Optimizing case airflow

SpyderCat · Post by **SpyderCat** » Mon May 19, 2003 1:57 pm

AtomicDude512 wrote:
JEN wrote:I think it was the headsink. I now have the slk800 and the temps are stable at 66C @ 100% load 0.5Hr
66c is extreme.

66°C is pretty high, not extreme.

When ASUS mobo's with the Q-fan feature start reducing the voltage fed to the CPU-fan at 60°C I wouldn't call 66°C "extreme".

http://www.asus.com.tw/support/english/ ... q-fan.aspx

Just my opinion...

dukla2000 · Post by **dukla2000** » Mon May 19, 2003 2:53 pm

JEN wrote:I think it was the headsink. I now have the slk800 and the temps are stable at 66C @ 100% load 0.5Hr

Which CPU temp is 66? Have you found the CPU thermister in MBM? (Is this the temp that got to 127 with the Zalman? And do you still have your 'old' CPU temp in the 50s?)

JEN · Post by **JEN** » Tue May 20, 2003 2:28 am

I still can not use the CPU diode option because the temp goes all the way up to about 120C ! and switches the computer off.

How do I stop the computer from switching off, I tried unticking the cpu warning but it still swtiches off.

Anyway, here are some temp readings using the normal sensor, not the diode !

1/2 HR CPU OVERLOAD TEST

Fans @ 7v

AMBIENT TEMP - 21C

IDLE
PSU - 34C
CPU - 47C
SYS - 41C
HDD - 32C

5 MINS @ 100% load
PSU - 36C
CPU - 56C
SYS - 48C
HDD - 31C

10 MINS @ 100% load
PSU - 37
CPU - 58
SYS - 51
HDD - 31

15 MINS @ 100% load
PSU - 37
CPU - 60
SYS - 51
HDD - 31

20 MINS @ 100% load
PSU - 37
CPU - 60
SYS - 51
HDD - 31

25 MINS @ 100% load
PSU - 37
CPU - 60
SYS - 51
HDD - 31

30 MINS @ 100% load
PSU - 37
CPU - 60
SYS - 51
HDD - 31

Fans @ 12v for another 5mins 100% load
PSU - 34
CPU - 50
SYS - 43
HDD - 31

Fans @ 12v for another 5mins 0% load
PSU - 32
CPU - 39
SYS - 38
HDD - 30

The only thing that conserns me now is the SYS temp @ 51 ?

SpyderCat · Post by **SpyderCat** » Tue May 20, 2003 3:15 am

JEN wrote: The only thing that conserns me now is the SYS temp @ 51 ?

I suspect hot air from the CPU-heatsink is being blown on the mobo sensor.

JEN · Post by **JEN** » Tue May 20, 2003 3:41 am

SpyderCat wrote: I suspect hot air from the CPU-heatsink is being blown on the mobo sensor.

What can I do to improve this, without adding noise

I doubt the "Silent Motherboard Heat sink ZM-NB32J" will make a lot of difference!

The only thing I can think of is making a side duct (above the CPU) and turning the CPU fan around so it blows hot air out of the case through the side duct.

As I now have L1A fans @ v7, how much noice will this add?

SpyderCat · Post by **SpyderCat** » Tue May 20, 2003 4:11 am

JEN wrote:
SpyderCat wrote: I suspect hot air from the CPU-heatsink is being blown on the mobo sensor.
What can I do to improve this, without adding noise

I doubt the "Silent Motherboard Heat sink ZM-NB32J" will make a lot of difference!

The only thing I can think of is making a side duct (above the CPU) and turning the CPU fan around so it blows hot air out of the case through the side duct.

As I now have L1A fans @ v7, how much noice will this add?

Hi Jen,

Now you are suggesting something I totally believe in.

Just some remarks:
* A side duct will probably present an easy path for noise to escape from the case.
* Constructing a good airtight duct to the sidepanel will be a pain.
* Avoid loading your CPU-heatsink with additional forces from gravity.
* Ducting to the rear of the case is preferred by me.
* The Badong-duct is probably the easiest way to this, although it isn't exactly cheap.
* The heatsink probably will need some additional shrouding in order to prevent cool air being sucked in. I mean to say: air that hasn't cooled the heatsink.

Think about it some more, and when you are serious about this; Go for it! We will help you!
I will read up on your previous postings to get a better view on your setup.
What does your tool-chest contain?

Regards, Han.

JEN · Post by **JEN** » Tue May 20, 2003 4:57 am

That does my tool-box contain?

1x Garden secateurs
1x Compass
1x Pencil

Also

My case setup has changed slightly:

3x Panaflo L1A (1 in PSU - exhaust, and 1 as case exhaust - just below the PSU, and the last 1 is mounted on the slk800 - blowing on it).

2x fanned back case exhausts as mentioned above, and 2x bottom/front case inlets - fanless.

Chipset heatsink has always been fanless.

GPU - GF4ti4200 has that massive heatpipe on it from Zalman and keeps the temperature at 46C under load while using 3DMark2001.

Cuda V 120Gb - suspended

And finally, the little noise made by the fans at 7v and the hard drive seek has gone because I put the case under the table.

If I turn the monitor and speakers off, there is no difference in noise when the computer is on or off

I have been folding for the last 2 hours and the temps are:
CPU - 53C
SYS - 47C
HDD - 32C

SpyderCat · Post by **SpyderCat** » Tue May 20, 2003 5:09 am

JEN wrote:That does my tool-box contain?

1x Garden secateurs
1x Compass
1x Pencil
..............
If I turn the monitor and speakers off, there is no difference in noise when the computer is on or off

I have been folding for the last 2 hours and the temps are:
CPU - 53C
SYS - 47C
HDD - 32C

Judging by your toolbox, you never felt an urge to mess around with mechanical / electrical things, or you are the sloppy kind that has her tools lying around all over the place

You can't hear the rig, and your temps are ok now.
Why would you do anything to your rig?
If it ain't broken; don't fix it!

Seriously, when you don't have a good time messing around with your rig, there is no reason to do anything to it.
Besides, your folding-production is going to suffer when you start modding.

Regards, Han.

JEN · Post by **JEN** » Tue May 20, 2003 5:41 am

I have to agree with you, my toolbox is not exactly amazing

But, how high can the system chipset temperature go?

its currently folding @ 47C and goes all the way upto 51C with CPUBurn

Post by **MikeC** » Tue May 20, 2003 7:56 am

JEN wrote:But, how high can the system chipset temperature go? its currently folding @ 47C and goes all the way upto 51C with CPUBurn

It's not necessarily the chipset temp, it is often some arbitrary point on the mobo, near the CPU, where a thermal diode/transistor has been placed. You need to check with the mobo maker or manual to see of the location of this sensor is identified, but the bottom line would be: if the system remains stable, it not really anything to be too worried about.

If you really need to know how hot things are getting on the board, just get one of those cheap digital temp devices with flat film temp sensors on wire leads, and try sticking them at various point on the mobo around the CPU. As you know from the "sys" temp drop with fans at 12V, airflow across this area is probably the only thing which will affect it much. Creative ducting could direct some of your existing airflow in the right path, but I doubt it is really worth the effort.

chiahaochang · Post by **chiahaochang** » Tue May 20, 2003 11:05 am

MikeC wrote:Creative ducting could direct some of your existing airflow in the right path, but I doubt it is really worth the effort.

Would it be worth ducting hot air to an exhaust fan? I've just finished building a fast and quiet PC and my system is a little on the warm side. CPU (P4C-3.0Ghz w/CNPS-7000Cu) is ~55C, mobo ~41C under load (folding). I have only 1 80mm Panaflo L1A as exhaust and a SilenX 14dB 400W PSU.

I was planning on making a duct out of cardboard to hopefully lower case temps. How good of a seal does one need of the duct around the heat sources (CPU heatsink, vid card heatsink) to be effective?

dukla2000 · Post by **dukla2000** » Tue May 20, 2003 11:39 am

JEN wrote:But, how high can the system chipset temperature go?

Jen - IIRC that mobo has the system temp sensor as a discrete thermistor somewhere close to the Southbridge. It is most likely it is trapped in hot air from your graphics card, maybe under IDE cables. But either way 47/50 is not too serious: far more likely a local problem to the sensor than a general case ventilation problem.

Your CPU cant be around the 50s if the general air temp in the case is 50 already!

So like Han says, just keep folding!

SpyderCat · Post by **SpyderCat** » Tue May 20, 2003 1:17 pm

chiahaochang wrote: Would it be worth ducting hot air to an exhaust fan?

Always!
Any arrangement to avoid hot air from spreading through the case will help.

chiahaochang wrote: I've just finished building a fast and quiet PC and my system is a little on the warm side. CPU (P4C-3.0Ghz w/CNPS-7000Cu) is ~55C, mobo ~41C under load (folding). I have only 1 80mm Panaflo L1A as exhaust and a SilenX 14dB 400W PSU.

I was planning on making a duct out of cardboard to hopefully lower case temps. How good of a seal does one need of the duct around the heat sources (CPU heatsink, vid card heatsink) to be effective?

Regretfully you have a heatsink that's very difficult to fit a duct on.
I've contemplated to use that heatsink but bought the AX-7 as it was much easier to fit a duct.
Still, it can be done. Are you handy working with with your hands, and willing to take some risk?
Do you have other heatsinks lying around?

Regards, Han.

JEN · Post by **JEN** » Wed May 21, 2003 5:03 am

I am going to try and re-direct some of the cool inlet air towards the sys temp sensor. The thing is, I have no idea where it is or what it looks like

I have taken some pics of the motherboard here!

Can anyone see it?

The pics are high res, but you need a yahoo email to see them at full res, please let me know your email so I can add you to the list of high res viewers!

@ dukla2000

It sounds worrying when you say the CPU temp can not be close to SYS temp

Currently ambient temp is 21C

all 3 fans are @ 12v

and the temps are

CPU - 35C
SYS - 33C
HDD - 29C

If I select the diode reader from MBM, how do I reset the program if it switches off the computer without having to re-install it again?

SpyderCat · Post by **SpyderCat** » Wed May 21, 2003 11:54 am

JEN wrote:I am going to try and re-direct some of the cool inlet air towards the sys temp sensor. The thing is, I have no idea where it is or what it looks like

Hi Jen,

Just admit it: You like fooling around with your rig

Nothing to be ashamed of, your just a bit over the top

You could probably find your sensor this way:

Set MBM to probing once per second.
Stick a piece of garden-hose into the sucktion-end of your vacuumcleaner.
Use the vacuumcleaner on it's low-power setting, and make sure the vacuumcleaner does get some air besides the air coming through the garden-hose, as it's engine needs to be cooled too.

Now slowly roam over the mobo with the garden-hose while watching MBM.
When you find the sensor you wil see the temp. drop FAST.
Make sure the vacuumcleaner does get some air besides the air coming through the garden-hose,
as it's engine needs to be cooled too.

Regards, Han.

Athlon Powers · Post by **Athlon Powers** » Wed May 21, 2003 1:17 pm

Is your case airflow good enough, is it negative or positive? I have two exhaust fans and my system never gets above 47C/32C CPU/Mobo on load.

I have a 56w Palomino CPU BTW, only 4w less than yours I think...

dukla2000 · Post by **dukla2000** » Wed May 21, 2003 1:51 pm

JEN wrote:It sounds worrying when you say the CPU temp can not be close to SYS temp

Probably misphrased: basically short of liquid nitrogen or similar cooling, your CPU temp has to be somewhere about 20C higher than the air temp of the air going into the CPU fan. (depends on heatsink, thermal goop and all kinds, but 20 is a fair ballpark.) So if your CPU (under load) is somewhere in the 50s, the air entering the CPU fan has to be in the 30s. i.e. your sys temp number is not an accurate reflection of all the air inside your case.

I can see the pics OK - but your chances of finding it by eye are slim. Sometimes they are bead thermistors - about the shape of a matchhead, half the size of a matchhead, and sometimes even blue. Then again, they can be nearly anything. SpyderCat's suggestion with a vacuum cleaner is a fair way to find which part of the mobo to try. (As before I would put a small bet on it close(r) to the southbridge, and/or to the Winbond W836(or7)xx Hardware sensor.)

The only thing I can suggest with MBM getting excited about displaying the CPU diode is to ensure all the temp alarms/alerts are disabled, as well as setting the range for the high values to 100. Otherwise will need to get into an MSI forum somewhere and search for the obvious mistake! Also make sure you haven't got any compensation set. (Or maybe even try a -60 compensation to try bring it into a reasonable range while you debug!)

ps - this could well be my bad memory, or vivid imagination (I only had that MSI board for about 2 days before I returned it), but in picture 2, between the bottom right of the southbridge and the fan header there is a miniture 44 gallon drum strapped to the mobo. Try put a finger/fan on that and see what happens to your Sys temp sensor

JEN · Post by **JEN** » Thu May 22, 2003 8:23 am

Cant really use the garden hose pipe idea, as I don't have any spare, and don't want to cut the 1 I use for watering the plants

I think my cases are flow is very good because with all 3 fans @ 12v under folding the temps are:

Ambient room temp - 21C

CPU - 43C
SYS - 39C
HDD - 30C

With the case off for 5 minutes the temps are:

CPU - 41C
SYS - 37C
HDD - 33C

Whats this all about positive and negative airflow???

I think the SYS sensor is close to the Winbond as the temp dropped about 5 degrees when I connected 1 of my old Zalman fans to the motherboard and pointed it in that direction. I also tried pointing it towards the southbridge, but the temperatures returned to normal again. So, its close to the Winbond chip !!!

I remember reading somewhere that airflow takes the shortest route, because of where the SYS sensor is, close to the Winbond chip, I don't think the air flow is passing over it, SO, how do I get some of the air flow over there?

dukla2000 · Post by **dukla2000** » Thu May 22, 2003 9:27 am

JEN wrote:SO, how do I get some of the air flow over there?

Dont bother - all you will do is fix your reported Sys temp, which is a waste of effort because you know it is basically misleading and not representative of your case air. (You could lift a bit of that duck tape nearby on the back panel to allow a slight leak that could help?)

Just keep folding - I still haven't seen your name come up in lights yet!

JEN · Post by **JEN** » Thu May 22, 2003 9:32 am

dukla2000 wrote:Just keep folding - I still haven't seen your name come up in lights yet!

Got a long way to go yet!!! Its just gone past 100/400

I have a question, which will most likely sound dumb, What does this folding prove. I know a lot of threads exists about folding, but I can't seem to find the answer to that question?

LushMD · Post by **LushMD** » Thu May 22, 2003 9:52 am

JEN wrote:I have a question, which will most likely sound dumb, What does this folding prove. I know a lot of threads exists about folding, but I can't seem to find the answer to that question?

The intent of the folding@home project is to examine protein folding (that is, how proteins assume their 3-dimensional structure, which dictates function). Furthermore, considering that the human genome has been sequenced and the human proteome (comprehensive database of human protein sequences/strucutres) is now being pursued, understanding the forces that dictate protein conformational state is becoming increasingly feasible/pertinent. This is a topic that is fairly timely, given the relatively recent outbreaks of prion diseases (which are thought to be caused by misfolded proteins), such as BSE and CJD. Examination of protein folding is also relevant to diseases of amyloid accumulation, such as Alzheimer's and multiple myeloma. More information may be found here. Take care.

JEN · Post by **JEN** » Thu May 22, 2003 10:04 am

LushMD · Post by **LushMD** » Thu May 22, 2003 10:43 am

Actually, I suppose that a better response to your question may be as follows: although it is known that protein structure is determined by the sum total of a variety of interacting forces (i.e. covalent bonding, hydrogen bonding, van der Waal's interactions, etc...) between various components (i.e. side chains, etc...) of a given protein's amino acid constituents, actually calculating the net effect of all of these forces requires considerable computing power (especially when examining larger proteins). This is where folding@home comes in. Of course, another application (the one that the folding@home website emphasizes) is that knowledge acquired from studies of protein folding may also be applied to examining protein misfolding (as occurs in various pathological processes). Take care.

wussboy · Post by **wussboy** » Thu May 22, 2003 12:43 pm

So basically, Stanford University wants to study how a protein folds but they don't have the computer power to do it. So they broke up monumentally huge tasks into tens of thousands of little tasks and sent them to thousands of computers like mine and yours to work on. When we get this all figured out, Stanford will build me that supermodel I ordered a few years ago.

rpc180 · Post by **rpc180** » Fri May 23, 2003 9:07 am

Like 'Weird Science' supermodel or supermodel computer?

Optimizing case airflow

The quest for the case-temp. sensor :lol: