Another radical cooling idea INPUT needed UPDATE- 1/2 DONE

theyangster · Post by **theyangster** » Tue Apr 11, 2006 5:22 pm

4/18 UPDATE BELOW

here we go again

so my past adventures delt with cooling the CPU mostly

here, I'm going for the whole she-bang

The purpose of this is to
1 - use one fan for general computing
2 - turn on a seperate fan to divide the critical components (CPU/GPU) airflow so each recieves seperate airflow, thus keeping components cooler
3 - have fun, be different

so the most unique part of this system is an air "valve" which will allow air to flow in only one direction, better explained in the pic below

(sorry for my crappy MS-paint skills)
the valves will be made out of cardboard and paper with string as the hinge (tape is not flexible/responsive enough)

so once the bottom fan turns on, the greater suction from the fan will open the bottom valve, allowing air thru
the middle valve will close due to the reverse airflow, opening up the rear valve, cooling the CPU

pic for reference

Benifits to this system
each major heat emitter is given it's "private" airflow, cooler components
minimal noise increase (+3 dbA assuming same fans)
sheer coolness

Problems
manual or automatic activation of the second fan (bottom fan)
the rear valve may open when it is not supposed to, solved by lowering the angle of the the valve or bigger/weighted flaps (heavier stock)?

hope you get what I'm trying to do!

Tzupy · Post by **Tzupy** » Wed Apr 12, 2006 10:03 am

I also think that the CPU and GPU(s) airflow should be separated for best performance.
Outputting GPU heat by the front intake fan, reversed, may only be possible in some cases.
I was thinking of cutting a hole in the right case panel and ducting GPU exhaust (for 7900 GTX style coolers).

theyangster · Post by **theyangster** » Wed Apr 12, 2006 11:00 am

the biggest problem is the rear valve, perhaps if I went near horizontal, it'd stay closed with only one fan on

the flaps are extremely sensitive due to the string hinge

as for cutting a hole in your case, it's up to you!

Bluefront · Post by **Bluefront** » Wed Apr 12, 2006 11:26 am

I suspect you'd have a bunch of trouble making your "air valves" work and keep working. The flap would have to be super-light, and stiff enough to retain it's shape to seal properly. Ever see a drier vent flap....somewhat like what you propose. They are all the time sticking open due to dust/distortion. I don't know about that one.

I've considered trying to construct one for computer use....but always gave up before I started.

McBanjo · Post by **McBanjo** » Wed Apr 12, 2006 11:34 am

Isn't there a risk that the flaps will be a noise issue? Paper moving in the winds sounds a lot

Maby I don't get the idea tho, me very tired

theyangster · Post by **theyangster** » Wed Apr 12, 2006 11:44 am

I think I've got in worked out

the flaps are made out of folder card stock, using standard string taped on the very edge
Since the sting has a twist to it, I've tried to add more twist in an attempt to have the flaps close automatically due to the twist

I have two prototypes that seem to work well in both blow and suck airflow

I'll get a pic up soon

The other problem is since this will be a negative pressure case, I'll probably have to seal up any extra holes

shadestalker · Post by **shadestalker** » Wed Apr 12, 2006 1:08 pm

I can't tell if the flaps on this are opened by air pressure, but this would be nice if it worked: http://doityourself.com/store/7890676.htm

IsaacKuo · Post by **IsaacKuo** » Wed Apr 12, 2006 1:09 pm

That's some pretty radical thinking you've got going on there!

First, I don't see any reason why the rear vent would ever close. Whether one or both of the fans is turned on, negative case pressure will open it up.

Second, and most importantly, is the PSU in the traditional location, just above the CPU? Looks like it. In that case, you're sucking up CPU warmed air into the PSU. This will either cause the PSU to ramp up or you've modded things so that the PSU will simply suffer the temperature rise defenselessly. The former won't be quiet; the latter means risking damage to the PSU.

Third, I don't see a need for the middle one-way vent. With both fans on, there wont be much airflow in either direction across that vent.

Here's my recommendation:

Code: Select all

            ______________________
           |       |              |
exhaust <--|  PSU  |              |
           |_______|              |
           :                      |
 intake -->:   CPU                |
           :                      |
           |-----------          []
 intake -->:   VM101             [] ---> exhaust/intake
           |-----------          []
           |______________________|

It's your layout, minus the partitions. There are a total of two fans--the PSU fan and the front fan.

In "single fan mode", only the PSU fan is running. The negative case pressure sucks air into all three of the openings. This provides cool fresh air for the hard drive but only a little air for the VM101. The PSU sucks up air warmed by the CPU and GPU, but this is cooled by air bypassing the heatsinks from the upper and front intakes. The CPU receives maybe half the airflow; it's a Ninja so I think it'll be okay. Conversely, the temperature of the air entering the PSU has only about half the temperature rise above ambient as the CPU "exhaust".

In "two fan mode", both the PSU fan and the front fan is running, with the front fan running stronger than the PSU fan. Much of the CPU heated air is pulled forward/down to the front exhaust fan. The VM101 gets plenty of airflow, which is also pulled forward to the front exhaust fan. The PSU receives air only slightly heated up by the CPU heatsink. The only loser is the hard drive--it receives air heated by both the CPU and GPU.

theyangster · Post by **theyangster** » Wed Apr 12, 2006 2:02 pm

shadestalker wrote:I can't tell if the flaps on this are opened by air pressure, but this would be nice if it worked: http://doityourself.com/store/7890676.htm

that's what bluefront is talking about, but the plastic flaps are too heavy and it is an blow only

That's some pretty radical thinking you've got going on there!

yup

First, I don't see any reason why the rear vent would ever close. Whether one or both of the fans is turned on, negative case pressure will open it up.

perhaps if the flaps were:
horizontal vs vertical (lll vs =) gravity might come into play
steeper angle l to \ to even ---

Second, and most importantly, is the PSU in the traditional location, just above the CPU? Looks like it. In that case, you're sucking up CPU warmed air into the PSU. This will either cause the PSU to ramp up or you've modded things so that the PSU will simply suffer the temperature rise defenselessly. The former won't be quiet; the latter means risking damage to the PSU.

This was the original plan

but unless I figure out how to successfully mount the seasonic PSU for better airflow (bare PSU), which would involve cutting that I don't want to do for something that may not work, not to mention dangerous capacitors not covered

now a pass thru PSU(such as a Antec HE) might fit the profile better, but I want to work with what I have and cheaply

I've made this defense before, would one expect a seasonic to be able to withstand high temps?
what is the ATX standard?

Third, I don't see a need for the middle one-way vent. With both fans on, there wont be much airflow in either direction across that vent.

It's funner this way

Here's my recommendation:
Code: Select all
            ______________________
           |       |              |
exhaust <--|  PSU  |              |
           |_______|              |
           :                      |
 intake -->:   CPU                |
           :                      |
           |-----------          []
 intake -->:   VM101             [] ---> exhaust/intake
           |-----------          []
           |______________________|
It's your layout, minus the partitions. There are a total of two fans--the PSU fan and the front fan.

In "single fan mode", only the PSU fan is running. The negative case pressure sucks air into all three of the openings. This provides cool fresh air for the hard drive but only a little air for the VM101. The PSU sucks up air warmed by the CPU and GPU, but this is cooled by air bypassing the heatsinks from the upper and front intakes. The CPU receives maybe half the airflow; it's a Ninja so I think it'll be okay. Conversely, the temperature of the air entering the PSU has only about half the temperature rise above ambient as the CPU "exhaust".

rather interesting

I feel that it'd leave to many openings to be effective

the majority of the air would just go thru the CPU, assuming all the opening are the same size
now if the intakes were smaller it might cool better

but now we come to the debate GPU vs PSU? (temp wise)
keep in mind, that I'm a bit of a temp freak, while I'm willing to accept higher temps, when things get too hot

In "two fan mode", both the PSU fan and the front fan is running, with the front fan running stronger than the PSU fan. Much of the CPU heated air is pulled forward/down to the front exhaust fan. The VM101 gets plenty of airflow, which is also pulled forward to the front exhaust fan. The PSU receives air only slightly heated up by the CPU heatsink. The only loser is the hard drive--it receives air heated by both the CPU and GPU.

more interesting, I like it, but if it doesn't work with the one fan concept, then what's the benefit?

Remember my goal is to be mostly operating on one fan with the best cooling efficiency which I think that my design with one fan would cool better

I'm working a third revision

Thanks for the input guys, keep that criticism/praise coming!

theyangster · Post by **theyangster** » Wed Apr 12, 2006 2:06 pm

McBanjo wrote:Isn't there a risk that the flaps will be a noise issue? Paper moving in the winds sounds a lot

Maby I don't get the idea tho, me very tired

sorry did notice your post

the flaps will be made of heavier cardstock (manila folders)
and I don't think air flow will be fast enough to cause any noise, though there will be a small "slap" noise when the flaps close, but that may be better for the peace of mind that it is working

EDIT

here are the prototypes

as you can see they are pretty simple

IsaacKuo · Post by **IsaacKuo** » Wed Apr 12, 2006 5:05 pm

theyangster wrote:I've made this defense before, would one expect a seasonic to be able to withstand high temps?

You're already using this computer, right? Have you done some sort of modification on the PSU fan? If not, then you can tell pretty easily what sort of temperatures it can stand by just closing off the rear case exhaust. With the PSU as the only exhaust, it will receive all of the heat generated by the components just like in your "one fan" plan. Any ducting or fancy one-way valves won't have any effect on the temperature of the air reaching the PSU.

The PSU will tell you when it's feeling uncomfortable with the heat by ramping up its fan.

My first efforts at quiet computing used a Fortron 120mm PSU as the sole exhaust in a single fan system. My experience with a 2.5Ghz Northwood was that under even mild loads, the fan would ramp up unacceptably.

You can perform a simple experiment just by closing off the rear case exhaust. That'll give you the answers you need.

The thought behind my proposal is that maybe 1/2 of the air reaching the PSU will bypass both the CPU heatsink and GPU heatsink. In other words, a large fraction of the air reaching the PSU will be fresh ambient air. This will help keep the PSU from ramping up. It comes at the expense of CPU and GPU temps, though.

I can tell you're thinking that air entering the rear case opening would pass through the CPU heatsink before reaching the PSU. I doubt it, though, since there's a big gap between the case rear and the CPU heatsink. That can be rectified with a small duct, of course.

Remember my goal is to be mostly operating on one fan with the best cooling efficiency which I think that my design with one fan would cool better

IMHO, the best way to get good results with the PSU fan cooling the whole system is to use the PSU as an intake. That way, the PSU receives fresh cool air and doesn't ramp up. The air leaving the PSU isn't terribly warm, so it can be used to cool the rest of the components.

For your case, simply remove all the partitions and flip the PSU fan. You'll have to duct the rear case exhaust to the CPU also. This way, the PSU blows relatively cool air into the case, where it leaves via the CPU and GPU ducts. The air heated by the CPU and GPU get immediately exhausted out the rear of the case. Unfortunately, there's a danger of recirculating warm air back into the PSU as it rises.

For extra cooling, you can have your secondary fan with a one-way valve acting as front intake. This would pull in extra fresh air, where it cools the hard drive along the way to working alongside the PSU to push air out the CPU and GPU ducts.

theyangster · Post by **theyangster** » Wed Apr 12, 2006 5:46 pm

IsaacKuo wrote: You're already using this computer, right? Have you done some sort of modification on the PSU fan? If not, then you can tell pretty easily what sort of temperatures it can stand by just closing off the rear case exhaust. With the PSU as the only exhaust, it will receive all of the heat generated by the components just like in your "one fan" plan. Any ducting or fancy one-way valves won't have any effect on the temperature of the air reaching the PSU.

you obviously haven't seen my other wacky ideas

but yes, I have a very low flow coolermaster UV silent there running as low as it can go

off a fancontroller
I have operated for brief amounts of time (I admit so

) on one fan with the back sealed up
here

The thought behind my proposal is that maybe 1/2 of the air reaching the PSU will bypass both the CPU heatsink and GPU heatsink. In other words, a large fraction of the air reaching the PSU will be fresh ambient air. This will help keep the PSU from ramping up. It comes at the expense of CPU and GPU temps, though.

ideally, yes PSU would/should be the intake, but there is no simple way to do it
I just looked up the ATX spec, the standards are 50C, now it would cut it close to the 40C load of my CPU, keep in mind that this also includes the air from the GPU in my current setup
If my 2 fan system goes into effect, then the PSU will only have to deal with the CPU, which is the whole point

I can tell you're thinking that air entering the rear case opening would pass through the CPU heatsink before reaching the PSU. I doubt it, though, since there's a big gap between the case rear and the CPU heatsink. That can be rectified with a small duct, of course.

yes, I realized that and am working on it
currently thinking of ducting the rear exhaust down and under the NINJA to open up slightly in front of the NINJA, it may be left that way to add fresh air to the mix and/or a valve might limit the air flow until it is really needed

PIC HERE

Remember my goal is to be mostly operating on one fan with the best cooling efficiency which I think that my design with one fan would cool better
IMHO, the best way to get good results with the PSU fan cooling the whole system is to use the PSU as an intake. That way, the PSU receives fresh cool air and doesn't ramp up. The air leaving the PSU isn't terribly warm, so it can be used to cool the rest of the components.

For your case, simply remove all the partitions and flip the PSU fan. You'll have to duct the rear case exhaust to the CPU also. This way, the PSU blows relatively cool air into the case, where it leaves via the CPU and GPU ducts. The air heated by the CPU and GPU get immediately exhausted out the rear of the case. Unfortunately, there's a danger of recirculating warm air back into the PSU as it rises.

if I flip the fan, then the PSU parts furthiest away from the rear exhuast (now intake) may not get the air they need?
also fans at a low voltages may not be as effective blowing down as blowing up

For extra cooling, you can have your secondary fan with a one-way valve acting as front intake. This would pull in extra fresh air, where it cools the hard drive along the way to working alongside the PSU to push air out the CPU and GPU ducts.

don't understand, perhaps one of you code pics could help

And lastly, would you or anyone else advise how to turn on the second fan?
ideally automatically and cheaply, bluefront mentioned here

BrianE · Post by **BrianE** » Wed Apr 12, 2006 5:47 pm

theyangster wrote:Ithe flaps are made out of folder card stock, using standard string taped on the very edge
Since the sting has a twist to it, I've tried to add more twist in an attempt to have the flaps close automatically due to the twist

If string doesn't work out, how about using cut elastic bands? Like the slightly wider type maybe.

I'm interested in seeing how this idea works out.

Flaps and shutters are something that briefly ran through my mind at one time, but I can't recall for what purpose, and I thought making them would be too hard.

theyangster · Post by **theyangster** » Wed Apr 12, 2006 6:04 pm

BrianE wrote:
theyangster wrote:Ithe flaps are made out of folder card stock, using standard string taped on the very edge
Since the sting has a twist to it, I've tried to add more twist in an attempt to have the flaps close automatically due to the twist
If string doesn't work out, how about using cut elastic bands? Like the slightly wider type maybe.

rubber bands? the string is best because it allows the flaps to open/close at a whim depending on direction of the airflow

Sorry, I don't think I said that clearly enough, i meant adding twist to have the flaps swing more closed than open, since due to the extremely senstive nature, they may inadvertly close the other way

I'm interested in seeing how this idea works out. Flaps and shutters are something that briefly ran through my mind at one time, but I can't recall for what purpose, and I thought making them would be too hard.

thanks

I'd say a big problem is keeping everything air tight so no leaks screw up the valves

IsaacKuo · Post by **IsaacKuo** » Wed Apr 12, 2006 7:26 pm

theyangster wrote:
IsaacKuo wrote:Have you done some sort of modification on the PSU fan?
but yes, I have a very low flow coolermaster UV silent there running as low as it can go off a fancontroller

In that case, I would SERIOUSLY consider either flipping the fan or moving the PSU to a nonconventional position in order to keep it fed fresh cool air.

theyangster wrote:
IsaacKuo wrote:IMHO, the best way to get good results with the PSU fan cooling the whole system is to use the PSU as an intake. That way, the PSU receives fresh cool air and doesn't ramp up.
if I flip the fan, then the PSU parts furthiest away from the rear exhuast (now intake) may not get the air they need?
also fans at a low voltages may not be as effective blowing down as blowing up

I find that 120mm fans generate a useful amount of turbulence immediately behind them (on the intake side). A PSU with 120mm fan is so cramped that the fan can't help move air past all components even if it's flipped.

You're right that blowing warm air downward isn't as effective as blowing it upward...theoretically. I haven't been able to produce a detectable difference yet, on the scale of a computer case.

For extra cooling, you can have your secondary fan with a one-way valve acting as front intake. This would pull in extra fresh air, where it cools the hard drive along the way to working alongside the PSU to push air out the CPU and GPU ducts.
don't understand, perhaps one of you code pics could help

This is almost the exact same picture as above:

Code: Select all

            ______________________ 
           |       |              | 
 intake -->|  PSU  |              | 
           |_______|              | 
           :                      | 
exhaust <--:   CPU                | 
           :                      | 
           |-----------          [] 
exhaust <--:   VM101             [] <--- intake 
           |-----------          [] 
           |______________________|

The only difference is that the airflow is reversed. The one-way valve would be right behind the front intake fan.

And lastly, would you or anyone else advise how to turn on the second fan?
ideally automatically and cheaply, bluefront mentioned here

If I were doing it, I'd use a manual switch. I don't like wondering if/when a fan will turn on or ramp up. I'd rather make the conscious decision, "Okay, I'm about to encode a video. I'll just turn up the fan now, and then I'll turn it back down after I'm done."

I personally find the noise less obtrusive when I'm in control of it.

theyangster · Post by **theyangster** » Wed Apr 12, 2006 8:05 pm

PSU location will have to stay, it's been thru much worse

Code: Select all

  
            ______________________
           |       |              |
 intake -->|  PSU  |              |
           |_______|              |
           :                      |
exhaust <--:   CPU                |
           :                      |
           |-----------          []
exhaust <--:   VM101             [] <--- intake
           |-----------          []
           |______________________|

Air would not be able to reach the VM 101 because the actual card in the way

Manual switch, but I'm a forgetful guy

I'll have to see, but it looks like manual is still the way to go

jaganath · Post by **jaganath** » Thu Apr 13, 2006 1:34 am

I personally find the noise less obtrusive when I'm in control of it.

Exactly. Even if something is loud or irritating, if you can turn it off, it is less stressful. So you being in control of the noise is the important bit; this is why jackhammmers digging up the road and aircraft taking off are so aggravating, even if you've got double-glazing- you're not in control of the noise/there's nothing or very little you can do to make it stop.

Bluefront · Post by **Bluefront** » Thu Apr 13, 2006 1:56 am

Funny....I'd rather "set it and forget it". I'd trust a temperature sensor and an electronic switch to turn on a fan at the appropriate time, as opposed to my memory.

Just as I typed this my number two computer ramped up it's single exhaust fan speed.....it's getting hot here in St Louis. Most of my computers adjust the various fans automatic. I like it that way....

IsaacKuo · Post by **IsaacKuo** » Thu Apr 13, 2006 4:31 am

theyangster wrote:

Code: Select all

  
            ______________________
           |       |              |
 intake -->|  PSU  |              |
           |_______|              |
           :                      |
exhaust <--:   CPU                |
           :                      |
           |-----------          []
exhaust <--:   VM101             [] <--- intake
           |-----------          []
           |______________________|

Air would not be able to reach the VM 101 because the actual card in the way

Air is a fluid, and as such is very good at going any which way to get from higher pressure to lower pressure. The flipped PSU fan sets up an overpressure within the case, and air will escape out both of the exhausts (the intake is blocked by the shutters).

theyangster · Post by **theyangster** » Thu Apr 13, 2006 6:34 am

hmmmm, I see

well, i'll try and get things working today

perhaps I do both auto and manual via DDPT switch?

keep in mind that the auto sensor is merely just to turn on the fan, not to control the fan

theyangster · Post by **theyangster** » Thu Apr 13, 2006 5:45 pm

soo, I have the chance to jump for a T-Balancer

should I do it?
considering it's only for two fans?

-theyangster

EDIT BOUGHT for $40 USD
expecting arrival

theyangster · Post by **theyangster** » Tue Apr 18, 2006 6:19 pm

It works!

with one valve...perhaps that's all I'll need

Picture of the overall setup

not quite the final version, need to seal up those little cracks

here the flaps are closed (the fan is spinning) curiously the flaps still close even with the case open

pic of the front (fan is spinning)

here was how I verified it actually worked, thru the tape "window"

pic of the foam divider

Notes

I like it

works well with one valve
with the front fan off, the flaps stay open with a bit of fluttering, but no noise thankfully!

the front fan must overcome the current two fans to close the flaps

the CPU temp fluctuates a lot, which is interesting considering the CPU has no actual intake

I am considering to open the rear exhaust with tin snips to divide it into two: half intake, half exhaust

bottom half would go under the Ninja
a fan in front of the Ninja would blow this air onto the Ninja and out the top half (of the rear exhaust), sparing the PSU the heat

Questions?Comments?Concerns?

theyangster · Post by **theyangster** » Tue Apr 25, 2006 9:02 pm

so, I've run this system for a week now

The valve is working perfectly fine, only mishap was an awry cable that held a flap open

As a test I left the the fancy stuff out and ran it as it was, noting that the Ati 9800 pro got very hot, opened the case, put the fancy stuff back in, turned the front fan on full blast and I immediately felt warm air being taken out

to me this is fine example of what I had in mind

now, I'm still trying to deal with the CPU and PSU issue

I thought of this

-Get a pass thru PSU, been eyeing a Antec NEO HE 430w Duct it that extends just far in front of the Ninja so CPU heat is completely pulled out by the rear exhaust and the Antec is pulling in mostly cool air
-use rear fan
-cut out bottom of the rear exhaust to make an intake (It'll have a "scoop" so it doesn't suck the rear exhaust air)
-make some sort of restricting valve to allow a bit of air to go thru, but once the front fan goes on, *bam* the valve opens up completely

Scary thing
T-balancer sparks and emits a small flame blacked little box thingy still works, but how do you disable the critical temp box?

Come on people, I want some input!

theyangster · Post by **theyangster** » Fri May 12, 2006 3:55 pm

Well, here we go

Would have showed you earlier, but my HD decided to corrupt itself (luckily it was only one bad cluster, so windows fixed it)
In the mean time I install Xp on a samsung MP0402H and wow, that thing is quieter than the barracuda!

anyway

they say pictures speak a thousand words

note- this pic was taken without the middle "valve" but rest assured it is there, nothing changed for that valve

this is the second version of the air valve, each one overlaps each other

side view

using this configuration, the digital sensor from the T-balancer reports 30C (mounted next to the VM101, not on core)

anyway, looks nice eh?