Viewing page 6 of 7 pages.
Previous 1 2 3 4 5 6 7 Next
THERMAL AND NOISE TESTING
The six fans in the TJ-07 provide many potential configurations, depending
on the specific needs of the system and the whims of the user. We have no doubt
that the TJ-07 is capable of cooling just about any system that is installed
in it — its large size, and numerous fresh air intakes almost
guarantees it. Whether it can do so quietly is the relevant question for SPCR.
Our testing was designed to answer these question:
- How well does the TJ-07 contain
noise?
- How much does the airflow design allow fan speed to be minimized?
- How quiet can the TJ-07 be without any serious modifications?
Our test system was fairly modest; we wanted to ensure that the TJ-07 could
handle an ordinary system before moving on to something more challenging.
Test System
- AMD Athlon 64 4000+ — TCaseMax reported a TDP of 50W, and a
maximum case temperature of 59°C
- ASUS A8R32-MVP motherboard — based on the ATI CrossFire Xpress
3200 chipset
- Scythe Ninja heatsink, no fan — the best low noise cooler we
know of
- Corsair XMS Xtreme PC-4000 DDR RAM — 2 x 1024 MB
- ASUS EAX1600XT Silent — passive graphics card
- Samsung Spinpoint SP0802N 80 GB 1-platter drive
- LG DVD-RW drive
- Seasonic S12-380 — 380W version of the quietest fanned power
supply we know of
- Arctic Silver Ceramique Thermal Compound
Testing Tools
Ambient conditions were 22°C and 20 dBA.
A number of quick tests were run initially to determine the best configuration
for cooling.
Where possible, the build-in fan controller on the motherboard (QFan) was used
to regulate fan speed. However, only the two rear fans had the appropriate headers to plug into the motherboard. We ended up hardwiring the top fans to specific voltages as required.
For each test, CPUBurn and RTHDRIBL were both run simultaneously so that the
cooling capabilities of the case were stressed as much as possible.
|
SilverStone Temjin TJ-07
|
|
Configuration
|
CPU Temperature
|
System Power (AC)
|
Noise Level
|
|
#1
|
Top Front (120mm) |
12V
|
Overheated
|
150W
|
31 dBA@1m
|
| Top Back (120mm) |
Blocked
|
| Rear Top (92mm) |
Off
|
| Rear Bottom (92mm) |
~1000 RPM
|
|
#2
|
Top Front (120mm) |
12V
|
47°C
|
152W
|
31 dBA@1m
|
| Top Back (120mm) |
12V
|
| Rear Top (92mm) |
Off
|
| Rear Bottom (92mm) |
Off
|
|
#4
|
Top Front (120mm) |
7V
|
60°C
|
157W
|
24 dBA@1m
|
| Top Back (120mm) |
7V
|
| Rear Top (92mm) |
Off
|
| Rear Bottom (92mm) |
Off
|
|
#5
|
Top Front (120mm) |
7V
|
44°C
|
149W
|
25 dBA@1m
|
| Top Back (120mm) |
7V
|
| Rear Top (92mm) |
Off
|
| Rear Bottom (92mm) |
~1000 RPM
|
Configuration 1
Our initial configuration used the top front and the rear bottom fans together
to draw air through the center of the system, across as many components as
possible. The back half of the top grill was blocked off to prevent any short-circuits
in the airflow.
Unfortunately, this was a better idea in theory than in practice, as the
CPU quickly overheated in this configuration. We also discovered that the
two rear fans sounded quite nasty, with a heavy buzz that may have been amplified
by the panel that they were screwed to. This noise was present even
when the fans were spinning slowly; lowering the fan speed reduced the volume
but not the intensity of the noise.
Configuration 2
The second configuration was a reaction to the first: The rear fans were
disabled completely, and the two top fans were run at full speed. Not surprisingly,
this configuration was too noisy to consider seriously, but the cooling improved
dramatically.
Configuration 3 (not in the table)
The next step was obvious: Reduce the 120mm fan speeds to an acceptable level.
Both top fans were undervolted to 5V, and the system was powered up. We were
quite impressed with the resulting acoustics at first. We didn't think the top fans would undervolt so well. However,
even with the processor idling in Cool 'n' Quiet, the temperature kept rising...
and rising.
Wondering why, we took a careful look at the system, and realized that neither
of the two fans had started at 5V. Oops.
Configuration 4
So, we tried again, this time undervolting the two top fans to 7V. This time
we made sure the fans started, and, sure enough, there was a small amount
of airflow rising from the top panel.
We were shocked to discover
that the system measured 24 dBA@1m. Normally, 24 dBA@1m is a little above
the ambient noise level in the lab and can be tuned out easily enough. However,
the subjective quality of the noise made it sound considerably noisier than
what we usually expect to hear at 24 dBA@1m. Instead of being just a little above
the ambient noise level, the noise from the system could be heard fairly clearly
in the next room.
The quality of the noise was terrible, characterized by a deep throbbing
thrum with a ringing overtone. The noise character could be changed (usually
for the better) by pressing down of various parts of the case to damp the
vibration, which indicates that much of the noise character was resonance
from the aluminum panels. Considerable improvements to the noise character
could be heard by pressing down on the top mesh, which the fans were screwed
to, and the side panels, which damped a significant amount of hum. Sometimes,
an audible difference could be heard after doing something as simple as shifting
the angle of the case or picking it up and putting it back down.
This configuration was about as quiet as we expected to get using the stock
fans, but the cooling was only marginal. The CPU peaked at ~60°C, which
was actually above 59°C maximum temperature specified by AMD for our chip.
Configuration 5
Because of the questionable cooling in Configuration 4, we decided
to try turning on one of the rear fans again — spinning slowly —
to see if that would improve things. Boy did it ever! The CPU temperature
dropped by ~15°C under load, to a reasonable 44°C.
With the rear fan spinning at 1000 RPM, the measured noise level
rose only a little. Once again, the different measurement did not reflect
our subjective impression of the change. Not surprisingly, the noise was quite
similar to Configuration 1, although the overall volume was lower.
| Help support this site, buy the SilverStone Temjin TJ-07 (Silver) from one of our affiliate retailers! |
|
