Experiments in CPU cooling in a P180

Got a shopping cart of parts that you want opinions on? Get advice from members on your planned or existing system (or upgrade).

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
rhys j
Posts: 31
Joined: Thu Oct 15, 2009 6:46 pm
Location: England

Experiments in CPU cooling in a P180

Post by rhys j » Tue Oct 20, 2009 3:11 pm

Hi all, I'm new here so not sure if this is the right forum, but thought you might be interested in what I found while playing around with my fan configs under stress testing.

Skip to the Conclusions section if you don't want to read it all.

System:

Image

Antec P180
2x Fractal Design 120mm 1000 rpm fans
Akasa Fan Controller Pro
Asus P5KPL-AM/PS Mini ATX motherboard
Intel Core2Duo E7400 (not overclocked for this test)
Nexus XIR-2300 CPU cooler
G.Skill 4Gb DDR2-PC8000
Samsung SpinPoint F3 1TB hard disk (in lower chamber)
Asus GTX 260 SP16 graphics card
Jeantech 500W PSU


PC configuration
-------------------

It's my first PC build. I bought the case second hand and it came with some Hiper / Akasa fans, the fan controller and the PSU.

I'm trying to make it silent for desktop work, which you may think impossible with a powerful 3D card, but actually the card is very quiet with the fan set to 25% / 850rpm using Rivatuner, and it automatically underclocks itself when not using a 3D app.

I decided to replace the original 4 fans with 2 lower speed ones because the old ones only went down to around 1000 rpm at 5v, and were still too noisy. I took out the one in the lower HDD/PSU section since my hard disk doesn't get particularly hot and it didn't seem to serve much purpose when the PSU has its own fan.

I tried to create a good airflow with good through the case with a push-pull config: one fan at the front-middle and one exhaust fan at the rear-top. To attempt to guide the airflow and create better acoustics, I blocked off the top fan hole (can't see it in the pic, but it's above the heat sink) and most of the venting in the back panel with acoustic foam. The graphics card has its own fan and exhausts air directly out the back, so will act overall as an extra exhaust fan.

In this configuration the loudest part of the system was the CPU cooler, which controls its own fan speed and didn't respond to my motherboard's Q-fan control. It seemed to run unnecessarily fast, and I was thinking about getting a tower cooler like a Scythe Ninja, when I tried unplugging the CPU fan altogether. (I'd tried this before but the BIOS stopped loading the operating system and reported a fan error at boot-up. This time I just ignored the warning.) To my surprise it worked fine.

I'd re-oriented the cooler so that the fins point towards the exhaust fan (x-z plane) rather than the upwards (y-z plane). Ideally I think it would be in the x-y plane as on most tower coolers, so that hot air can both rise and be sucked towards the fan. I'm not sure which is the best orientation under low air flow... I think it might be best to put the exhaust fan in the top hole and orient vertically (but like this, the air going into the cooler will come off the graphics card, which is the hottest part of the system). Not easy to change since it's a motherboard-out job.

Test method
--------------

So, for the test, I tried a variety of configurations, varying the front (F) and rear (R) fan speeds, the CPU load using Prime95 (in-place large FTTs torture test), and GPU load using Furmark (GPU overclocked to 671 MHz vs default 576). I allowed the tests to run until temperatures stabilised.

Then I ran some tests with the side casing off, before removing the acoustic foam over the top fan hole when it was apparent that hot air was pooling in the top of the case.

The top hole then seemed to act as an air intake rather than an outlet as intended, so I turned the rear fan around to act as an intake, with the top hole as the main exhaust.

I did a few more tests in a second session, with possibly different room temperature, and a cleaned front dust filter (it wasn't really very dirty before however), to find the effect of removing the front dust filter and door, since I'd found that the front fan didn't seem to be blowing much air at all into the case.

Temperatures were monitored using Speedfan for CPU core temperatures, PC Probe II for "CPU temperature" (I think this is the CPU Case temperature, since it is generally cooler and lags the core temperatures), motherboard temperature warnings and voltage measurements, and Rivatuner for GPU and Ambient temps.

I stopped a test when the Core temperatures rose above 75 C. Intel's specified Case temperature is 74 C, and apparently the CPU will begin throttling with core temps around 85 C.

Results
--------
Image

Notes:

- Fan speeds: Max = 1080 rpm; Min = 600 rpm; 50% = 890 rpm

- Where core temperatures are given as 30-37 or 31-37, take the lower reading. This is because one of the core temperature probes was stuck (common problem apparently), and never read below 37. However it closely followed the other core probe reading above this.

- I don't trust the Ambient temperature reading as a reflection of ambient temperature. It very closely followed the GPU core temperature (about 11C lower) and I suspect the probe was located right next to the graphics card.

- "Door open" is the last two tests was with the case door open, the fan door open, and the dust filter removed.

- GPU temperatures were non-linear since they were constrained by a fan speed profile that stepped up/down at given temperatures

Conclusions
--------------

1. The Nexus cooler was (just about) sufficient to passively cool the Core Duo processor under sustained stress testing, with two 1000 rpm fans running in a push-pull configuration - although with these at full speed it was not quiet.

2. At normal workloads, the fans could be slowed to 600 rpm, which was very quiet.

3. The front fan had almost no effect on temperatures in the case. This was partly because the dust filter and fan door greatly impeded the flow of air into the case; I could feel a great difference with my hand in front of the fan with the side casing off. With the filter and door removed, the GPU temperature dropped at least 8C for a given GPU fan speed (see note above). The front fan acted to cool the GPU very slightly, but actually increased CPU temperatures.

4. Even a very small amount of air from the CPU fan, at 5V / 630 rpm, produced a 26+C drop in CPU temperature.

5. Unblocking the top vent reduced CPU temperatures by around 5 C.

6. Reversing the rear fan to act as an inlet reduced CPU temperatures by 8-9C at maximum fan speeds. At minimum fan speeds the CPU ran cooler under stress than it did at maximum fan speeds as an exhaust. However Ambient / GPU temperatures were higher by around 6-8C at low fan speeds.

7. Real game tests with Company of Heroes and Crysis Warhead ran successfully with minimum fan speeds and the CPU cooler running passively, with maximum temperatures at least 30C below their limits.

Final thoughts / future work
--------------------------------

- CPUs cool much better with fresh air blown at them rather than warm air from the case sucked through them.

- The front fan in the P180 case is essentially useless and should be disconnected. Don't put filters / doors in the way of inlet fans.

- I could remove the acoustic foam that I stuck onto the inside back of the case to increase ventilation, but since temperatures aren't that high I'm not sure I need to?

- Try with fan as exhaust in top opening.

- The noisiest part of the system when not gaming is currently the PSU. I got this cheaply with the case so I can probably justify an upgrade. :)


Question
----------

Should I go for a semi-passive PSU or just a quiet one with a fan? The Nexus Value 430 has been recommended but I'm not sure it's going to be powerful enough (GPU box recommends 550W with 36A on the 12v rail. However there's a review (link below) that shows a card similar to mine with an overclocked quad-core CPU in a system drawing 312 W - so my system is probably a bit less.
http://www.hardocp.com/article/2009/03/ ... ight_1gb/9

Hope these results were interesting / useful, and I'd be interested to hear any comments!

dragonfire
Posts: 62
Joined: Tue Nov 27, 2007 12:23 pm
Location: Massachusetts, USA

Post by dragonfire » Thu Oct 22, 2009 7:18 am

I have a Corsair HX520. Like others have said many times, it is by far the quietest item in my system. Even under load, makes no discernible noise. I highly recommend it.

Thanks for your detailed post. REally good info.

new2spcr
Friend of SPCR
Posts: 230
Joined: Wed Apr 22, 2009 9:14 am
Location: Sweden

Re: Experiments in CPU cooling in a P180

Post by new2spcr » Thu Oct 22, 2009 10:57 am

Interesting test. Especially where you used the exhaust fan
as an inlet and using the top vent as an exhaust.
This actually makes sense if you want to cool the hotter components such as the CPU and chipsets quicker, but yeah, the idea probably suits overclocker better.
Thanks for sharing.

nomoon
Posts: 210
Joined: Thu Apr 14, 2005 8:35 pm
Location: Allen, TX US
Contact:

Post by nomoon » Thu Oct 22, 2009 12:10 pm

I like your methodology of tracking your configurations and results in a spreadsheet. It sounds like finding a way to slow down your CPU fan would be a high priority. I make an argument in this thread that you don't want to remove the fan completely, or otherwise, your case fans will have to run too fast to compensate. Are you sure that you have the correct fan type for your motherboard : 3-wire vs 4 wire?

You also may consider getting something like a Fanmate2 to slow down your case fans (and possibly your CPU fan too). In my opinion, 1000 rpm is too fast for any fan for a quiet system. In my P180, my two case fans run at a fixed 720 rpm. My CPU fan idles at 540 rpm on a HR-01 Plus tower cooler.

Jason

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Thu Oct 22, 2009 3:35 pm

The imbalance between the P180 inlet and outlet cross sections and impedances has been widely acknowledged. Heck, even Antec made improvements in this area with the P183.

The standard fix is to remove the swing-out front doors (but leave the filters in place), and put a Kama Bay into the top 3 5.25" bays (your photo shows two of the four bays occupied; you'd have to give up one device).

As you've discovered, inlet fans do nothing but add noise.

Once the front inlet area is large enough, the system performs best with the top and rear fans blowing out.

To see this approach carried to its extreme, with lots of ducting and very slow fans, see this article http://www.silentpcreview.com/article672-page1.html.

rhys j
Posts: 31
Joined: Thu Oct 15, 2009 6:46 pm
Location: England

Post by rhys j » Thu Oct 22, 2009 8:30 pm

dragonfire wrote:I have a Corsair HX520. Like others have said many times, it is by far the quietest item in my system. Even under load, makes no discernible noise. I highly recommend it.
Thanks, will take a look at it.
nomoon wrote: It sounds like finding a way to slow down your CPU fan would be a high priority. I make an argument in this thread that you don't want to remove the fan completely, or otherwise, your case fans will have to run too fast to compensate. Are you sure that you have the correct fan type for your motherboard : 3-wire vs 4 wire?

You also may consider getting something like a Fanmate2 to slow down your case fans (and possibly your CPU fan too). In my opinion, 1000 rpm is too fast for any fan for a quiet system. In my P180, my two case fans run at a fixed 720 rpm. My CPU fan idles at 540 rpm on a HR-01 Plus tower cooler.
I did actually manage to slow it down by attaching it to the fan controller, so now it doesn't go above 690rpm (compared to the stock minimum speed of about 1200-1400 rpm), which is very quiet. It's a 4-wire connection to the motherboard so I don't know why the Q-fan doesn't work, but the cooler controls its own fan speed (not sure how exactly). Anyway, if the CPU doesn't overheat running passively it'll be fine permanently on minimum speed. I think Nexus messed up in the design, especially since they marked this cooler as a silent one, since I showed that the fan doesn't even need to run at all except under extreme loads and restricted case airflop.

I have the Akasa fan controller so I can get the 1000 rpm fans down to about 600 rpm (they run ok but don't start up at this level), but having done these tests I agree that 1000 is unnecessarily fast and I could have gone for even slower ones.

cmthomson - thanks for the link and advice. I don't think I really need the extra inlet fan for my system but will bear it in mind if I build something more powerful. Another complication when thinking about the airflow is that the graphics card must exhaust a significant proportion of air in the case when running at high speed.

MisterB
Posts: 15
Joined: Thu May 20, 2004 12:16 pm
Location: Netherlands
Contact:

Post by MisterB » Fri Oct 23, 2009 12:19 am

Just wanted to thank you for sharing your results; very enlightening and complete. I will probably get a P183 soon myself and this is a big help in getting the configuration right.

Post Reply