Viewing page 6 of 6 pages.
Previous 1 2 3 4 5 6
TESTING & MEASUREMENTS
As noted, we were unable to monitor thermal conditions or fan speed in Windows,
which made it impossible for us to determine how effective the heatsink is or
what its potential for undervolting is. The system was stable throughout 30
minutes of running CPUBurn. No benchmarks were run; performance of the motherboard
should be typical of the chipset it is based on.
Most of the testing was done using the Smart Fan setting, as this is the only
one that we expected to change under load. The fan will run at full speed regardless
of the BIOS setting if the temperature gets hot enough to cause damage to the
CPU.
Ambient temperature was 20°C, and ambient noise level was 20 dBA/1m.
During preliminary testing, we realized that one of the major
sources of noise in the system was our drive, the Barracuda IV, which resonates
at a low frequency even in the steel case. In order to test the acoustic properties
of this XPC as fairly as possible, we swapped the Barracuda for a 40GB Samsung
notebook drive placed on a bed of foam in the same rear drive cage for the rest
of the test. This was the same HDD used for our
recent Shuttle SN95G5 review.
We do not believe it is possible at this time to build a quiet
SFF system using a bare hard-mounted 3.5" hard drive; SFF systems are simply used too close
to the user for even the quietest 3.5" drives to be used. A resilient suspension mounting would change this, but in the vast majority of SFF systems, there is no space for such a setup.
SOUND MEASUREMENTS
|
Shuttle XPC SB86i: Sound Measurements
|
|
Test Conditions
|
AC Power Draw (W)
|
SPL (dBA/1m)
|
|
Seagate Barracuda IV, Idle
|
78
|
29
|
|
Samsung Notebook, Idle
|
72
|
29
|
|
Samsung Notebook, Load
|
150
|
33
|
Idle
Despite a considerable subjective improvement in noise levels after we switched
the hard drive, the SPL measurement that we made was identical with either hard
drives. We trust our ears more than we trust our sound meter: With the low hum of the Barracuda HDD gone, the noise is dominated by the 92mm heatsink
fan. Unfortunately, this noise is
less benign than hard drive hum. It is best described as a low, grating buzz
that is quite rough in character. It is not an especially loud sound, but its
uneven character makes it hard to ignore.
Were this a tower system designed to sit at our feet, not our ears, we could
probably call this a quiet system. 29 dBA/1m is not a bad result, and it is
below the 30 dBA/1m threshold that we usually consider quiet. However, the typical
operating position of a SFF system is not under a desk but on
it. For this reason, we are more stringent about SFF systems. We consider the SB86i to be only borderline quiet at best.
Load
To get a rough idea of the thermal performance of the Thermal Module, we
took a quick peek at the CPU temperature in BIOS after a 20 minute session
of CPUBurn. Based on the reading here — 55°C — we can guess
that the CPU temperature never exceeded 65°C under load. This is a high
but perfectly safe temperature. By comparison, the baseline temperature in
BIOS using Smart Fan was 49°C.
Our test processor never got hot enough to cause the CPU fan to speed up
and Throttlewatch showed that the CPU never throttled, so it would appear
that the Smart fan setting is conservatively tuned.
The increase in noise under load comes entirely from the power supply. Because
the power supply fan is meant to exhaust the heat from the entire case, it
must speed up to compensate for the extra heat that the CPU produces.
Unlike the CPU fan, the power supply fan has a fairly smooth character; most
of the extra noise comes in the form of a smooth mid-frequency hum and turbulence
noise. Compared to the narrow aquarium-pump buzz of the CPU fan, it is more broadband noise that could be tuned out fairly easily. The problem with the
noise under load is the volume, not the quality.
It is possible that a lower noise level under load could be achieved by running
the CPU fan slightly faster. However, given the terrible quality of the noise
that the CPU fan produces, we would not recommend this.
Shuttle's own published acoustic report on the XPC SB86i jibes with our assessment that it is not a very a quiet system. Here are their SPL readings at ~0.6 meter distance of a hotter system configured with a P4-3.6, a gig or RAM, and a louder Seagate 120GB S-ATA drive: With Smart Fan enabled, 37 dBA at load; 33.1 dBA at idle. (Subtract 4~5 dBA to get approximate 1 meter distance SPL readings.)
THERMAL MODULE FAN SWAP
Because the main source of noise in the system was so obviously Intel's stock
fan, we decided to see if noise performance could be improved by swapping it
with the quietest 92mm fan that we know of: A Nexus.
The Nexus fan is much quieter — but provides much less airflow.
The improvement in noise quality was quite dramatic. With the system
in BIOS, system noise dropped 3 dBA/1m to 26 dBA/1m. Most importantly, the ugly
buzz that was characteristic of the system disappeared almost completely.
Although the main source of noise was still the front fan, the noise character
with the Nexus was much more tolerable.
|
Nexus Fan Swap
|
|
Fan Speed (RPM)
|
BIOS Temp (°C)
|
SPL (dBA/1m)
|
|
1550
|
66
|
26
|
Because our Nexus 92 had only a traditional three-wire header rather than a 4-pin
PWM header, the various fan settings available in BIOS had no effect after our
fan swap. The voltage supplied to the fan remained constant at 12V no matter
which setting was selected.
However, even at 12V, performance with the Nexus was marginal. The temperature
in the BIOS stabilized at 66°C, only a couple degrees short of the ~68°C threshold
when some P4s start throttling. AC Power draw in BIOS was 110W, which suggests
that the CPU was under a moderate load. If power draw (and thus heat dissipation)
was increased to the levels seen under CPUBurn in Windows, we would expect to
see quite severe throttling as the CPU struggles to keep itself cool. The Nexus
does not provide adequate airflow to cool this system, even at full voltage.
While the stock fan was separate from the Thermal Module we took the opportunity
to listen to it in free air. Without the plastic duct surrounding it, the buzz
that we observed previously was much reduced. This suggests that the duct itself
is a large part of the system's acoustic signature. The resonance that it causes
would likely be a problem no matter which fan is used. (Note that this is not the plastic material of the duct vibrating, but the air in the duct resonating.)
MP3 RECORDINGS
MP3 recordings were made with the system at idle with both hard drives, and
with the system under load using the notebook drive.
Idle Comparison
MP3:
Shuttle XPC SB86i with Seagate Barracuda IV, Idle: 29 dBA/1m
MP3:
Shuttle XPC SB86i with Samsung Notebook Drive, Idle: 29 dBA/1m
MP3: Shuttle
XPC SB86i after 92mm Nexus Fan Swap, Idle: 26 dBA/1m
The main difference to listen for is a low frequency hum that is present
in the Barracuda file but not the Samsung file. You may need to turn up the
volume a little to hear the difference, especially if you're using low quality
speakers.
The difference between the recordings with the stock fan and the one with
the Nexus is quite dramatic and needs no further explanation.
Load
MP3:
Shuttle XPC SB86i with Samsung Notebook Drive, Load: 33 dBA/1m
Comparisons
MP3: Silverstone LC-11 test system at any load - 23 dBA @ 1 meter SPL
MP3:
Arctic Cooling Silentium T2 with test system (3.5" HDD suspended) at idle - 23 dBA/1m
MP3:
Arctic Cooling Silentium T2 (3.5" HDD suspended) at max load - 34 dBA/1m
MP3: Shuttle SN95G5 SFF system (notebook drive suspended) in idle - 27 dBA/1m
MP3:
Shuttle SN95G5 SFF system (notebook drive suspended) at max load - 30 dBA/1m
|
SPCR MP3s: HOW TO LISTEN & COMPARE
The recordings above were made with a high resolution studio quality digital recording system. The microphone is 3" from the edge of the fan frame at a 45° angle, facing the intake side of the fan to avoid direct wind noise. The ambient noise during all recordings is 20 dBA or lower.
A quick and simple way to use these recordings for valid listening comparisons is to play the quietest recording on only one speaker (or a pair of headphones) and set the volume so it is just barely audible a meter away. You must turn off any special sound effects, and set equalizer / tone controls to neutral or flat. Don't touch the volume setting afterwards, and use the same one speaker when you listen to any of the other files. The end result will be reasonably close to the actual recorded sound levels.
Here is a recording of a very quiet sound that is barely audible from 1 meter away even in a super quiet room.
For full details on how to calibrate your sound system to get the most valid listening comparison, please see the yellow text box entitled Listen to the Fans on page 3 of the article SPCR's Test / Sound Lab: A Short Tour.
|
CONCLUSIONS
The XPC SB86i has plenty of potential. A sturdy steel case, modular
BTX construction, a limited number of large fans, the nice air intake design of the front bezel, and low-restriction, straight through airflow path — all of these make this case a good candidate
for modding.
But the charm of a barebones SFF system is that it should not need
modding. This type of system is sold to the end user as a package, albeit one that
needs to be completed with a few key purchases. The acoustics of the system, however, should be quiet enough
when used with the kind of minimal heat components we chose for the test system. Unfortunately, in its stock form, the SB86i is not a quiet system.
The deal killer here is Intel's 92mm, PWM-controlled fan in the BTX Thermal
Module. However, there are few alternatives on the market yet, so fixing this flaw is not a simple as just replacing
the Thermal Module. Our experiment with the much less powerful Nexus did improve
the noise quite a bit, but at the cost of acceptable cooling performance. Furthermore,
the resonance caused by the plastic duct makes it difficult to recommend using
a more powerful fan.
There may be some question about how well the system could handle a more aggressive
set of parts. We've already pointed out that the highest power VGA cards cannot
be supported by the power supply. From a thermal standpoint, adding a significant
amount of heat downwind of the CPU may be a bad idea, as the air used to cool
these devices would already be heated by the CPU. We also noticed that our hard
drive ran much warmer than usual. On the other hand, there's no question that with a hotter CPU, the fan will simply blow harder. Cooling is probably not going to be the issue with hotter components; noise will be one for many users.
Overall, this is an interesting effort from Shuttle with
a lot of firsts. We look forward to seeing more steel cases, and we expect the
quality of BTX-based systems to improve.
|
PRO
Modular and upgradeable
Steel chassis
Wide feature set
Good component layout
Good looking
|
CON
Stock Intel fan is noisy
Poor hard drive airflow
Incomplete PCIe support
Missing thermal monitoring software
PATA hard drive cannot be used
|
Much thanks to Shuttle
for providing us the XPC SB86i sample and to Newegg.com
for the Intel 520 CPU loaner.
* * *
Discuss this article in the Silent PC Review Forums.
| Help support this site, buy the Shuttle SB86i BTX SFF system from one of our affiliate retailers! |
|
