New direction for case air flow

[Snafu]

Did a few test on case air flow that I thought some might find interesting.

Stress testing involved running F@H and Prime95 until temps were stable (1-2 hours).

The fans in the case are 2-80mm up front, 2-80mm in the rear, 2-80mm on the side, 1-80mm at the top and the PSU has two 80mm fans.

Started using MBM5 and Asus Probe to monitor temps. Due to conflicts between the two (alarms and reported CPU temp at 83C!) ended up using only Asus Probe. Since I did not learn this until after the first test the baseline results are somewhat questionable. During one test Asus probe alarm went off when it (mis)reported a CPU temp of 83C .

Baseline:
Full-load: CPU: 47C, MB: 29C (RT: 18C)
Idle: CPU: 25C, MB: 27 (RT:17C)

Here's a pic of the case's innards before anything happened. What a mess.

[Snafu]

I cleaned up the insides of the case. Re-routed the wiring and encased them, closed the PSU vent (clear packing tape), removed one drive bracket (not needed right now) and moved the PCI cards (sounds and modem) to the bottom slots.

Front/rear fans on high and side/top fans on low
Full-load (#1): CPU: 52C, MB: 33C (RT: 19C)
Freezes after a few minutes when CPU hits 50C.

Dropped OC from 235 to 230, vcore from 1.65 to 1.60 and PAT from turbo to auto
Full-load (#2): CPU: 48C, MB: 31C (RT: 19C)

Not sure why temps are higher than the baseline. Can only speculate that either the baseline results were not accurate or the PSU vent was actually helping pull hot air out of the case when the top fan is on low. Can't recreate the baseline results unless I undo all the case clean-up.

One thing to try is to see if uncovering the PSU vent helps.

Front/rear fans on high and side/top fans on high
Full-load: CPU: 47C, MB: 30C (RT: 18C)
Idle: CPU: 25C, MB: 27C (RT: 17C)

Side and top fans seem to help with the cooling but will need to see if opening the PSU vent helps or hinders the side/top fans.

I have since replaced the ribbon cables with rounded cables.

[Snafu]

Reversed the flow of the front and rear fans (rear fans blowing in and front fans blowing out). Front/rear fans on hi and side/top fans on low. Since the front fans are now blowing outward I opened the front door of the case (don't you love the smell of a electronics)

Full-load (#1): CPU: 44C, MB: 34C (RT: 17C)

Closing the front door (restricting outbound air)
Full-load (#2): CPU: 46C, MB: 35C (RT: 17C)

Adding the side/top fans on hi (fron door closed).
Full-load (#3): CPU: 47C, MB: 35C (RT: 18C)

Seems the side/top fans are not helping here but room temp was a little higher.

So far the best set-up for a cooler CPU temp is reversed air flow. The down side is that the MB temps are higher and you get the exhaust in the face (don't you love the smell of ozone in the morning ).

Not sure if MB temps are to be concerned about. Lower CPU temps are nice.

The next things on the to do list...
- see if I can get back up to 235 stable with the cooler temps
- open PSU vent to see if it has any impact
- use a drumel to remove the vent grills
- add a CPU cooler
- add video card cooler
- heat sink and fan set-up to cool the NB (if it will fit under the CPU cooler)

[Paragon]

This is a great thread Snafu with a huge amount of potential for all our users, for that reason: I've made it STICKY keep up the good work bud

[Snafu]

Thanks for the compliment.

It would be greta if someone could validate the temp drop when you reverse the case flow. Unfortuantely I do not have anything elaborate to see if the changes affect the ram temps, NB temps or anything else (won't know if they are affected until eitehr I get a higher FSB that's stable or something goes pop ).

Will keep posting (& folding)

P.S. where is everyone - off on vacation or having babies?

[flexkill]

hey man...nice job,pc looks good too.

[Snafu]

Thanks . Wish I could download clearer pics but limited to 50k & I do not have a webpage where I can post (gotta change that).

[Snafu]

Running back at 235 (vcore@1.60). Temps are a little higher but seems stable (no freezing).

Full-load (#1): CPU: 52C, MB: 38C (RT: 18C).

So it looks like increasing the FSB by 5MHz causes a 7C increase in temp for this CPU. Surprised the temp went up this much. This CPU could be temperature sensitive and maybe this is why it is not stable over 235.

Interesting.

[Ice]

Wow, great findings, Snafu. Keep us updated on your progress!

[Snafu]

Wow, great findings, Snafu. Keep us updated on your progress!

Thanks Ice. Here's the latest from last night.

Sad to note that it froze at 5am this morning . There was no indication of any issues (temp, volts fan speed) during the run as all recorded and normal using Asus probe up until 5am (>7.5 hours okay then nada).

Under full load CPU: 52C MB: 38C RT: 18C

Might be some of the bios settings. I will try to up the vcore to 1.625 (and then 1.65) - droops to ~1.52 under load. PAM is set to auto.

Maybe I should post this in another thread but thought I would ask about a couple of bios settings:

Q-fan: enable or disable (this is to let temps control the cpu fan speed right and disabled means the cpu fan runs full tilt, right?)

Ram timings: I have them set at 3-2-2-5 (8). These 3700EB sticks (1GB kit) are labelled with 3-3-2-8. I can reset the timings no prob to see if I can get stability back. I wanted to check which timings are tighter (fyi - cannot go below cas 3. Tried 2.5 and received a bunch of error messages).

Thanks

[Snafu]

Ram timings: I have them set at 3-2-2-5 (8). These 3700EB sticks (1GB kit) are labelled with 3-3-2-8.

Must have forgot to memtest the timings. tested and found errors everywhere. Reset the timing to 3-3-2-8 and no issue in memtest.


Now running Prime95 and F@H to confirm stability and temps.

Humbly

[RotorHead]

Snafu, great thread!

I have been working some airflow issues my self, mostly to try and quiet things down a bit.

It’s great to see other folks doing a little airflow testing on their systems. To many times I’ve seen people concentrate on the heatsinks, believing that the factory setup for the case fans is good.

I just picked up a couple of 120mm L1A’s, which are quieter and have a lower CFM rating then the NMB fans that came stock with my case. After swapping them out I noticed about 3c difference in my CPU load temps.

Configuration: http://members.cox.net/ebecquet/rearshot.jpg

120mm L1A intake front, 2 x 120mm NMB exhaust rear
Load Temps 52-53c on both CPU’s

120mm L1A intake front, 2 x 120mm L1A exhaust rear
Load Temps 49-50c on both CPU’s

There’s only one thing I can think of that would cause this drop in temp with the lower CFM fans: Circulation within the case and around the processors has been changed because of lower airflow.

Next on my list is to install a fan on the top configured as an intake, and tape over the stock vents on the side of the case.

[Snafu]

Thanks! I would like to see alternative approaches especially if you can rig (McGiver) something up from common materials at a hardware store ( I am trying to make a fan shroud out of 3-inch PVC flange used toilets. Not sure if it will work but for a couple of bucks and some time it might actually work). IMO I think these are the coolest mods (I like the fan trick posted by Jim ).

I have a number of ideas and components to install on my to do list (it grows faster than what I get accomplished ).

Although I am no aeronautical engineer, I think there is another aerodynamic factor that is not openly provided in the fans specs. This is the pattern of airflow that comes out of the fan itself. I noticed that the 80mm fans that came with the Thermaltake case tend to blow air in a cone shape away from the fan. This is what got me thinking about a fan shroud (using the toilet PVC flange) to try and get a more direct airflow.

I tried a 120mm Vantec Stealth fan (using a spare connection on the MB) and it seemed to flow more directly (more cylindrical pattern) away from the fan. This could be due to the size difference but it is more likely due to blade shape and the shape of the ventrii that the fan sits in. I think a shroud will help the Thermaltake fans.

Because the stealth fans seem to blow more directly outward I bought four of its baby brother the 80mm with the intent to replace the Thermaltake fans (I have yet to test them out to confirm that they do blow more directly outwards). Although the stealth fans have lower CFM ratings than the Thermaltake fans (27cfm vs 32cfm) I think the air is more direct and therefore moves through the case better (fyi - I assume the Tt fans with my case are the standard 32cfm Tt versions - I checked the website for model number and mine are different than those posted by Tt)

You might be seeing the results of this in you case (although I do not think you have a cooling issue if the sides are left open )

I think the best set-up is with 120mm fans (they can move more air at lower speeds = good cooling and quieter). My next case will take this into consideration.

With your case, is that another drive that is hiding below the PSU? Do you have a fan on the drives? I heard that they are more susceptible to heat than other components. Mind you if you have 2-120mm exhaust and 1-120mm intake the negative pressure should pull air in from all sources and probably past your drives.

Not sure if an intake at teh top will help. You can always reverse the fan and make it a blow-hole if it does not work.

[Snafu]

Just a quick update. Started opening the fan grills to see if there is any improvements. I removed most of the grill. After installing it I noticed that the air flow from the fan was greatly improved.

Will post a pic or two when I get the film back (I gotta get a digital camera one day). I need to paint it a bit since my hatchet job with the dremel .

The temps while running with FSB@230 and 2 F@H:
CPU: 49C MB: 40C RT: 21C (almost time for the AC)
Front and rear fans on high and air flow reversed (from rear to front). Side and top fans on low.

I think the RT is having an affect on the CPU fan speed. In earlier runs it was at 2500rpm even under full load. It is now spinning at 3300 (annoyingly). It might also be due to the the PSU fan as I turned it down to half speed (earlier tests were on high). When I cranked the PSU fans back up to full, the CPU fan speed started to drop. Coincidentally one of the F@H runs just finished and temps/fan speeds started to drop. I will have to look into this a bit more.

I tried lowering the front a rear fan speed. Almost instantly the CPU fan would crank to an annoying level. It looks like the reverse air flow across the CPU is definitely having a direct effect on CPU cooling.

[Snafu]

Here's a pic of the vent that I opened up. It flows far more than the oem vent. I will probably see the biggest effect once I get all 4 front and rear vents open. Right now the temps while running with FSB@230 and 2 F@H, front and rear fans on high and air flow reversed (from rear to front). Side and top fans on low:

CPU: 44C / MB: 33C / RT: 17C (RT is helping out)
CPU fan: 2600

Added CPU fan speed as it is very sensitive to temps and would ratchet up to annoying levels if the air flow is too low (anything above 3000 is very annoying )

Tried upping the oc for a short bit but was not stable after 7.5 hours. Rather than tryin to find the max completely stable oc I decided to back down to 230 so I can fold 24/7 without worrying about freeze-ups. Will try again later and play with some settings in the bios.

The challenge with opening up the vents will be the rears as they cannot be removed from the case. So I will probably have to cover everything up very carefully and hack away. Will need to add grills for finger protection and to hold filters (any precautionary advice ).

ALmost lost some flesh reinstalling the fan. The rig was hibernating when I started to install the fan and it decided to wake up as I was screwing teh fan to the vent. It was funny trying to juggle a spinning fan but luckily no flesh or blood loss