Zalman TNN-300 Fanless PC Enclosure System

Cases|Damping
Viewing page 7 of 10 pages. Previous 1 2 3 4 5 6 7 8 9 10 Next

CASE / SYSTEM TESTING

Test Tools

1. MIMALIST CONFIGURATION

As mentioned earlier, this configuration consisted of the following:

Each of the load tests were run for 45 minutes. The ambient temperature in the room was 20~21°C, and the noise was 19~20 dBA.

RESULTS #1: MIMALIST CONFIGURATION - <20 dBA SPL@1m
LOAD
AC POWER
CPU
MB
HDD
Idle / CnQ
57W
38°C
38°C
38°C
Idle
62W
40°C
39°C
39°C
CPUBurn
91W
51°C
50°C
40°C
VCST
95W
51°C
50°C
39°C
3DMark05
95W
50°C
48°C
40°C
Idle / CnQ: Idle with Cool 'n' Quiet enabled.
LOAD: The load on the system.
AC Power: The amount of power drawn from the wall outlet is equal to the heat inside the system.
CPU: Temperature of the processor.
MB: Readout from motherboard sensor; it may next to the Northbridge chip.
HDD: Hard drive temperature.

The results data table speaks volumes. The TNN-300 coasts, thermally speaking, with this system. It's a very efficient system, drawing only 95W from the wall at full load. This suggests that at full CPU load, the components are only drawing perhaps 70W in DC voltage, given the claimed 75% minimum efficiency of the power supply. The case remained barely warm to the touch under all conditions. It's safe to say that this system would probably be cooled safely even in tropical conditions.

If we take the 44.1W TDP for this particular A64-3000+ processor (as extracted by A64 TCaseMax v1.18) to be accurate, then we can make some predictions about how well the TNN-300 will cool hotter processors.

°C rise over ambient = 30 >>> TDP = 44.1W >>> Therefore, °C/W = 0.68

For a completely passive cooling system, it is impressive. If we wish to limit the maximum temperature of the CPU to 65°C (at the same ambient room temperature), then the TDP should not exceed 67W. This prediction jibes perfectly well with Zalman's recommendations, which call for CPUs with TDP no greater than ~70W.

Performance-wise, this system was perfectly suitable and plenty fast enough for all the usual applications ° web browsing, playback of audio / video files, email, and office applications.

While 3DMark05 is an interesting test, it is not very useful for precise GPU loading and temperature measurements. The problem is that the load it imposes, while reaching high peaks, varies up and down. More importantly, it is not possible to view the GPU temperature monitoring screen while 3DMark05 is running. Only after it is stopped can the GPU temperature be seen. Even if the delay is only a few seconds, by the time we can see the GPU temperature, it has already dropped from the highest points reached during the test.

Video Card Stability Test (VCST) is quite different. The load on the system remains steady, much like CPUBurn for the CPU. As a result, we can establish a simple time-based test (45 minutes, in this case) and run easily repeatable tests while having a constant view of the GPU temperature monitoring screen. The temperature generally stabilizes in half of that time, and it reaches higher levels than viewable with 3DMark05

The combination of sorbothane damping and the extremely quiet 2.5" drive made noise a non-issue in this configuration, even in the ultra-quiet post-midnight environment of the test room. It was possible to have the TNN-300 right next to the monitor, at head level just a couple feet away, and hardly notice the noise. Occasionally, some HDD seek noise could be heard, but it was too soft and muted to be obtrusive. There was no apparent noise of any kind from the power supply.

For the record, a Samsung Spinpoint 80GB SATA 3.5" desktop hard drive was mounted using the Zalman rubber dampers. A quick listen revealed a higher subjective level of noise. In this configuration, I would personally not want to have the PC atop the desk. The measured noise went up to 23 dBA@1m in idle and 25~26 dBA@1m during seek. Both numbers were a bit higher than measured for this drive on its own. The increase in noise is directly attributable to the conduction of vibration between the drive and the case, and to the cavity air resonance / amplification effects of the drive being in an enclosed box. The overall noise level was still quite low, but not in the champion class as with the notebook drive.


Samsung 3.5" HDD mounted as per Zalman's instructions.
With the tight spacing between the drives, two 3.5" drives might not be ideal due to heat.

2. WITH NF6800GT-256 PCIe GRAPHICS CARD

This configuration added one additional component to the above setup:

This graphics card had been modified with an Arctic Cooling NV Silencer 5 (Rev. 3) heatsink/fan, and this was the way it was first installed in the system. The Arctic Cooling heatsink/fan actually exhausts the heat of the GPU out the back via the PCI slot opening directly below the main video card slot, and in doing so, helps with the cooling of the whole case by a small degree as well. The Arctic Cooling HSF was manually limited with a rheostat to 6~7V for a very slow speed to minimize noise. In order to establish baseline performance before using the TNN-300's GPU cooling system, the AOpen card was used with the AC NV Silencer 5 heatsink/fan for about 24 hours.

The same range of tests was performed as before. Each of the load tests were run for 45 minutes. The ambient temperature in the room was 21°C, and the noise was 19~20 dBA.

RESULTS #2A: TNN-300 w/6800GT + AC NV Silencer 5 -- 27 dBA SPL@1m
LOAD
AC POWER
CPU
MB
GPU
GPU Ambient
Idle / CnQ
95W
38°C
38°C
51°C
48°C
Idle
101W
40°C
39°C
51°C
48°C
CPUBurn
133W
52°C
46°C
53°C
50°C
3DMark05
163W
47°C
46°C
63°C
62°C
VCST
146W
47°C
46°C
68°C
64°C

The hard drive temperature was left off to keep the table simple; it hovered at 41-42°C throughout testing with the NF6800GT. Note that two more items were added to the table: GPU or graphics processing unit, and GPU ambient, both from the 6800GT graphics card. The latter is from a sensor somewhere on the card that is a measure of its immediate operating temperature.

Power (and total heat in the system) nearly doubled compared to the first configuration. The CPU load temperature hardly changed, which is not surprising as the cooling systems in the case were not asked to do any more work. The GPU was cooled by its own HSF. The temperatures of the graphics card were quite low, relatively speaking. There is rarely any on-screen misbehavior until the GPU temperature approaches 80°C or higher.

The main price of the added graphics card with the AC heatsink/fan was an increase in the overall noise. Even with the fan undervolted, a chugging kind of hum prevailed. It wasn't really loud, but it was always clearly audible.

The AC NV Silencer 5 was then removed, and the TNN-300 GPU cooling system brought into play. The tests were run once again.

RESULTS #2B: TNN-300 w/6800GT -- <20 dBA SPL@1m
LOAD
AC POWER
CPU
MB
GPU
GPU Ambient
Idle / CnQ
94W
38°C
38°C
51°C
48°C
Idle
99W
40°C
39°C
51°C
48°C
CPUBurn
132W
52°C
46°C
53°C
50°C
3DMark05
163W
47°C
46°C
63°C
62°C
VCST
145W
51°C
50°C
71°C
74°C

A few changes in results can be seen. The most salient ones showed up in the VCST testing, which are highlighted below.

COMPARISON #2C: Video Card Stability Test, 144W AC load
CPU
MB
AC NV Silencer 5
CPU
MB
TNN-300
GPU
GPU Ambient
SPL@1m
GPU
GPU Ambient
SPL@1m
47°C
46°C
68°C
64°C
27 dBA
51°C
50°C
71°C
74°C
<20 dBA

First, the CPU and MB temperatures rose to about the same level as during CPUBurn. This may have been the result of GPU heat no longer being evacuated by the AC NV Siliencer 5 heatsink/fan.

Secondly, the GPU rose by a couple of degrees, which is of little consequence. GPU ambient temperature rose substantially, however, by 10°C. This increase is not trivial, and suggests that heat is being trapped beneath the video card. Still, there was no misbehavior of any kind visible during the test. Also, keep in mind that the heat dissipation of the NF6800GT is about twice Zalman's official recommended level.

FAN MOD

A final impromptu mod involved jamming a Scythe 80mm fan under the NF6800GT card, as close to the PCI slot covers as possible. This was with all the PCU slot covers removed, and with the fan blowing out. The very quiet fan (subject of a review soon to come) was undervolted to ~6V. A piece of double sided sticky tape held it to the floor of the case. Here's the only photo taken of this "mod", taken after the system was being dismantled:


Impromptu exhaust fan for the VGA card area.

What was the effect of this fan?

  • Noise went up slightly to 20~21 dBA @1m. It was less audible from the front than from the back or side, for obvious reason. Subjectively, it was much nicer than the Arctic Cooling NV Silencer 5; it measured considerably lower, too.
  • Temperatures dropped substantially. See the comparison table below.
COMPARISON #2C: Video Card Stability Test, 144W AC load
CPU
MB
TNN-300 + 80mm fan
CPU
MB
TNN-300
GPU
GPU Ambient
SPL@1m
GPU
GPU Ambient
SPL@1m
48°C
47°C
68°C
66°C
20-21 dBA
51°C
50°C
71°C
74°C
<20 dBA


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

Cases|Damping - Article Index
Help support this site, buy the Zalman TNN-300 from one of our affiliate retailers!