Rear case fan intake duct? See message...

TMonte · Post by **TMonte** » Sat Jan 21, 2006 3:20 pm

Hi,

I've recently modded my old Lian Li PC-6000 quite a bit in order to improve cooling capabilities and reduce noise. One of the things I also just did was assemble and install a water cooling system.

I'm using an ASUS A8N-SLI Premium, which includes ASUS' heatpipe setup for the NF4 MCP. The heatsink at the top of the pipe that's also attached to the power FETs requires constant air flow or both components just feed into each other and the heatpipe mechanism is effectively stifled.

One of the mods I did was to open the back of the case up to allow the use of a 120mm fan (stock was 80mm). Having it simply exhaust air does nothing to help the situation, so I've decided to make a duct to either draw or feed air through said heatsink. Does anyone think it would be more effective pushing air in? If so, can the power supply (Antec TPII 480) handle all exhaust duties? I realize this will create a positive pressure situation, which I don't mind at all.

What I'm doing temporarily is using an old Zalman bracket with a 92mm Nexus fan to blow air at the heatsink. Using an Omega digital thermometer and the highly effective (:lol:) method of holding a thermocouple against the NF4 block, I've measured a full 10C drop with the Nexus (48C+ down to ~38C).

See pics...

Now:

Duct pattern:

Full setup:

In the third pic you can see the whole deal. If anyone is interested in full details of the mods, I can post another thread...but yes, that is a Raptor suspended by O-rings in the stock drive cage. It was moved back .5" for radiator and intake fan clearance so I could keep the drive in the intake stream.

Thanks in advance...I appreciate any advice or comments!

Tom

TMonte · Post by **TMonte** » Sun Jan 22, 2006 10:21 am

Thanks for all the responses!

Just kidding!!!

Anyway, I finished up the duct this morning. I set it up as an intake to direct air over the "PWM"/Power MOSFETs & MCP heatpipe heatsink. (there's a mouthful)

Here are the final results. I'll estimate accuracy at +/- 1C based on holding the tip of thermocouple to the MCP block with my hand 12" down the thermocouple wire.

No airflow: 48.3C
92mm Nexus @ ~1480RPM in free air on Zalman bracket: 38.2C
120mm fan @ ~1050RPM mounted to case and ducted: 36.5C

So there's a full 10C drop from no fan to the Nexus 92 at full speed, and then another ~2C drop with the 120 ducted into the case.

The 120mm fan is a Mad Dog "Whisper Fan" (CompUSA). It's 25mm sleeve bearing model. It runs 1850RPM at full speed, but I'm using SpeedFan to run it at 45%, which nets 1054RPM. The fan has a bit of bearing noise that disappears below ~1200RPM, so I'm going to leave it there. I have it soft mounted with Nexus "long" rubber fan mounts, as the fan mounting holes are the full width of the frame. For reference, full speed puts the measured temperature another 2C lower (34.4C).

The duct is spaced about .5" from the heatsink, and 'overlaps' by about .375"...of course that's just eyeballing it.

I think I'll stick with this unless someone thinks it would be better set up for exhaust. Whaddaya say?

By the way, the rig is running an A64 San Diego @ 2700MHz at ~1.57v (10x270), with 1GB PDP PC3200 at DDR450 speeds (~225MHz. Timed at 2.5-3-3-7-1T). It also has a 7800GT at 490/1220. The CPU block is a Swiftech Storm (aka Stew Forster's G4 design) and the GPU block is a DangerDen Maze4. Pump is a Laing D5 at setting 4 (seems to be the best balance of subjective noise vs. flow), radiator is a BlackIce Pro X-flow (single pass) with a Panaflow at 7v, and the tubing is 1/2"-3/4" Clearflex. CPU idle temps are 30C, peak at 37C (running Prime95 for ~12 hours). GPU idles at 35C and peaks at 43C (rthdribl to generate a heat load). The GPU idled at 42C using a VF-700Cu at 5v, so this is very good IMHO. I realize I could get even better cooling performance with a Thermochill PA160 or a double 120 radiator, but I'm very satisfied with the results.

Edit: forgot to mention ambient temp of 22C.

More Pics:

Later,
Tom

Weldingheart · Post by **Weldingheart** » Sun Jan 22, 2006 5:47 pm

Thanks for sharing the mods Tmonte, how do you attach the duct to the fan?it seems just only being clamped by the foam.
Aluminium shroud near electrical components is not really safe in my experience

Nice temp even with single BIP.
Just curious,why do you measure the NB temp with thermocouple instead of just with onboard Asus utility?is the onboard monitoring is not trushworthy enough? I find 3-12C difference between internal and external monitoring,really depend on the location where I attach the probe, every about 0.1mm(no,didn't really measure every single 0.1) give difference on thermometer reading..

TMonte · Post by **TMonte** » Sun Jan 22, 2006 6:22 pm

Hi,

Thanks for the comments. Yes, the duct is only held to the fan by the foam. Believe it or not, it's quite a tight fit and seems very stable. Put it this way, it won't fall off. I plan on checking it every few months to make sure the foam doesn't deteriorate. The bottom line is that I wanted it isolated from the fan to prevent vibration transmission (even thought the fan is spinning relatively slowly), and this was the best idea I could come up with.

To be honest, I didn't know it could be measured onboard. I don't have the Asus utility installed, and I'm using Speedfan. I'd bet it's the temp showing 36C though.

Do you know where I could find each documented so they can be labeled properly in SpeedFan? I suppose I could just install the Asus tool.

Thanks!!

Tom

TMonte · Post by **TMonte** » Sun Jan 22, 2006 6:56 pm

Here's what I have in SpeedFan, and then in Asus PC Probe II, respectively:

Obviously Temp1 and Temp2 in SpeedFan are the CPU and "Motherboard" readings. Is Temp2/Motherboard possibly the NF4 chip reading? Don't know what Temp3 is, HD0 is self explanatory and the second Temp1 is coming from ACPI according to SpeedFan's temp config page. Is that possibly the PWM/power circuitry temp?

Thanks again,
Tom

TMonte · Post by **TMonte** » Thu Jan 26, 2006 8:03 pm

Update. I reversed the fan to exhaust. The "motherboard" temp is 1C lower, but the NF4 MCP block went up 2C ( 38 ). The fan also struggles at 12v during boot...a lot of restriction there.

What I need to do is finish off the side panel and the bezel, button it up and then test again. I'm still concerned that having it intake air, while a bit more quiet and seems better for the MCP temp (with the side cover off), is going to adversely affect the internal case temp.

Thanks,
Tom

epiphane · Post by **epiphane** » Fri Feb 10, 2006 9:20 am

just came across u'r duct...nice job in planning/building it

Cheers
Peter

Aris · Post by **Aris** » Fri Feb 10, 2006 9:51 am

all that work for cooling a northbridge....

anyhow, whatever floats your boat.

as far as direction goes, i'd say have it exhaust. while your northbridge temps may go up slightly, your entire system temps will drop. your PSU temps will also drop which means that if you have a PSU that varies its fans speed based on its temp that the fan speed will stay at a lower rpm at higher loads, lowering overall system noise.

another suggestion i would make is to drop the front intake fan all together. all it really does is "direct airflow" over the components immediately in front of it, it doesnt actually increase overall case airflow at all. its not really clear from the pictures, but it looks like you MAY have your radiator at that location. If you do you would probably see an increase in cpu/gpu temps, but system temps will remain the same, and noise will deminish quite a bit since you wont have a fan sitting right at the front of the case which is usually located fairly close to where you sit. you'll still get cool airflow through the radiator without a fan on it because the 2x 120mm exhaust fans will be pulling air from the largest opening in the case, which appears to be where the radiator is. so it will still keep the cpu/gpu cool. i'd say at the very least experiment with dropping the front fan and see what happens after you make both the rear fans exhaust air. you might be pleasently suprised at the noise drop and sustained cooling ability.

TMonte · Post by **TMonte** » Fri Feb 10, 2006 11:59 am

Thanks for the suggestion. I had been kicking that idea around (dropping the front intake fan) and as of late have been running the rear fan as exhaust without the duct. There doesn't seem to be any impact to overall system stability, and at the time I thought it may have been impacting my overclocking attempts. I'm not bent on oc'ing it to the hilt, so I may just drop back to stock clocks and enable CnQ again.

Tom