pull the cpu heat out of the case,or at least to help. I was thinking about a passive heatpipe type cpu coolor and the Coolermaster PSU that has the adjstable speed 140 mm fan. The size and airflow,even at low rpm,from that 140 could likeley move enough air that a Venice 3000 could cool okay,but the intake path would need to be planned out. It would be a homemade case,desktop style,designed to run in this setup.. Probably for safety,I'd mount a very low rpm fan,motherboard controlled to come on if things start getting hot.

I assume you will have a fan type heatsink on your CPU? Will it be fixed speed or thermal control?


Top
 Profile  
 
 Post subject:
PostPosted: Fri Jun 10, 2005 5:48 am 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
ronrem wrote:
I see no point in sendig air to the CPU that has already been preheated by the PSU. ... Heat=noise
I don't quite agree. It's airflow=noise, heat itself is actually quite silent. And of course having parallel airflow streams cooling PSU and CPU would keep components cooler, but it means doubling airflow, increasing the number of fans, and thus increasing noise.
Quote:
I'd totally understand using the PSU fan to pull the cpu heat out of the case,or at least to help.
MikeC started a long discussion about this a long time ago. It contains many of the reasons why I was looking for different arrangements.
Quote:
I assume you will have a fan type heatsink on your CPU? Will it be fixed speed or thermal control?
The idea was to use only the PSU fan, and to arrange/duct things such that air leaving the PSU can only go through the HS and then leave the case. As my computers are thermally stable, folding 24/7, I'd fix the fan speed.

steef wrote:
Interesting idea (PSU > CPU airflow) and it would not difficult if your case allows you to put your PSU in the bottom and at the front.
Yes, this arrangement would make things easier, but I was not too keen on having the HDs sitting in warm air. They are the least heat resistant in a system, and should get fresh air, IMHO (unless you follow IsaacKuo's advice and use notebook drives).
Quote:
For optimum results I think you should get rid of the PSU's original casing and guide the airflow as much as possible.
I can already hear people scream about fire, shock and other hazards. But yes, personally I think this is an option if done carefully.
Quote:
I guess absolute heat dissipation of a CPU is not that different from the heat dissipation of a PSU (?).
Depends on the components and the way the system is loaded. I run [email protected] 24/7, so the CPU accounts for the majority of the DC power draw. If I calculate 60W for a somewhat undervolted Barton @2.2GHz, 40W for the rest of the system, and a 75% efficient CPU, I get 33W for the PSU, which is just over half of the heat dissipation of the CPU.

IsaacKuo wrote:
I really don't understand why people here think this is a strange or novel idea. I'm in the middle of a major restructuring of several systems, and all three of the new/modified systems use this single fan strategy.
I was going by the lack of systems described in the Gallery using such an arrangement. But maybe I didn't search well. Do you have links hany where such an arrangement was shown or discussed? This was actually one of the questions in my original post.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Fri Jun 10, 2005 7:28 am 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
JanW wrote:
IsaacKuo wrote:
I really don't understand why people here think this is a strange or novel idea. I'm in the middle of a major restructuring of several systems, and all three of the new/modified systems use this single fan strategy.
I was going by the lack of systems described in the Gallery using such an arrangement. But maybe I didn't search well. Do you have links hany where such an arrangement was shown or discussed? This was actually one of the questions in my original post.


I haven't actually submitted anything to the Gallery yet since its creation. I've been holding off for my major system restructuring (which I've finally just now gotten around to).

But you're right--I don't think anyone else here uses PSUs with flipped fans to take in air rather than exhaust it. Maybe this is because most people here use more modern, hotter equipment than I do. If you need the airflow of 2+ fans anyway, it seems to make more sense to feed the CPU and PSU two separate streams of fresh air.

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Mon Jun 13, 2005 1:39 am 
Offline

Joined: Mon Aug 09, 2004 5:49 am
Posts: 26
@ JanW
The point I could have made is: if the PSU does not dissipate more heat than the CPU, it is not illogical to put it in front of the CPU. Since it is my experience that [email protected] does not push a CPU to the max, it is my guestimate the heat dissipated by your PSU is lower but quite close to the heat dissipated by your CPU. But that does not really matter, does it? As long as it is lower you can put your PSU in front of you CPU anway.

Other question - can you (or why can't you) keep your HDD out of the airflow?


Last edited by steef on Mon Jun 13, 2005 7:29 am, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Mon Jun 13, 2005 3:26 am 
Offline
*Lifetime Patron*

Joined: Sat Jan 18, 2003 2:19 pm
Posts: 5316
Location: St Louis (county) Missouri USA
Have any of you people seen this setup? It has a 120mm fan above the heatsink blowing upward....and a PSU with a bottom 120mm, located further back in the case. Both these 120mm fans blow their exhaust into a common exhaust chamber.

The twin 80mm fans in the photos, only turn on at higher temps (rarely). This arrangement assures both the PSU and the CPU are cooled with case temp air, and the exhaust off both the CPU and the PSU cannot heat up each other or anything else in the case.

This setup required a specific case and CPU heatsink....but it works perfect. Everything runs cool and quiet.


Top
 Profile  
 
 Post subject:
PostPosted: Mon Jun 13, 2005 5:52 am 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
Bluefront wrote:
It has a 120mm fan above the heatsink blowing upward....and a PSU with a bottom 120mm, located further back in the case. Both these 120mm fans blow their exhaust into a common exhaust chamber.


That's nice, but it doesn't really have anything to do with the idea JanW is proposing.

I've just finished my Dad's computer and my new HTPC. My Dad's computer is an old Slot1 Gateway with PSU exhaust ducted to the CPU (flipped 92mm fan). I've taken pictures and will post them tonight. It's no "Bluefront project", but it's about as close as I'll ever get.

My new HTPC is in an Enlight desktop case. It has a Fortron 300 with its 120mm fan flipped. The fan blows air at the CPU heatsink with a partial airflow guide to help direct the airflow.

Both of these computers essentially follow JanW's idea of having a single fan that blows PSU exhaust ducted to the CPU heatsink. The CPU isn't the last thing in line in either case, though. It doesn't matter so much with my older processors, of course.

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Mon Jun 13, 2005 6:54 am 
Offline
*Lifetime Patron*

Joined: Sat Jan 18, 2003 2:19 pm
Posts: 5316
Location: St Louis (county) Missouri USA
IsaacKuo.....the idea proposed in this thread seems to be designed for low-power systems. Trying to keep a Prescott for example, cooled with already heated exhaust air from the PSU, simply means you have to run the PSU fan faster (noiser). The same thing happens with a lower heat CPU....you just wouldn't have to run the fan as fast.

IMHO....it's much easier (quieter) to cool a PSU with limited airflow, as compared to cooling a CPU with the same airflow. Plus airflow through any PSU can be modified/improved for better cooling.....easily without any expense.

To me it seems a no-brainer to put the coolest running component last in line in a single fan setup.

Of course there are many different ways to achieve a quiet setup.... :lol:


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jun 14, 2005 1:05 am 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
steef wrote:
The point I could have made is: if the PSU does not dissipate more heat than the CPU, it is not illogical to put it in front of the CPU.
Even though I agree with the conclusion, I think the argument is flawed because it does not consider heat tolerance, which is at least as important as heat disspation, IMHO. If I was confident in feeding my PSU with slow moving intake air of 40°, I'd put it last, regardless whether its heat dissipation is lower or higher than the CPU's.
Bluefront wrote:
To me it seems a no-brainer to put the coolest running component last in line in a single fan setup.
:shock: The coolest running componnent would be the harddrive. I'd insist that, on the contrary, low-power low-heat-tolerance components should be first. Having seen some of your systems, I know you agree on this for harddrives which you place in the birdcage at the entrance, not in the exhaust chamber. Then, if you compare the PSU to the CPU, it has both lower power dissipation and lower heat tolerance, and it should be before the CPU.

Of course, all of this only works in a low-power system where a single-fan low-airflow setup is feasible. But my guess would be that my currently loud system (with a SS ST-300 instead of the Antec PSU and a better HSF) is borderline in that a single fan setup would be too ambitious with the PSU last, but could work with the CPU last. Unfortunately it'll still be at least two months before I get the time to start working on this.

Bluefront wrote:
Have any of you people seen this setup?
Yes, I did a similar thing in my "Box2" (see sig and gallery post), but I used a total of three fans.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jun 14, 2005 7:51 am 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
JanW wrote:
Of course, all of this only works in a low-power system where a single-fan low-airflow setup is feasible.


Thinking about the theory of reusing airflow, it can actually makes sense even with multiple fans. Suppose you have a two fan system. You can either:

1. Split the airflow with one fan for the CPU and one fan for the PSU.

or

2. Combine the airflow for the CPU and PSU in series.

It seems to me that for a given total airflow, you might provide better cooling with option 2 (PSU exhaust ducted through CPU). The PSU is cooled by twice as much air, so it will run cooler. The CPU is cooled by twice as much air, although it's a bit warmer--this could go either way.

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jun 14, 2005 5:00 pm 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
IsaacKuo wrote:
...
2. Combine the airflow for the CPU and PSU in series.

It seems to me that for a given total airflow, you might provide better cooling with option 2 (PSU exhaust ducted through CPU). The PSU is cooled by twice as much air, so it will run cooler. The CPU is cooled by twice as much air, although it's a bit warmer--this could go either way.
But but but... two fans in series don't increase airflow much if there are no severe restrictions. According to MikeC, a single "high speed" fan is quite a bit louder than two parallel half-the-airflow fans. This makes sense, as noise is mostly related to turbulence and thus the speed of the airflow. In a parallel setup, the total crossection of airflow is higher, so air flows slower for the same CFM. And adding a second "high-speed" fan in series, won't do any good.

In conclusion, if a single low-speed fan is not enough, I think serial cooling the way I proposed here is not a good idea. On the other hand, if a judicious arrangement of the components makes this feasible, there should be a slight reduction in noise that could make the difference between an audible system and a virtually silent one.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jun 14, 2005 5:42 pm 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
JanW wrote:
But but but... two fans in series don't increase airflow much if there are no severe restrictions.


I'm not talking about putting the FANS in series, but rather putting the CPU and PSU in series. The two fans should be side-by-side in "parallel". For example, with a "straight-thru" type PSU, two side-by-side 80mm fans can be put on the inner side. These could be pointed inward, and ducted to the CPU with a wide "U" bend tube.

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jun 14, 2005 5:51 pm 
Offline

Joined: Fri Jan 23, 2004 7:50 am
Posts: 1705
Location: Baton Rouge, Louisiana
Of course...ducting air to/from two side-by-side fans may be a bit kludgy. A more elegant method is to remove the fans from the CPU and PSU altogether. Then all you need is a simple duct from the bottom of the PSU to the CPU heat sink. You use two or more exhaust fans to create negative pressure to suck air through the PSU/CPU.

For example, you could rig up one exhaust fan in the bottom front "intake", and jury rig another large exhaust fan in the 5.25" bays. You could even have a third large exhaust fan on the side of the case. These fans are all working in parallel to pull air outside the case. The PSU "exhaust" becomes the one and only intake, so all air gets pulled in through the PSU and then CPU.

_________________
Isaac Kuo


Top
 Profile  
 
 Post subject:
PostPosted: Wed Jun 15, 2005 1:32 am 
Offline

Joined: Mon Aug 09, 2004 5:49 am
Posts: 26
@ JanW
Actually I did consider heat tolerance but I did not mention it, because one imho should stay clear of such temperatures (where heat tolerance becomes an issue) anyway.

Maybe I should be a bit more elaborate on my idea (of course I am reinventing the wheel here). The case can act as a chimney. On top is the CPU and below the PSU, both heating air in order to get it moving a bit. Depending on the CPU heatsink and whether or not your PSU is a passive design, you need an airguide and probably one fan too to get enough airflow through the CPU heatsink and PSU. The fan is there to support the airspeed and to promote a laminar airflow inside the airguide. I think the preferred position of the fan is somewhere halfway the case, so the noise it is generating can be well damped inside the airguide before it reaches the front intake and rear exhaust. Of course the correct execution of this idea includes removing drive bays - otherwise the room for an airguide is too restricted.

A variant would be to drop the PSU into the external drive bays and create an airduct behind it towards your CPU. The airflow can be made laminar. This however would not take advantage of the natural tendency of air to rise.

A final variant would be to have a duct above the CPU exhausting to the rear and to have the PSU in the usual place with a separate duct passing the CPU. Both CPU duct and PSU duct can then share their intake. This setup does take advantage of the fact that heated air risese. It will however not result in a laminar airflow, because there the airflow has to be divided between CPU and PSU and the rather sharp bends that the airflow has to go through. Of course this idea is not very advantages if you PSU has a (12cm) bottom intake.

Personally I am a bit uncomfortable with a single-fan system if the airflow meets bends. To get the same amount of cooling the fan will probably have to move more air.

I have of course seen the pictures of your system. You have already created a lot of room for damping, i.e. the air intake. Also you already have a lot of experience with ducting and damping. Actually I see no reason at all why I would advise you other than being plainly arrogant myself. If you would just get yourself a heatpipe tower CPU heatsink ...

Good luck with your conquest


Top
 Profile  
 
 Post subject:
PostPosted: Fri Jul 22, 2005 2:26 pm 
Offline

Joined: Sun Jan 16, 2005 2:59 am
Posts: 1066
Location: Santa Cruz
If you are driving a car across the Mojave in July,overheating is a major concern-while it would be no issue in April in Maine. Same engine,load,airflow but the Radiator-the heatsink as it were,is dealing with HOT ambient air and the reduced temperature differential makes the heat exchange process far less effective.

You seem to be intending to pull your air in via the PSU-blowing it at the Passive CPU heatsink-without a second fan to draw the twice heated air from the hot areas,CPU,chipset,etc,leaving it to swirl about the rugged terrain in there,maybe some heat ultimately seeping out.

Try some experiment,and you will see that it is not easy to get a controlled passive exhaust flow.and you will need multiple outlets or you get the equivalent of a backwater eddy,corners where that hot air swirls about-but does not escape. I see a basic plan that is unique,but that's bcause it's nearly impossible to get to work well-while the suck-it-out the PSU scheme has been done a lot.

Coolermaster has a 120 mm fan that purrs along at 720rpm,stock,could run quieter/slower-adding no significant noise. Run the Seasonic in its normal exhaust mode-replacing the fan if you feel bold. Create an INTAKE pipe/duct-having an inlet on the FLOOR-with a small T to send some cool air to the chipset area,and then an elbow directing the rest of the inflow right at the CPU heatsink unit,thus there is a decent temp differential,and the PSU fan and the low speed 120 give you the draw to pull in ample air, to assure that air is coming up the pipe-you silicon seal your assorted airleaks-or use a non hardening putty where stuff may need future removal. You leave a SMALL frontal intake that lets some air circultate. around the drives.

Going a tad quieter with the PSU fan should offset what little sound a 720rpm 120 brings and it's less likely you toast the CPU or mobo.

I'm assuming you are not going to run heavy games,folding,high load stuff?


Top
 Profile  
 
 Post subject:
PostPosted: Fri Jul 22, 2005 5:07 pm 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
ronrem wrote:
... the reduced temperature differential makes the heat exchange process far less effective.
Yes, of course. For constant airflow, the amount of energy transferred is simply proportional to the difference between the temperatures of the heatsink surface and that of the cooling air. That's what my point is about: components that can stand running hot should be cooled last, because there will still be a sufficient temperature difference even if the cooling air is pre-heated. I'm looking at a mid-power CPU here, that is nowhere near its temperature limit when cooled with an efficient heatsink and cool air.

ronrem wrote:
Try some experiment,and you will see that it is not easy to get a controlled passive exhaust flow. and you will need multiple outlets or you get the equivalent of a backwater eddy,corners where that hot air swirls about-but does not escape.
Ahh, this is an interesting argument that goes back to the positive pressure vs. negative pressure debate. If you have a sealed box with a single inlet and a single outlet, I would argue that it does not matter on which side of the box the fan is (that is, whether the exhaust is passive or the intake is passive). The only (small) differences are the absolute air pressure inside the box, and the profile of airflow speeds across the intake and outlet. Airflow patterns in the box at some distance from inlet and outlet (eddies in the corners) should be identical. So in the end, it depends on how close the fans are to the critical components, and on the precise geometry. Think about the box around the CPU and heatsink in my silent rig (see link to gallery post in the sig). For the Zalman7000 I use in my silent rig the profile of airflow speeds actually does matter and I do need the fan on the heatsink, (that is, the intake to the box) not for absolute airflow but to push the air close to the core where fins are tightly packed.

ronrem wrote:
Coolermaster has a 120 mm fan that purrs along at 720rpm,stock,could run quieter/slower-adding no significant noise.
Granted. There is a lot to be said in favor of multiple fans (as the sticky by MikeC I linked to earlier in this thread shows quite impressively). At the level of a very slow 120mm fan (say, undervolted Nexus), noise is so low that hardly anyone would go to any length to reduce the number of fans from two to one, as you point out. But this is about what is feasible and how to go about pushing the limits. Maybe I should have entitled the thread "Best design for a single-fan setup?"

ronrem wrote:
I'm assuming you are not going to run heavy games,folding,high load stuff?
Gaming, no. But of course I'll continue to fold 24/7!!!

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
 Post subject:
PostPosted: Mon Jul 25, 2005 10:02 am 
Offline

Joined: Sun Jan 16, 2005 2:59 am
Posts: 1066
Location: Santa Cruz
I can see giving it a shot as an experiment-seems the hard way to me but I guess it can-maybe-work. Do be sure you can monitor temps so things don't fry if the concept doesn't pan out-and good luck :wink:


Top
 Profile  
 
 Post subject:
PostPosted: Tue Jul 26, 2005 1:32 am 
Offline

Joined: Fri Dec 03, 2004 12:38 pm
Posts: 296
Location: France, Europe Folding for SPCR
ronrem wrote:
I can see giving it a shot as an experiment ... good luck :wink:
Thank you :D . I keep postponing this project, though, because I'm really busy at the moment and need that computer pretty much on a daily basis.

_________________
ImageSilent rig: muffled Sonata; relocated Seasonic-SuperTornado; ducted 7000A


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 31 posts ]  Go to page 1, 2  Next

All times are UTC - 8 hours


Who is online

Users browsing this forum: No registered users and 7 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group