Viewing page 2 of 2 pages.
Previous 1 2
UNDERVOLT & UNDERCLOCK THE CPU
In order to meet my
requirements,
I felt it necessary to underclock and undervolt the system to lower the CPU heat. Numerous articles in Silent PC Review have shown that any change in CPU core voltage has a nearly exponential effect on the
power
-- whether you reduce or increase voltage. The relationship between CPU
clock
speed and power is linear, however. Decrease clock speed by 10%, power goes down
by
10%.
For an AMD XP Thoroughbred 1700+,
the
default specs are 1466 MHz, 1.5V and 49.4W. If you reduce clock
speed
to 1200MHz, the power drops to 40.4W, which is exactly 18% -- same as the
clock
speed drop. But if you reduce the Vcore by just 20% from 1.5V to 1.2V, the power
drops
to 31.6W -- this is a 36% reduction! If you apply both of these together, (1.2V &
1200MHz),
the power drops to 25.8W -- a 47.6% drop. 1200MHz and 1.2V is quite fine
for
me.
You can see the temperatures
and
other information in the table below. In order to properly compare the two
settings
Stock and Underclocked states, I left the Zalman fanmate at the same setting, so
the
fans were running at the same speed, same noise. Load temperatures were
attained
by running CPUburn until the temperature stabilized. I have calibrated the
NF7s
according to Russ Kinder's article, Calibrate
Your CPU Temp Reporting.
|
NOTE:
Undervolting and underclocking
are much discussed topics in the SPCR community. More
information on these topics can be found in the following
articles:
|
Measurement & Analysis
Tools
- CPUBurn
processor stress software
- Motherboard
Monitor 5 software to monitor CPU temperature
- Enermax UC-A8FATR4 multifunction monitor/fan controller w/ thermal sensors
- 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 (SLM)
The ambient temperature in the
test
lab was 21°C. Maximum load temperatures
were
recorded >20 minutes into a CPU stress test with
CPUBurn.
|
Setting
|
Clock
Speed
|
FSB
|
Multiplier
|
Vcore
|
Idle
Temp
|
CPUburn
|
|
Stock
|
1.46
GHz
|
133
|
11
|
1.5 V
|
45°C
|
57°C
|
|
Custom
|
1.2
GHz
|
133
|
9
|
1.2 V
|
36°C
|
41°C
|
While the underclock/volt is
minimal
— only 260 MHz and 0.3V — it is enough to reduce idle temps by a huge 9°C and load temperatures by an even bigger 16°C. Subjectively, the computer runs about as fast as when the settings were at stock.
ACOUSTICS &
DAMPING
Having underclocked the CPU and
reduced
the speed of the fans, the system is relatively quiet. But quiet to my
ears
with the computer in a cabinet across the room
is
a different thing from sitting right beside the computer in the open. So, in
order
to give you a good idea of how loud this system was running, I brought it to
the
SPCR lab to get measurements, audio recordings and analysis by the man himself.
While we were starting to do the acoustic testing, Mike suggested
applying
some Nexus DampTek
in the case as part of SPCR's ongoing exploration of acoustic damping materials.
This sounded good to me.
The material consists of a 1.5 mm "thermoplastic unvulcanized rubber" mass barrier layer to isolate the noise and a 6 mm foam layer to absorb multiple frequencies. It is similar to the AcoustiProducts Acoustic Composite sheets that Mike showed me samples of, although the AP material is thicker and a bit heavier.
Nexus DampTEK on left, AcoustiProduct Composite on right.
Here are some details about DampTek kit from Nexus:
- 3 sheets of DampTek noise absorption material
- each sheet consists of a barrier to isolate the noise and a foam layer to absorb the multiple frequencies
- size per sheet: 40 x 50 cm / 15.7 x 19.7 inch
- Easy to cut and shape
- Self-adhesive backing
- Easy to remove and to refit
Fire-resistant
- Thickness ca. 7.5 mm (1.5 mm barrier layer and 6 mm foam layer)
- Total is compressible down to approx. 2mm
One sheet was more
than
enough to line the inside cover of the Silverstone LC-01 case. The area over the PSU, the optical drives and the center reinforcement bar had to be kept clear. The process took only about 20 minutes with the aid of a felt pen, a ruler and a utility knife. Naturally, the case sounded much better damped afterwards, with a dull thud when rapped. It was a bit heavier as well.
Acoustic measurements, recordings and listening was done before and after the damping was applied, with and without the suspension mounting for the hard drive. Each configuration was measured and tested with the
system
at idle (quietest) and with the hard drive being defragmented
(loudest).
The table below shows the results, including sound pressure levels (SPL).
|
Configuration
|
Idle
(dBA/1m)
|
Defrag
(dBA/1m)
|
MP3 Audio Recording*
(save to disk, then listen)
|
Notes
|
|
A. Case
undamped,
HDD hard mounted
|
24
|
28
|
A. Jordan's HTPC at "stock" - 24~28 dBA/1m |
Very quiet, pretty smooth sound. Defrag sound is quite sharp. Note that noise level does not usually change with high long term load.
|
|
B. Case
undamped,
HDD suspended
|
24
|
28
|
B. Jordan's HTPC with suspended HDD - 24~28 dBA/1m |
Subjectively
much quieter than above, especially during HDD activity. The SPL measurements does not seem to correlate well here with what we heard.
|
|
C. DampTek
applied,
HDD hard mounted
|
24
|
28
|
C. Jordan's HTPC with DampTek
- 24~28 dBA/1m |
We heard no improvement in noise despite the damping. Our subjective assessment was confirmed by the SPL measurements.
|
|
D. DampTek
applied,
HDD suspended
|
22
|
24
|
D. Jordan's HTPC with DampTek and
suspended HDD - 22~24 dBA/1m |
This was definitely the smoothest and quietest config. The noise reduction was more noticeable during drive activity (which could not be heard) but it was apparent even without. The SPL readings support our perceptions this time.
|
|
* Audio Recording Notes: Each recording is 30 seconds in length. The recordings were made with the system in idle for at least 10 seconds. Then HDD defragment begins. The sequence is not identical for each sound file, but it is close. You may wonder if disk defrag was actually turned on for B and D; rest assured it was. We can just barely hear it in B. We cannot hear it at all in D.
|
|
How to Listen and Compare
The MP3 recordings were made with a high
resolution lab quality digital recording system. The microphone was 3" from
the front center panel of the system. The ambient room noise during all recordings was ~18 dBA.
To set the volume to a realistic level (similar to the original), try playing this reference file, Nexus 92mm case fan @ 5V (17 dBA/1m), and setting the volume so that it is barely audible. Then don't reset the volume and play the other sound files. Of course, tone controls or other effects should all be turned off or set to neutral. 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 four of the article
SPCR's Test / Sound Lab: A Short Tour.
|
Mike and I discussed the results above. Several questions were raised:
- Why did the Nexus DampTek damping material have no effect on the noise with the hard drive normally mounted?
- Did this mean the material is not effective?
- Then why did it reduce the noise when the hard drive was soft mounted?
A logical explanation for these results was reached. As Mike has a lot more experience on this topic, I'll quote his words here:
|
"The damping makes no difference with the HDD hard mounted
because the loudest part of the noise that we're hearing is actually caused by vibrations transmitted into
the
case from the hard drive. The damping does
little
to stop the conduction of vibration through the metal chassis and the
subsequent
noise that the vibrating panels cause. The case cover is acting
like
the sounding board in a musical instrument, and the frequencies of the noise are mostly below 200Hz, where the bit of mass damping in the DampTek (or the AcoustiProduct stuff, for that matter) has very little effect. When the HDD is suspended, you greatly
reduce
the conduction of vibration, so the airborne noise becomes predominant. The Nexus DampTek does have a
greater damping
effect on this airborne noise, which is why we hear less noise.
The degree of noise reduction is small at idle because a lot of that noise is actually coming right out the air vents of the case. There's little the damping can do to stop this. During HDD defragmentation, the noise reduction is greater because it's now all airborne noise, and the HDD does not have a direct path to our ears. Thus, the damping is able to absorb more of the HDD noise before it escapes out the air vents in the case.
|
What this tells us is that acoustic damping panels are not magical cure-alls as so many people seem to think. They are tools that can be helpful when used in the right way. You can't expect them to do miracles. Certainly, acoustic damping materials applied to case panels cannot make a loud computer quiet. For best results, you have to do your due diligence in reducing the noise and vibrations at the sources before turning to acoustic damping materials. And be prepared to experiment!
CONCLUSIONS
I feel this project has been a
resounding
success. I have met the objectives for the system detailed at the beginning and
I
have made my parents happy. The undervolting/clocking gives me piece of mind
knowing
that the system runs nice and cool. It also has more than enough
power
to view pictures and slide shows as well as stream movies from my main
system.
Working with the damping material
to
this system was an interesting exercise. The effectiveness of the
panel damping
depends partly on the nature of the noise and the way that it is propagated within the case. With the system in a home theater cabinet four meters
from the nearest listener / viewer, the damping material probably isn't
necessary.
I have decided to keep the hard drive
hard mounted in the Silverstone chassis. The HDD suspension mount device is better used in a system placed closer to the user. Even at the louder 24 dBA/1m idle and
28
dBA/1m defrag, I cannot hear the system from where I sit in that room.
However,
having the dampening material in place is a good thing as I can now access
a
quiet desktop system if I need and all I need to do is suspend the
hard drive.
* * *
Discuss this article in the SPCR Forums.
| Help support this site, buy from one of our affiliate retailers! |
|
