Arctic Cooling Silentium T2

THERMAL AND NOISE TESTING

Our test system was quite basic, and was intended to reflect a typical low-to-mid power system. The system is the test platform used in our recent review of the Arctic Cooling Super Silent 4 Ultra TC with the exception of the PSU and HDD damping system. (Only the integrated VGA was used.) Temperature measurements should thus be comparable between the two reviews, and some judgment of how the Silentium compares to an open test bench can be made. Noise levels between the two reviews are not comparable, however, as this review is intended to show how effective the Silentium is at damping noise, while the SS4UTC review was designed to test the noise levels of a specific heatsink.

Test Bench

• Intel P4-2.8A The Thermal Design Power of this P4-2.8 (533 MHz bus) is 68.4 or 69.7W depending on the version. As the CPU is a demo model without normal markings, it's not clear which version it is, so we'll round the number off to ~69W. The Maximum Power, as calculated by CPUHeat & CPUMSR, is 79W.
• AOpen AX4GE Max motherboard - Intel 845GE Chipset; built-in VGA.
• Arctic Cooling Super Silent 4 Ultra TC heatsink
• OCZ DDRAM PC-3700, 512 MB
• Seagate Barracuda IV 40G 1-platter drive
• Arctic Silver Ceramique Thermal Compound

Testing Tools

• CPUBurn processor stress software
• Motherboard Monitor 5 software to show CPU temperature
• Steinberg Wavelab 5 recording and analysis software
• T.H.E. KP-6M reference omni microphone
• M-Audio Tampa mic preamp
• M-Audio Firewire 410 external digital sound interface
• B&K 2203 sound level meter

Ambient temperature was 20°C. Total system draw was 58W at idle, 118W during folding, and 132W during burn, measured at the AC outlet.

TEST #1: ACOUSTICS BASELINES

Our first test was done with the system at idle, placed on the carpeted floor in the quietest room available in order to establish a reference noise level for the case itself. Ambient noise level was 16 dBA. A 40G Seagate Barracuda IV was used to test the effectiveness of the suspension system.

Although the exhaust fans and the PSU fans appear to be identical, they do not start off spinning at the same speed. We were unable to determine exactly where the thermal sensors for the two sets of fans are located, but the power supply fans appear to have a minimum voltage that is significantly higher than that of the rear exhaust fans. The result is that the power supply fans are consistently the main source of noise in the system.

At 23 dBA, the starting noise level of the Silentium appears deceptively low. Subjectively, however, the noise is a specific pitch centered somewhere around middle A (440 Hz) and is immediately recognizable and irritating.

To discover just how the noise is distributed between the PSU and exhaust fans, a second measurement was taken with the PSU fans stopped by jamming them with the reviewer's fingers (don't try this at home, kids). The change in noise level that this produced was quite dramatic. Not only did the measured noise drop to 18 dBA, but the irritating hum that was so evident in the first test completely disappeared. Even with the side door removed, the fans were barely audible from a distance of one meter.

To illustrate just how dramatic the difference that the power supply fans make, we used Steinberg Wavelab 5 to produce 3-D spectrum analysis graphs of ~5 second noise segments. (Click on the images below to enlarge in a separate browser window.)

The Silentium noise profile with all four fans running. The character of the noise is dominated by a spike around 440 Hz.

The Silentium without the power supply fans running. The 440 Hz spike has completely disappeared, and the graph is now dominated by an intermittent band of noise centered around 100 Hz, most likely associated with the HDD. This is not very audible.

Note: The vertical scale of these graphs is not quite identical.

We also took this opportunity to test the effectiveness of the hard drive suspension system. Our conclusion is that the HDD Muffler is as effective as any method of suspension. Running the case with the hard drive unplugged did not produce any noticeable change in noise level, and the measurements bear this out. The 1 dBA difference between the two situations is well within the margin of error of the measuring equipment, and subjectively the two noise levels were too close to call. At any rate, the noise contribution of the hard drive was made inaudible by the dominating drone of the power supply fans.

TEST #2: COOLING & NOISE AT LOAD

After establishing the baseline levels, the system was moved into a different room for thermal testing. This room is equipped with desktops, keyboards, monitors, access to network and high speed web. Ambient noise level in this room was ~20 dBA, somewhat higher than in the previous room. Not only was the ambient noise level higher, but the case was placed on a table. There was also a trace of audible noise from the table resonating due to conduction of vibration from the case. These different conditions are reflected in the 4 dBA higher noise level measured at system idle.

Idle: Our test system does not produce enough heat at idle to push the case airflow capabilities. At 33°, the CPU is roughly the same temperature as in free air. More interesting is the temperature reading for the hard drive. This drive typically idles in free air around 38-40 °C, but, according to the SMART sensor on the drive, the temperature is just 29 °C, easily 10 °C cooler This suggests that the HDD Muffler is effective as a heatsink as well as a silencing tool. It also suggests that there is a considerable amount of air being drawn across the Muffler by the power supply fans, as illustrated by AC's airflow diagram.

CPUBurn: Once the CPU is fully loaded, the unconventional airflow of this case begins to make a difference. After 20 minutes of CPUBurn, the CPU temperature was 51 °C, three degrees lower than in our open bench setup . This means that the cool air being drawn into the case and directed around the heatsink is doing its job. This is quite an impressive result.

The noise level at full load is less impressive. The character of the noise does not change much ? it is still dominated by the hum of the power supply fans, although the pitch of the noise has risen correspondingly with the increase in fan speed. There is a trace of residual noise from the HSF underneath the power supply fans, but not enough to affect the measurements by more than a couple of dBA. At no point did the rear exhaust fans become a distinguishable source of noise; they remained at a considerably slower speed than the power supply fans throughout the test.

At full load, even our modest test system cannot be considered quiet. This is not a fault with the case and case fans; the noise is directly traceable to the fans in the PSU. Because the power supply is fully integrated into the case and its thermal sensors are not easily modifiable, this is quite a serious drawback with this case.

At full load, the spectrum of noise has shifted further into the frequency range to which humans are most sensitive. The power supply fans are now spiking around 600 Hz.

MP3 RECORDINGS

To judge the case noise for yourself, please download and listen to the MP3 files below.

MP3: Arctic Cooling Silentium T2 at Idle, 23 dBA/1m

MP3: Arctic Cooling Silentium T2 at Idle,PSU fans stopped, 18 dBA/1m

MP3: Arctic Cooling Silentium T2 at Full Load after 20+ min. of CPUBurn, 34 dBA/1m