Sub-$20 CPU Coolers: A Reader's Roundup

Viewing page 2 of 13 pages. Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 Next

Testing Rig

My chosen test rig was the machine that would be the final home of the winning cooler: my home theater PC. It's a custom-build rig in a slim mATX case with terrible cooling capabilities and an atrocious power supply. It currently houses a "Conroe" Core2 Duo E4600 running at stock speeds and voltages (2.4GHz and 1.4v) on a Foxconn 45CMX motherboard built around the venerable (read: old) Intel 945 chipset.

Test platform

"Holly," as I call this system (shout out to the Red Dwarf fans out there!), was pieced together over the last few years from low-cost or cast-off parts and has only recently been shoehorned into the role of "ultra-quiet media PC" -- a job that, at first, it didn't do well or easily. Several fan resistors, a little PSU surgery (replacing the stock PSU fan with a 92mm Noctua) and a new 120mm side fan later and things are finally looking pretty good for the little HTPC.

Here are the full system specs if you're interested:

CPU Intel Core 2 Duo "Conroe" Socket 775 (65W TDP)
Motherboard Foxconn 45CMX
Chipset Intel 945
RAM 4 GB DDR2 667 (chipset consumes 800MB)
GPU HIS iSilence fanless Radeon HD 7750 1GB
HDD WD Caviar 160GB PATA/133 (7200 RPM)
Case Sentey SS2-2220 Slim mATX case
Power Supply Sentey 250W ATX12V unit included with case
OS: Windows 8.1 Pro
Network Generic WiFi G USB adapter


Test Methodology

I removed the motherboard from the case for my own sanity, to make swapping coolers with potentially-questionable mounting systems a little simpler.

I measured ambient air temperature using the built-in thermistor on the machine's case by placing it between 3 and 4 inches from the cooler's fan intake. I measured CPU temperature using SpeedFan's reading of the internal thermal diode on the Core2. These two temperatures were not the most accurate means of measuring temperature rise above ambient, but it was all I had at the time, and I did my best to make sure this test setup was the same with each cooler.

I used Noctua NT-H1 thermal paste for all coolers in the test. It's a high-grade paste and should avoid the inherent irregularities seen when testing with the pre-applied pastes that ship on these low-end coolers. I applied the same amount of paste to the CPU with each cooler, allowing it to spread over the CPU's heatspreader by the force of the cooler upon mating, as I've done for over a decade of PC building now.

Noctua thermal interface material

With the cooler mounted, I connected the fan to my Zalman FanMate 2 voltage-regulated fan controller and set the controller to maximum speed. I would then boot the system and have a good listen. Windows 8's fast boot feature made this task quite a bit less-arduous than it could've been, but it was still time-consuming.

With the system booted and the internal hard drive of the HTPC settled down, I took to characterizing the sound of the cooler's fan. I adjusted the speed, making sure to note relative sound properties between the lowest and highest speeds achievable on the FanMate. Then I'd try to make an estimation of the cooler's "quietness." This is where the testing became completely subjective. With the HTPC's hard drive running (but idle), I adjusted the fan speed down to the point where I had serious difficulty discerning fan noise over the hard drive.

At this point, I noted the RPMs necessary for "silence" (really, quiet) and let the system idle for five minutes before taking the ambient temperature and idle core temperature for "quiet" fanspeed. I then fired up FurMark's CPU Burner and let it run for ten minutes with the fan at this "silent" speed. After ten minutes, I measured the load CPU temperature for "quiet" and stopped FurMark.

Test programs

At this point I turned the fan up to maximum speed and noted the peak RPMs while I let the CPU cool for another five minutes before recording the core temperature once more, for maximum speed idle. I then fired up FurMark again and let it run for another 10 minutes before recording the core temperature reached for load at maximum fan speed.

Once I was satisfied with my test results (and made sure to note any changes in the cooler's fan sound since I'd turned it on), I stopped FurMark, let the system cool for a bit and shut the system down with Fast Shutdown, making sure the FanMate was again on maximum. I snapped pictures of the thermal paste spread on the cooler, cleaned everything up and repacked the cooler in its packaging before unpacking the next one.

In all, pretty basic testing procedures. I performed the testing in my basement, with all air-handling and laundry equipment (air conditioning, dehumidifier, washer and dryer) turned off for maximum quiet. I was able to complete this testing over the course of three days, in my free time.

Anyway, without further ado, let's get to the testing!

Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 Next

Cooling - Article Index
Help support this site, buy from one of our affiliate retailers!