Optimizing case airflow

Enclosures and acoustic damping to help quiet them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Devonavar

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Wed Apr 30, 2003 1:47 am

I havn't received a reply from Rusty yet. Does he not use this forum anymore?
looks pretty good, I mean the duct tape isnt exactly attractive, but if it works
I have the case's back hidden behind the monitor, so no one can see the back :D

[edit]

I also cleaned up the cables a bit and adjusted the CPU fan. Here are the results after 1 hr of 100% CPU usage:

Old temps - CPU - 67, SYS - 43, HDD - 35 (room temp 23C)

New temps - CPU - 64, SYS - 43, HDD - 30 (room temp 23C)

If I spend a little more time with the cables, I am sure the temps will drop further

[/edit]

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Wed Apr 30, 2003 4:41 am

Temperature of air 1cm from PSU fan - 40 Degrees C
Temperature of air from floor case intake - 23 Degrees C
Jen - in general your temps are OK, but only that. In my opinion the 40C 'intake to the CPU fan' is abnormal.

On the one hand it figures that your CPU runs in the 60s under load (the CPU running 20 something degrees above the air being blown on its heatsink is absolutely 'in spec'). On the other, I am confused why the CPU temp was not MUCH better with the case side off. I figured your results from that test were just consistently 'biased because of how you measure', but I presume now the 40C was measured with some other thermometer? In theory if the CPU fan intake was 23C (= room temp) I would expect the CPU to be in the 40s under load :? Something is wrong.

Personally I figure you aren't letting enough room air (@ 23C) INTO the case. You have 2 fans trying to extract (psu & rear case) but they can only exhaust as much as can get in. To keep them as efficient as possible you need the equivalent of 2 * 80mm fan holes for intake. You have only 1 (your nice new bottom intake). I would recommend:
- remove tape from bottom 'slot' (front/bottom plastic) and use a Stanley knife to get that slot as big as possible.
- Cut out the front tinwork (in front of your drive, behind the USB plastic)
- Probably increase the size of your existing hole in the bottom of the case, to say 100mm diameter.

And yeah, I was away for a few days after I stirred the Barracuda debate: personally I cannot hear my 7200.7 (suspended from rubber bands).

Last - Rusty is around, mostly hanging round the F@H forum methinks.

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

Post by SpyderCat » Wed Apr 30, 2003 5:04 am

dukla2000 wrote:
Temperature of air 1cm from PSU fan - 40 Degrees C
Temperature of air from floor case intake - 23 Degrees C
Jen - in general your temps are OK, but only that. In my opinion the 40C 'intake to the CPU fan' is abnormal.
Hi Dukla,

Why are you talking about CPU-fan when the quote is talking about PSU-fan ?
Is the rest of your text still valid ?

I read the quote as a temperature taken at the outside of the case, 1 cm behind the PSU-fan. :roll:

Regards, Han.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Wed Apr 30, 2003 6:24 am

SpyderCat wrote:Why are you talking about CPU-fan when the quote is talking about PSU-fan ?
uh - cos i cant read :oops: (At least now I know what is wrong!! Note to self, must avoid these 4am starts! Thanks Han for the gentle nudge.)

No - the rest of my post is less relevant: the original result that taking the side of made little difference stands as valid, and thus airflow through the case is on the good side of normal.

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Sat May 10, 2003 1:25 am

I have the MSI KT3 Ultra(MS-6380E) motherboard. Will I have problems getting the slk800 heatsink to fit on it?

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Sat May 10, 2003 3:17 am

Jen - theoretically the sink sticks out towards the ATX power connector & DIMM slots but I am 99% sure this will not be a problem.

As far as I can see the only problem will be actually mounting it. The lugs on the KT3 that you need to locate last with the SLK-800A are above the socket cam, at the top of the mobo. On the assumption your psu is there, you probably have to mount the SLK on the mobo outside the case. (Avoid the 800U or 900U BTW - the fins go the other way and could bash your psu.)

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Mon May 12, 2003 7:44 am

My computer seems to be having a bit of a stability problem. It works perfectly with normal use, but when I use the CPUBurn program, it restarts when the CPU temp gets to 62C. Why does this happen?

I think it might be the flower heatsink, but I will find out for sure when I get the slk800 :)

Kostik
*Lifetime Patron*
Posts: 391
Joined: Wed Apr 16, 2003 7:51 am
Location: Paris, France

Post by Kostik » Mon May 12, 2003 7:50 am

Jen : you probably have you BIOS set up to reboot the computer when the CPU reaches a certain temperature. An overheating CPU will usually freeze the computer, not reboot it (and I wouldn't call 62° "overheating"). You can change this setting in the BIOS setup.

Jan Kivar
Friend of SPCR
Posts: 1310
Joined: Mon Apr 28, 2003 4:37 am
Location: Finland

Post by Jan Kivar » Mon May 12, 2003 7:53 am

Kostik wrote:Jen : you probably have you BIOS set up to reboot the computer when the CPU reaches a certain temperature. An overheating CPU will usually freeze the computer, not reboot it (and I wouldn't call 62° "overheating"). You can change this setting in the BIOS setup.
I think that BIOS would rather shutdown the PC than reboot, if the user-set temp limit has been exceeded.

Jan

Kostik
*Lifetime Patron*
Posts: 391
Joined: Wed Apr 16, 2003 7:51 am
Location: Paris, France

Post by Kostik » Mon May 12, 2003 8:47 am

Jan Kivar wrote:I think that BIOS would rather shutdown the PC than reboot, if the user-set temp limit has been exceeded.
Jan
Thinking about it, you are right :)

Jen : you might also be actually experiencing stability problems, Windows XP will reboot the computer when an unrecoverable error occurs. But I find it strange that your processor crashes at only 62°c.

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Mon May 12, 2003 9:41 am

I don't see why I would be experiencing stability problems because about a week a go, I was able to run the CPU at 100% load for half an hour with the temperature stable at 67C.

Also I have Windows 2000 and when the computer restarts, I get no error message, no blue screen, no nothing. Its asks just the way it would if you pressed the restart button on the case yourself ???

fmah
Friend of SPCR
Posts: 399
Joined: Fri Mar 28, 2003 9:32 pm
Location: San Diego, CA

Post by fmah » Mon May 12, 2003 5:50 pm

I got this with bad RAM, but I also wondered if it was the power supply being too weak (perhaps).

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Tue May 13, 2003 4:59 am

300W PSU powering

1x hdd
2x odd
1x fdd
3x fans
1x gpu
1x cpu
1x monitor

whats it ???

I still think its the flower heatsink which has probably lost its pressure on the cpu because of its badly designed clip at its weight !!!

cameron
Posts: 7
Joined: Tue May 06, 2003 1:52 am
Location: Perth, Western Australia
Contact:

Post by cameron » Tue May 13, 2003 5:43 am

300W should be more than enough for that - I have a generic nameless 250W supply powering a similar-ish system.

I've seen systems become unstable due to dodgy RAM, poor CPU cooling, or bizarre hardware/driver issues. I'd be surprised if it's the CPU heatsink, as your temps are fine and I have a similarly chunky heatsink which has been hanging off a delicate-looking clip for well over a year; the system is perfectly stable.

You can check your memory easily enough, though: grab a copy of memtest86 (http://www.memtest86.com/) and see what it thinks. It'll probably take a few hours to check depending on your CPU speed and how much RAM you have, but it's more thorough than programs than run on top of a normal operating system (it's a bit hard to test all of your RAM when Windows wants to keep a hundred-odd megs to itself :-p).

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Tue May 13, 2003 8:27 am

Can you get me a direct link to memtest86 as I cant seem to download it from anywhere ???

I have AMD XP1900+ and 512Mb PC333 (2700) RAM !

aphonos
*Lifetime Patron*
Posts: 954
Joined: Thu Jan 30, 2003 1:28 pm
Location: Tennessee, USA

Post by aphonos » Tue May 13, 2003 8:42 am

JEN wrote:Can you get me a direct link to memtest86 as I cant seem to download it from anywhere ???
It looks like the memtest86 site is down right now (May 13, 2003 12:43pm GMT+4/EDT, USA). May want to try again later.

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Sat May 17, 2003 12:31 am

I tried out the memtest over the night for 9.5 hours. It passed 23 tests with no errors! Is that good enough?

How many tests are there altogether in standard mode?

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Sat May 17, 2003 5:59 am

It runs in a loop (IIRC), so after 3 or 4 you are unlikely to find anything new. BTW, I believe GoldMem is marginally better, but seems unlikely to be a mem problem if Memtest is happy.

Going back to the original problem: in normal usage a PC is minimally stressed. Stuff like CPUBurn severly stresses parts of your PC other sw doesnt reach, and can uncover latent problems. Prime suspects are any components running out of spec - overclocked CPU, memory, overloaded psu etc. I would also agree your 300W psu should be idling, but fire up MBM and keep an eye on your voltages, even good psus can go bad. But could also be a badly designed component that is running in spec but still cant cope for which the only cure would be to avoid that specific stress until you change the environment or that component!

The reboot at 62 - is that 'rock solid' repeatable? (watching temp in MBM, 60 OK, 61 OK, 62 bang?) Could be BIOS (as before, but I haven't seen you comment if/not possible in your BIOS) or MBM or other software (does CPUBurn itself have this option?).

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Sat May 17, 2003 6:40 am

I tried it again, but this time it rebooted at 61C ???

I have VCool running which I set to shutdown at 70C, I set the MB to shutdown at 75C, I also have SpeedFan. I dunno if that has shutdown option. I dont think it is any of them because the computer never shuts down, it restarts ???

I also monitored the voltages with speedfan while CPUBurn was running

Before CPUBrun

VCORE - 1.74V
+3.3V - 3.25V
+5V - 4.92V
+12V - 12.65V
-12V - -12.28V
-5V - -5.35V
+5VSB - 5.46V
VBAT - 3.36

After starting CPUBurn

VCORE - 1.71V
+3.3V - 3.25V
+5V - 4.82V
+12V - 13.01V
-12V - -12.77V
-5V - -5.55V
+5VSB - 5.48V
VBAT - 3.34

small differences only !!!

I have not overclocked anything!

What is MBM

fmah
Friend of SPCR
Posts: 399
Joined: Fri Mar 28, 2003 9:32 pm
Location: San Diego, CA

Post by fmah » Sat May 17, 2003 8:24 am

Wow that +12V is pretty high, you get +13V? I don't know that that should affect anything, but that may be rare.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Sat May 17, 2003 8:49 am

MBM is Motherboard Monitor. Gives you a better set of voltage/temp monitor tools than Speedfan, averages, high/low, you can control a bit more etc.

Speedfan wont actually control fans on that mobo I hope (or at least that is why I sent my MSIs back!). Also neither Speedfan nor VCool actually read the on die thermistor (or at least they dont on the ASUS I replaced my MSI with), whereas MBM does.

The voltages are OK - the high 12V is actually out of ATX spec (should be +-5%, 12.6V max) but should be harmless: the cheapo psu in my sig is running 13.1V but I figure it is only my Seagate motor at risk and that is rated +-10% so if I get over 13.2 I'll start worrying. What is more of a problem is how wildly they are fluctuating. (You can set MBM on a 1 second readout to catch that.)

Personally I've ditched VCool - you should be Folding for SPCR anyway :)

Now, the root cause problem is still up for grabs: the 61/62 difference could be because those are socket temps and it is actually happening at 75 core temp reliably. Hopefully MBM can help analyse that. If it is, then (IIRC you have a flower?), that Zalman has to go or at least you have to crank the fan up a bit on it.

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Sat May 17, 2003 12:02 pm

I got an error message while trying the install MBM!

Message:

An I/O error occurred while installing a file. This is normally caused by bad installation media or a corrupt installation file.

How do I fix this?

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Sat May 17, 2003 12:50 pm

Try another download? Or try again? My file is mbm5320.exe, right click & properties it is 1,783,511 bytes. (Think it is the same version for Win98, me, and 2K?)

Maybe disk problems? Try a scandisk first while I clutch at straws!

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Sun May 18, 2003 12:39 pm

I got MBM to work.

I think I have it set up correctly !

This pic shows the values while the CPU was idle, and this pic shows the values while the CPU was under 100% load.

The values look almost identical to the values I got with VCool and Speedfan.

This probably means I didn't set it up correctly ???

I didn't display the fans because I have 3x panaflos.

I also downloaded the software for folding, entered the correct number for this team and then nothing!!!!

I checked the members list, and my name wasn't there ???

What is supposed to happen ???

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Sun May 18, 2003 1:33 pm

JEN wrote:I got MBM to work.

I think I have it set up correctly !

This pic shows the values while the CPU was idle, and this pic shows the values while the CPU was under 100% load.

The values look almost identical to the values I got with VCool and Speedfan.

This probably means I didn't set it up correctly ???

I didn't display the fans because I have 3x panaflos.

I also downloaded the software for folding, entered the correct number for this team and then nothing!!!!

I checked the members list, and my name wasn't there ???

What is supposed to happen ???
Chances are you do have it set up right. There's usually only one temp feed per item off any mobo and all the various software taps into the same feeds. If you see wee differences, it's just small timing/display differences. Anything less than 2C you can pretty much ignore.

Your member info won't show up until your first work units are posted to folding; your system ought to be running 100% CPU now with folding on all the time. Watch your temps. It won't go as high as with CPUBurn, but folding does kick in any time free CPU cycles are available.

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Sun May 18, 2003 1:44 pm

Your MBM setup looks mostly OK.

What you want to try find is the CPU thermal diode. On the Settings screen, Temperatures, you want to look at all the options under "Should display board sensor". (Often they are duplicated - that MSI has a Winbond controller but you might also have some LMxx detected by MBM.) Basically if you attach each available "Board sensor" to a different MBM sensor, then my favourite view is actually the High/Low tab. You have a feeling for a CPU temp in the 50s, you are looking for another temp of similar order. Actually, if you run idle for 10 mins all the temps should settle down. Then fire up Folding: the CPU diode should rise a lot faster than the CPU socket temp - you may need to set the MBM, General, Basic tab, Interval to 1 second to see this. Then hopefully you will be able to label a temp for your CPU diode (and another for the socket).

And welcome to the SPCR Folding effort. Did you download/instal the Graphical Client? It should have downloaded some work for itself and be sitting at 100% CPU load for somewhere between 12 & 24 hours. Then it will upload the results back to Stanford, then your name will hit the boards! Check through the Folding Forum here for as much background as you can cope with: actually I would recommend you swap to the 'no-nonsense' console and Electron Microscope in a day or so when you get the hang of it.

And BTW - Folding loads your CPU a bit less than CPUBurn: going back to your 'reboots' you may need to fire CPUBurn to really load things up. In general the Folding client checkpoints itself regulalrly to disk and restarts where it was, so is reasonably robust from that point of view!

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Mon May 19, 2003 1:09 am

I selected CPU diode and the temp shot up to 127C and the PC switched off !!!

I had to start up in safe mode to remove MBM from the start folder.

This can't be very good, can it ?

dukla2000
*Lifetime Patron*
Posts: 1465
Joined: Sun Mar 09, 2003 12:27 pm
Location: Reading.England.EU

Post by dukla2000 » Mon May 19, 2003 1:27 am

Not good, but it may be closer to reality than the 60s you thought you had! Would be hard to believe 127C is accurate, but could be worth checking that flower seating before proceeding further!

If you power off for 10 mins (or a bit longer, to try get everything nice and cool, and/or crank up all your fans), then fire up and fire MBM, then sit and watch the High/Low screen (at idle) till it stabilises (and clear all the aggravating/erroneous alarms as you get a feeling for which sensors are really meaningful!).

Then fire up CPUBurn or Folding, and again get familiar with what is what.

JEN
Posts: 212
Joined: Mon Apr 21, 2003 12:45 pm

Post by JEN » Mon May 19, 2003 1:42 pm

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

Athlon Powers
Posts: 376
Joined: Sun May 04, 2003 4:33 pm
Location: Athlonville, My Computer
Contact:

Post by Athlon Powers » Mon May 19, 2003 1:46 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
66c is extreme.

Post Reply