Antec NSK3300: Quiet Out-of-the-Box

THERMAL & ACOUSTIC TESTING

Thermals and noise comprise the core of most SPCR equipment reviews. Two variations of the same system were installed and tested in the Antec NSK3300. The base components are listed below. They are the same components used in our recent reviews of the Zalman HD160 home theater case and the Lian Li PC-101, so these cases will be used as reference points. Note that, because the NSK3300 includes a power supply, the Seasonic S12-330 was not (and could not) be used. The stock power supply is louder than the S12, and this probably affects the results.

DFI RS482 Infinity MicroATX motherboard
This new ATI Radeon Express 200 chipset model from DFI has the most flexible and user-adjustable BIOS we've seen on any microATX board, comparable to the best of the full-ATX boards. It allows the CPU core voltage to be manually set without disengaging Cool'n'Quiet, which simply applies the manual voltage adjustment to the various CPU power states. It allowed the X2 4800+ to be undervolted by 0.1V throughout the testing, for very modest power consumption in every load. It has no fans.

AMD Athlon 64 X2 4800+ processor
Not long ago, this was AMD's current second fastest desktop processor, one small step down from the flagship FX-60. This particular dual-core sample has a rated TDP of 85W. Previous testing showed it easily undervolts by 0.1V or more, with resulting power draw at full load of just ~60W at the 2x12V motherboard socket.

OCZ Technology Gold PC4000 2 x 512MB DDR matched dual channel memory.

Samsung SP2504C 250GB SATA 3.5" hard drive
A quiet 3.5" desktop reference. It measures 21~22 dBA@1m.

Other components included:

AOpen Aeolus PCX6800GT-DVD256 video card with Zalman VF900 VGA cooler at 5V. The SPL of this HSF at 5V measures 20 dBA@1m. It sounds a bit like a whispery rubbing of paper.

Zalman CNPS9500 CPU HSF
A heavy duty heatsink for a hot processor, this is an effective cooler even undervolted to 5V. It measures 23 dBA@1m.

Windows XP Pro SP2 was installed and fully updated, and our usual gamut of software tools installed:

• SpeedFan 4.28 for CPU and other hardware monitoring.
• CPUBurn for processor stress testing.
• ATI Tool provides a steady high load to the GPU in a reduced window, allowing other tools to be in use at the same time.
• RivaTuner allows the core temperature of the GPU to be monitored over time.

Other tools:

• Extech Power Analyzer / Data Logger 380803 power meter, for precise, unmanned, long term AC power monitoring.
• SPCR updated digital audio recording system to record and save noises in high quality MP3 files.
• B&K 2203 Sound Level Meter to measure sound pressure levels.

TEST RESULTS

Ambient conditions were 24°C and 18 dBA. It is summer, and the lab is a few degrees warmer than in winter. This has an effect on both thermals and noise, especially thermally speed-controlled fans.

The system was initially configured without the VGA card installed so that the system could be as quiet as possible. The CPU fan was undervolted to 5V, and the system fan was set to "Low". Various other configurations were tried in turn to see how the case responded to changes in airflow and system heat. Details of each configuration are listed in the table below.

Configuration #1

The dominant sources of noise in the system were the system fan and the power supply. Even so, the system was surprisingly quiet without requiring any special tinkering. Unlike most cases where we have to undervolt the stock fans to make the system even moderately quiet, the NSK3300 achieved a noise level of 24 dBA@1m without any special modifications.

The noise character was also quite good. Most of what we could hear was the soft whoosh of airflow; aside from a low growl, there were few pure tones to speak of. Much of the noise at idle came from the TriCool fan.

With the CPU under load, the roles of the two noisemakers was reversed: The power supply became more dominant and noticeable as its fan increased in speed, while the system fan receded into the background. The quality of noise did not change appreciably; the deep, humming growl may have gotten slightly louder, but the dominant noise was still the whoosh of airflow at a slightly more urgent pace.

We were pleased to note that, although the power supply did begin to get noisier when the system was under load, the increase was limited to a relatively minor 2 dBA@1m. Our test processor, an AMD X2 4800+, is one of the hottest AMD processors currently on the market, and represents far more heat that is likely to be generated by a midrange system. It should not be difficult to select a slower processor that does not cause an increase in noise.

Thermally, the NSK3300 was good enough to prevent our processor from overheating, which is really all that is needed. Comparisons with the Lian Li PC-101 and the Zalman HD160 were too close to call given slight variations in configuration, noise level, and ambient temperature. All three cases ended up with the CPU in the high 50's — an expected result given the combination of processor and heatsink.

Configuration #2

It is difficult to know how to improve the NSK3300 beyond its stock configuration. Pushing the noise level to below the ambient noise level would require several modifications that are beyond the scope of what most users are willing to do. However, there was one easy modification we could perform: Swapping the system fan for a Nexus 120 mm at 5V. This effectively takes the system fan out of the equation as far as noise is concerned, as the fan is practically inaudible from a distance of one meter.

This change had very little effect on the noise, although we did measure a drop of ~1 dBA@1m. Subjectively, the most significant change was not the volume but quality of noise, which was no longer so turbulent. The lack of air noise uncovered the low hum that could be heard in the previous configuration, bringing it to the forefront. Overall, we were not sure whether or not the change was an improvement; the slight decrease in noise was paid for in a noise character that was rougher and more irritating.

One thing we were sure of was that the much slower fan had a significant effect on temperatures. The processor temperature rose to 60°C, right on the cusp of overheating. For some reason, though, the power supply did not seem to ramp up as much during this test. The result is reflected in the lower measured noise level under load.

Configuration #3

Our last configuration made a hot system even hotter by throwing a video card into the mix — a GeForce 6800GT cooled by a Zalman VF900 at 5V. Like the CPU cooler, this cooler is quieter than the rest of the system, and is unlikely to affect the noise level in the system. The new graphics card doubled the amount of heat in the system at idle — the system now drew 85W from the wall instead of 42W.

At idle, this change had no effect on noise at all; the measured noise jumped back up to 24 dBA@1m, but we did not notice any subjective change. However, this was not the case under load, when the addition power required by the card caused the power supply to ramp up to the point where it was difficult to consider the system quiet any longer. Perhaps this change was not all that significant; after all, the only time a 3D card is likely to be used is during gaming, when game sound is likely to drown out the sound of the computer in the background.

Adding the new graphics card had a significant effect on cooling. The CPU jumped up to 66°C — stable, but not a safe temperature for long-term use — even when the graphics card was not in use. Presumably, upping the voltage on the Nexus fan would be enough to reduce the temperatures to an acceptable level; an increase to 7V or 9V could probably have been done without affecting noise.

The video card itself showed every indication of being properly cooled. As with the CPU in Configuration #1, the thermal results for the VGA card were too close to the Lian Li and Zalman cases to draw any conclusions about which case is better for cooling.

(Editor's Note: These are only a few of the many configurations that could be used. With three fan mounting positions, the number and variety of ways to configure this case is limited only by the user's imagination. Setting a Nexus fan to 7V, for example, or adding one or two quiet 92mm input fans.)

MP3 RECORDINGS

Antec NSK3300, Config 1 (System Fan @ L): 24 dBA@1m: One Meter, One Foot

Antec NSK3300, Config 2 (Rear Fan swapped to Nexus @ 5V): 23 dBA@1m: One Meter, One Foot

COMPARATIVES

Lian Li PC-101, Config 1 (No Intake Fan): 24 dBA@1m: One Meter, One Foot

Lian Li PC-101, Config 2 (Intake Fan @ 5V): 26 dBA@1m: One Meter, One Foot

CONCLUSIONS

Whether or not the NSK3300 is appropriate for you depends on how well you fall into the mainstream market segment that Antec is targeting. If you're building a monster gaming rig, a stealthy recording machine or an blistering overclocker, the NSK3300 isn't for you.

On the other hand, if you're building an ordinary system that needs to be quiet — not silent — and you need to do it on a budget, the NSK3300 could be just the thing. An out-of-the-box noise level of 24 dBA@1m (matched with the proper components, of course) is very competitive, especially considering that the power supply chamber can probably prevent the power supply from ever ramping up in an ordinary system.

Size is also likely to play a factor. The NSK3300 is MicroATX only, which means that prospective users will need to be interested in building a small system — without resorting to a SFF system.

If the case's biggest strength is its noise level out-of-the-box, its weakness is its out-of-the-box power supply which has trouble staying quiet under heavy load and short cables. Given the dearth of good quality, quiet SFX-compatible power supplies, it's probably wise to avoid planning a power-hungry system around the NSK3300. (Editor's Note: On the other hand, the too-small top panel vent for the top chamber and higher than normal ambient temperature in the lab might be responsible for much of the PSU fan's ramp up.)

The NSK3300 is a good low-cost quiet case option. It can currently be found starting from ~US$65 — pretty good for a case and power supply combo. It isn't the quietest or the coolest case that we know of, but all of the alternatives are significantly more expensive. For many users, 24 dBA@1m out-of-the-box might be good enough for its price. It's an intelligent and logical addition to Antec's line of cases.

Many thanks to Antec for supplying the NSK3300 sample.

* * *

Articles of Related Interest

Zalman HD160 Home Theater PC Enclosure
Lian Li PC-101: Aluminum *Can* be Quiet!
Antec P180: The Whole Nine Yards
Antec P150 mid-tower case w/ Neo HE 430 PSU
Antec SLK3000B mid-tower case
Cases: Basics and Recommendations

* * *

Discuss this article in the SPCR Forums.