A LOW-tech guided airflow rig...

Show off your quiet rig.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
SpyderCat
Posts: 208
Joined: Sun Feb 23, 2003 12:22 pm
Location: The Netherlands

A LOW-tech guided airflow rig...

Post by SpyderCat » Sat Mar 08, 2003 8:34 am

Annoyed as I was with the noise from my computer I tried to optimize the cooling of my computer.
I thought it was probably best to guide the air through the case, thereby preventing the contamination of fresh air with used air from the heatsink.
Only recently I found SilentPCreview.com, and here I learned to decouple the harddisk.
Lets have a look.

Image

An overview of my modified rig with the “divided case” system. In the lower right corner the decoupled harddisk, and to the left the cardboard devider that guides used air to the PSU. The rig has only two fans, one in the PSU, and the fan on the heatsink.

Specs:
AMD Athlon XP 1800+
ASUS A7V333 256 MB RAM ( Raid present, not used )
IBM Deskstar 120 GXP / 120 MB
LiteOn DVD
LiteOn CDRW
Floppy drive
Coolermaster Al. heatsink with Coolermaster fan ( 70mm x 20 mm)
Nexus NX-3000 PSU.
Pine Nvidia Geforce2 MX400 / 64 MB
Etech modem
Intel NIC
Cheap generic case.



Lets have a look at the harddisk first.

Image

I suspended the harddisk from elastic bands hung from the floppy-drive cage.
At the rear there wasn't enough material to support the elastic bands, so wooden sticks fastened with tie-ribs were used to lengthen the support.
4 pairs of holes in the bottom of the case with tie-ribs were used to fasten the elastic down below.
With a heated nail I burned 4 holes on the elastic, and the drive was fastened with it's original screws.
It's a pretty cool place with sufficient airflow there as the drive (IBM 120 GXP-120 MB) typically has a temperature of some 36*C/96*F

Image

Here you see the cardboard divider, separating the used warm air from the cooler air in the case. It is supposed to force the warm air directly to the Nexus PSU to be extracted; and I must say it does it’s job remarkably well.
The graphics-card seals the “hot chamber” at the bottom. To the right the are some, not intended, openings.
Looking up from below you see this.

Image

Again a few openings, but it’s clear that most warm air is being sucked into the PSU.

Image

In an attempt to prevent fresh air from entering the PSU without cooling anything, most
holes above the graphics-card have been closed with adhesive tape.
The heatsink and fan are from CoolerMaster, The fan is a “silent” type (70mm x 20 mm), with 3100 rpm maximum.
Writing this (almost idle) my CPU has 35*C/95*F and the MoBo 29*C/84*F.
Not too bad.

Image

The upper holes in the casing's side are closed with adhesive tape too.

Image

And while we are at it, these gaps above the heatsink-clamp are closed with tape too.


Funny that almost everything I did is in contradiction with the “general advice” on this web-site.
I restricted airflow in a BIG way and didn’t add fans.


I kind off grew fond of modding and tried something else.
What if I could completely seal the air coming from the heatsink, and bring it directly to the PSU.
At http://www.procooling.com/reviews/asset ... banner.jpg I read that Joe tried several times to reverse the airflow over the heatsink, and the results were not bad at all. Having the fan suck air through the heatsink would make it much easier to capture the used air and bring it directly to the PSU.
This would become the “pull-duct-pull” system. Lets have a look.

Image

The fan on the heatsink has been reversed so it pulls air through the heatsink. With a thin, light foam a duct was fabricated, so it connects directly to the PSU’s intake at the bottom of the PSU. The rear PSU’s intake was (almost) completely closed with a piece of foam in order to prevent the heatsink’s exhaust air from re-entering the case. With some obstructions in between, now both fans are connected in a serial way, thereby offering some redundancy should one of them fail.

details:

Image

Image

In these last two photographs the case is standing upside down to have a better lighting and a better angle.


The results:
Noise: Hard to say. I guess about the same or somewhat less.
CPU idle temp: 1-2 degrees C higher, from 35* à 36*-37* :(
CPU load temp: 3 degrees C lower !! :D
MoBo temp: 7 degrees C lower, and no rise in temperature under load !! :D

Image

A game of Warcraft2 with the cardboard divider.
Note how the MoBo temp slowly rises

Image

And here a game of Warcraft2 with the pull-duct-pull system.
MoBo temp doesn’t go up, and CPU-temp is 3 degrees lower!


CPUBurn gives a very heavy load on the CPU, a load not found in day to day computing.

CPU temp after 10 minutes CPUBurn (idle) 70*C/158*F
CPU temp after 10 minutes CPUBurn (normal) 70*C/158*F
CPU temp after 15 minutes CPUBurn (high) 70*C/158*F

Image

These are the three CPUBurn tests.
At the end of the third test the PSU gets warmer and starts revving up.

As said before, this is by no means a totally silent system, it’s a quiet system.

There is still plenty of room for future devellopment.
* I would kill for a few Panafloos (No Panaflo here in Europe !)
* With a panaflo I could dump the heatsinks exhaust-air directly out the rear of the casing.
* Two panafloos pushing air through the PSU would be worth a try too.

Comments are appreciated.

Regards, Han.

GamingGod
Posts: 2057
Joined: Fri Oct 04, 2002 9:52 pm
Location: United States, Mobile, AL

Post by GamingGod » Sat Mar 08, 2003 10:34 am

I have to say that it doesnt look as bad as some of the stuff ive made with cardboard. Have you tried covering all the little holes in the back so the air has to be drawn across the hardrive?

SpyderCat
Posts: 208
Joined: Sun Feb 23, 2003 12:22 pm
Location: The Netherlands

Post by SpyderCat » Sat Mar 08, 2003 10:57 am

GamingGod wrote:I have to say that it doesnt look as bad as some of the stuff ive made with cardboard. Have you tried covering all the little holes in the back so the air has to be drawn across the hardrive?
I would say more than half of the holes in the back have been covered. Low down the holes are open as the graphics-card needs some air too.
Judging by the MoBo temperature of 28*C I would guess that the free air in the case is a mere 26*C. Outside the case it's 20*C.
Besides, the harddrive is pretty as it is at 36*C.
Do you mean to say that you would try to lower your harddrive temperature when you had 36*C/96*F ?

I have to admit that the next project will indeed be the harddrive, but than I'll focus on noise reduction, not on temp. reduction.

Regards, Han.

slippy
Posts: 42
Joined: Fri Jan 03, 2003 3:35 pm
Location: Toronto
Contact:

Post by slippy » Sat Mar 08, 2003 12:12 pm

i may have missed the explanation, but why not make a duct to the rear exhaust fan instead of the psu?

this way the psu won't get overloaded with all the heat from the cpu and video card

SpyderCat
Posts: 208
Joined: Sun Feb 23, 2003 12:22 pm
Location: The Netherlands

Post by SpyderCat » Sat Mar 08, 2003 3:16 pm

slippy wrote:i may have missed the explanation, but why not make a duct to the rear exhaust fan instead of the psu?
this way the psu won't get overloaded with all the heat from the cpu and video card
Well, I wrote:
There is still plenty of room for future devellopment.
* I would kill for a few Panafloos (No Panaflo here in Europe !)
* With a panaflo I could dump the heatsinks exhaust-air directly out the rear of the casing.
* Two panafloos pushing air through the PSU would be worth a try too.
For the moment the Nexus seems to be able to cope easily. It can, and does, increase its rpm when the temp rises.

Having the current CPU-fan exhausting to the rear would increase the noise, I think.
This solution needs a silent fan capable of sucking enough air through the heatsink, or a large heatsink with an efficient airflow where a slight under-pressure is enough to consume large amount of air.

I have some trouble finding the right words here, but I'm trying to use the "counter-flow principle". It is used in heat-exchangers everywhere, EXCEPT in most heatsinks. When two fluids have exchanged heat using the "counter-flow principle" (could somebody tell me the right phrase?) using an appropriate heat-exchanger the heat exchange is almost 100%.
One should design heatsinks, or use heatsinks, in such a way that the cooling fluid travels a long path along the exhange surface, and meets an ever increasing temperature. Having passed the heat-exchanger the cooling-fluid will be so hot it isn't any good for cooling anymore and should be dumped overboard WITHOUT CONTAMINATING the environment any further.
This way you need the least amount of cooling-fluid, resulting in few low revving fans.

I hope this makes my intentions clear.

Regards, Han.

d_kay
Posts: 40
Joined: Tue Dec 03, 2002 8:17 am

Post by d_kay » Wed Mar 12, 2003 1:26 am

okay, this may sound stupid, but:

Is the cpu-fan blowing air onto the heatsink, or is it extracting the air from the heatsink?. Which is the most common?.

NeilBlanchard
Moderator
Posts: 7681
Joined: Mon Dec 09, 2002 7:11 pm
Location: Maynard, MA, Eaarth
Contact:

Most blow air down through the HS

Post by NeilBlanchard » Wed Mar 12, 2003 5:27 am

Hello d_kay:
d_kay wrote:Is the cpu-fan blowing air onto the heatsink, or is it extracting the air from the heatsink?. Which is the most common?.
Most HS fans blow air down through the HS, though there are some that work better when air is sucked up through the HS; like the Alpha PAL8045. It has a shroud that gets the air in down at the base of the pins (it has pins w/ many tiny ridges, rather than fins) and this works better mainly because the coolest air hits the hottest part of the HS first, and therefore has the most rapid heat transfer.

SpyderCat
Posts: 208
Joined: Sun Feb 23, 2003 12:22 pm
Location: The Netherlands

Re: Most blow air down through the HS

Post by SpyderCat » Wed Mar 12, 2003 1:18 pm

NeilBlanchard wrote: Most HS fans blow air down through the HS, though there are some that work better when air is sucked up through the HS; like the Alpha PAL8045. It has a shroud that gets the air in down at the base of the pins (it has pins w/ many tiny ridges, rather than fins) and this works better mainly because the coolest air hits the hottest part of the HS first, and therefore has the most rapid heat transfer.
Hello Neil,

I'll chime in if you don't mind.

Through Google I found the Alpha PAL8045
Image
source http://www.overclockers.com/articles454

I respectfully disagree with you on:
... and this works better mainly because the coolest air hits the hottest part of the HS first, and therefore has the most rapid heat transfer.
From the center the heat spreads through the baseplate, and than up the pins. As the route to the pins in the center is shortest, the center pins will heat up most. The cooling air meets the outer pins first and cools them, thereby allowing more heat to flow to the outer rim. The slightly heated air further inwards meeting ever hotter pins all the time, but due to the difference in temperature between the air and the "ever hotter pins" the air is still able to cool them. This is the "counterflow principle".

Would the fan be blowing onto the pins, then the coolest air would meet the hottest pins first, thereby picking up so much heat that the air isn't able anymore to cool the outer pins effectively. This situation is definately not in accordence the "counterflow principle"

Still, there room for improvement on the Alpha PAL8045.
* There is no reason why the air would travel to the center bottom pins which are hottest.
* The hub of the fan blocks the airflow to the hottest center pins.
* The easiest path for the air (shortest route) is along the coolest pins (tops of the pins in the outer rim).

A better cooler along the lines of the Alpha Pal8045 could look like this:
Image

The baseplate with pins (blue), thick in the middle, guiding air upwards.
A shield over the pins (red) sloping down to the center, and speeding up airflow.
Then a nozzle (also red) slowing down the air again as it gets wider
And (in black) the fan on top.

Instead of the fan on top, you could probably also use a duct with a 80mm Panaflo at the other end. 8)

Now, all we need is a manufacturer to make this fan for a reasonable price. :lol:

Regards, Han.

NeilBlanchard
Moderator
Posts: 7681
Joined: Mon Dec 09, 2002 7:11 pm
Location: Maynard, MA, Eaarth
Contact:

point taken!

Post by NeilBlanchard » Mon Mar 17, 2003 9:10 am

Hello Han:

Your point is well taken, but I think we're on the same page. I just didn't state my point as precisely: the tops of the pins are cooler than the bases of the pins, though they certainly are going to vary in temperature, depending on their distance from the surface of the CPU chip.

One of my earliest posts here on SPCR went into much more detail on this:

===================
Okay, Alpha recommends that you mount the fan to pull air through the PAL8045. Dan's Data (dansdata.com) says that this gains you about 5% better cooling. It seems to me that better cooling can be quieter...

Why is it better to pull the air through a heat sink? Several reasons, I think:

1. The cool air enters the heatsink at the hottest part (in this case at the base of the fins) and where there is a greater difference, there is better heat transfer.

2. Air doesn't "like" to be pushed. It tends to go places that you don't intend it to go. OTOH, air always flows towards a vaccum or a lower pressure. So, when you blow down into the heatsink, some air leaks (off to the side, for instance) instead of flowing all the way down to the base of the heatsink. The 8045's shroud causes the air to flow up from the bottom, instead of sucking it in from the sides right near the fan.

2a, b, & c: One could experiement a bit with shrouds that cover more of the sides. Also, if one were to make the shroud so that the fan was moved up and away (a little) from the tips of the pins, then things might get quieter. Another line to try would be to have the intake areas be non-symetrical so as to case some swirling and to make sure that the air flow makes it's way all the way to the center of the heatsink (it is a big one!) where the heatsink touches the CPU.

3. This is likely to be less important than #1 and #2: when a fan blows down, the air pulls some heat off the top of the heatsink -- and it moves down to the middle part and gains some more heat -- and then it finally makes it's way down to the bottom hottest part (if it hasn't leaked out the side!) and it has a harder time pulling off as much of the heat because it has already gained (some) from higher up. This is kinda' the reverse of #1, but it needs to be pointed out.

I guess this sorta' goes in the direction similar to power supply efficiency:if the heatsink plain ol' works better, then you don't need as much air flow. (For power supplies that run cooler, ditto.) Now, I know the Thermalright SLK-800 is a righteous heatsink, but I'm not sure that all the "fight" is out of the 8045. I really like the 8045's mounting setup -- very secure, whereas hanging a pound of anything on a clip is not so elegant.

Can the SLK-800 work as well in the "suck" mode as the 8045? Not without some fancy shroud, 'cuz an 80mm fan overhangs it by quite a bit. The two things the SLK-800 has going for it are the fact that two sides are already covered and soldering on some covers at the top of the fins on the other two sides would likely be easy enough. Also, the shape of the "core" of the 800 is conducive (sp?) to air flow in at the sides and then up through the top.

Thanks for reading this far!
===================

I would theorize that the Alpha PAL8045 could be improved by extending the shroud down a bit further, and the lower edge should flare out in a curve to reduce turbulence (and noise) at the sharp edge.

The problems that I see with your section are the constriction (this will raise the static pressure) and at that narrowest place, there is a sharp corner. It should be curved to increase velocity and reduce drag and noise. I like the profile of the base!

SpyderCat
Posts: 208
Joined: Sun Feb 23, 2003 12:22 pm
Location: The Netherlands

Re: point taken!

Post by SpyderCat » Mon Mar 17, 2003 10:52 am

NeilBlanchard wrote:Hello Han:

Your point is well taken, but I think we're on the same page. I just didn't state my point as precisely: the tops of the pins are cooler than the bases of the pins, though they certainly are going to vary in temperature, depending on their distance from the surface of the CPU chip.

One of my earliest posts here on SPCR went into much more detail on this:

===================
<snipped previous posting>
===================
Hi Neil,

I've completely missed your previous posting (didn't travel that far back in the archives), and we are in agreement for sure. The problem isn't that complicated that we can't come up with an optimum design. The problem is that the design has to be manufactured for a reasonble price.
NeilBlanchard wrote: I would theorize that the Alpha PAL8045 could be improved by extending the shroud down a bit further, and the lower edge should flare out in a curve to reduce turbulence (and noise) at the sharp edge.
Absolutely, and as the circumference is rather large we could bring the shroud down quite a bit, and still have enough intake area.
As the heat transfer from heatsink --> air also takes time, one shouldn't try to reduce volume, for other reasons than keeping the weight down.
NeilBlanchard wrote: The problems that I see with your section are the constriction (this will raise the static pressure) and at that narrowest place, there is a sharp corner. It should be curved to increase velocity and reduce drag and noise. I like the profile of the base!
I'm not sure here. Speeding air reduces pressure.
You are right regading the sharp corner, but I choose to draw it this way as it is easy to produce in large quantities.
And regarding the base; Yes, it looks nice but how do you make such a base in a cheap way?

Still, it's good to think about an optimum design, so we know what to look for in an existing design. Ralf pointed us to the new Thermalright SLK-800(U) Bolt-On Heatsink, and with a little modding I think this would be a very good "suction heatsink".
I'll will look around here in The Netherlands to see if I can buy it; just for fun.

Thanks for your reaction, Han.

Post Reply