You are here

SPCR's Fan Round-Up #2: 120mm Fans

November 28, 2006 by Devon
Cooke

May 5, 2008

Our fan airflow measurement system has recently undergone a major
revision to improve accuracy and repeatability. We've updated airflow data
for some but not all fans; only fans that fared well acoustically were retested
with the new system. There will be a new methodology article coming soon.

Although 120mm fans are now commonplace, it wasn't long ago that 80mm fans
were standard and a case that accepted larger fans was worthy
of note on a sticky in our forums
. The advantage of the larger size should
come as no surprise to most: A larger fan produces more airflow. A larger fan
means that hot components can be kept cooler, allowing greater headroom for
overclocking. Or, from a silencing perspective, a larger fan can provide the
same cooling at a lower speed, and thus a lower noise level.

We usually recommend using a 120mm fan if possible — but which one to
choose? Our first fan round-up
was published about two weeks ago, but that only looked at 80mm fans. Now, we've
examined another half dozen (or a dozen, depending on how you count) fans,
a surprising number of which are suitable for a quiet system.

As with the last roundup, we started with the Nexus that has been our fan
of choice for the past two years. We also looked at an Antec Tri-Cool (which
comes stock in many Antec cases), some much hyped fans from Scythe and
Noctua, a highly reputed Papst, and a couple dark horses from ARX.

This roundup is primarily a summary of our test results with a few interesting
tidbits about each fan thrown in. We have kept theory to a minimum, so you do
not need to know how a fan works to get the most out of this article. You need
to know two things:

  1. Fans are designed to push air — the faster the fan, the more air it
    pushes
  2. Fans produce noise — the faster the fan, the more noise it produces

For our purposes, the best fan is the one that pushes the most air for the
least noise. For users who are interested, a more technical discussion
of fan technologies can be found in our recent article, Anatomy
of A Silent Fan
. Users who want to know exactly how the fans were tested
should refer to our test methodology
article
. The rest of you: Sit back and enjoy! We hope you find our work
useful.



A large pile of 120mm fans, not all of which made it into this test.

HOW TO USE THIS REVIEW

Each fan in this roundup has its own data table and write-up that summarizes
what we learned about it. Use these to find specific information about the fan
you're looking for. In addition, every fan was recorded four times, according
to our standard Audio Recording
techniques.
These recordings can be used to make A/B comparisons between
fans to help illustrate the differences between them. The four recordings are
as follows:

  1. Alternating ambient noise and the fan running at 5V, 7V, 9V, and 12V,
    recorded at a distance of one meter.
  2. Alternating ambient noise and the fan running at 5V, 7V, 9V, and 12V,
    recorded at a distance of one foot (30 cm).
  3. Five seconds of ambient noise, followed by the fan running in the constant
    airflow test, recorded at a distance of one meter.
  4. Five seconds of ambient noise, followed by the fan running in the constant
    airflow test, recorded at a distance of one foot (30 cm).

As always, we recommend that you listen and compare the recordings in a specific way. The green box below describes how we make our recordings and what you're supposed to do with them.

At the end of the roundup is a conclusion that summarizes the best and the
worst that we found. This is where to look if you just want to cut to the chase
and find out which fan we liked best.

HOW TO LISTEN & COMPARE

These recordings were
made with a high resolution, studio quality, digital recording system,
then converted to LAME 128kbps encoded MP3s. We've listened long and
hard to ensure there is no audible degradation from the original WAV
files to these MP3s. They represent a quick snapshot of what we heard
during the review. Two recordings of each noise level were made, one
from a distance of one meter, and another from one
foot
away.

The one meter recording
is intended to give you an idea of how the subject of this review
sound in actual use — one meter is a reasonable typical distance
between a computer or computer component and your ear. The recording
contains stretches of ambient noise that you can use to judge the
relative loudness of the subject. For best results, set your volume
control so that the ambient noise is just barely audible. Be aware
that very quiet subjects may not be audible — if we couldn't
hear it from one meter, chances are we couldn't record it either!

The one foot recording
is designed to bring out the fine details of the noise. Use this recording
with caution! Although more detailed, it may not represent how the
subject sounds in actual use. It is best to listen to this recording
after you have listened to the one meter recording.

More details about how
we make these recordings can be found in our short article: Audio
Recording Methods Revised
.

THE FANS

The following fans were included in the roundup:

The Conclusion can be found on page 6.

NEXUS REAL SILENT CASE FAN D12SL-12

Ambient noise at the time of testing was 19 dBA.



The Nexus comes in two unusual color schemes.

Brand Nexus Power Rating 0.30A
Manufacturer Yate
Loon
Airflow Rating 36.87 CFM
Model Number D12SL-12 RPM Rating 1,000 RPM
Retail Availability Yes Noise Rating 22.8 dBA
Bearing Type Sleeve Header Type 3-pin & Molex
Hub Size 1.58" Starting Voltage 5.5V
Frame Size 120 x 120 x 25 mm Number of Samples 4

Our thanks to EndPCNoise for supplying these samples.
Voltage
Noise
RPM
CFM
Power
12V
1080 RPM
47 CFM
0.95W
9V
850 RPM
35 CFM
0.75W
7V
680 RPM
27 CFM
0.62W
5V
490 RPM
16 CFM
0.51W
@25 CFM (6.6V)
640 RPM
25 CFM
0.60W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1080 RPM
29 CFM
0.95W
9V
850 RPM
23 CFM
0.75W
7V
680 RPM
19 CFM
0.62W
5V
490 RPM
13 CFM
0.51W
@20 CFM (7.6V)
740 RPM
20 CFM
0.59W

Nexus' 120mm fan (and the Yate
Loon models
that it is derived from) is one of the most popular fans among
SPCR's regular readers, and for good reason. It may be a little on the expensive
side, as individually packaged fans often are, but the combination of a decent
noise level at maximum speed, a smooth, low frequency noise character, and its
ability to be inaudible at lower voltages make it a well loved favorite.

Despite spinning more slowly, the 120mm Nexus is slightly louder than its 80mm
counterpart at full speed. Presumably, the heavier fins require a more powerful
motor to drive them, or perhaps their larger surface area produces more turbulence
noise. Fortunately, it also blows double the amount of air, which more than
makes up for the noise difference. This is a prime example of why we recommend
120mm fans!

The point of inaudibility is somewhere around ~8V, so many users use the
7 volt trick
for the best compromise between noise and airflow. We do not
recommend running the fan at 5V, as some samples have trouble starting at this
level.

As mentioned, the noise character is very low and smooth. This is ideal, since
the lower frequencies do not cover distance easily. From most usable distances,
the noise is a slight whoosh that is easy to ignore. A word of caution: Much
of the low frequency noise is present as vibration as well as sound and can
cause resonance if it is hard-mounted. At lower fan speeds, the vibration is
so low that this is not a concern, but those using the fan at high speed may
find that soft-mounting with silicone grommets may help cut down the low frequency
rumble that this fan produces.

Some users who care about appearances have shunned the Nexus for its bright
orange color. We're not sure what prompted Nexus to choose orange, of all colors,
but there has been enough outcry that Nexus now offers a black-and-white version
with the same noise characteristics as the original.

We did not notice any significant variation that could be tied to color, but
we did notice that sample variance seemed unusually high, and not all samples
we heard sounded as good as the best. One sample we heard just seemed to be
louder overall without any identifiable change in noise character, while another
had a slight ticking that marred the smoothness of the noise. Other users have
also reported some problem samples with ticking. The difference is not large,
and at some speeds it may be insignificant, but it is enough to be noticeable
in a quiet system.

Aside from that, the only disadvantages are shared with Nexus' 80mm fan: Closed flange screw holes that make wire clips unusable without modification, and sleeve bearings
that are inappropriate for horizontal mounting or high heat situations. Together,
these ensure that, while the Nexus is an excellent choice for a simple case
fan, it is less suitable for mission-critical operations such as a CPU heatsink
or a hot power supply.



The closed flange can make mounting a bit tricky.

Noise Recordings

ARX FD1212-A Series

Ambient noise at the time of testing was 19 dBA.

LOW SPEED (FD1212-A3053E)
Brand ACT-RX Power Rating 0.16A
Manufacturer ACT-RX Airflow Rating 70.14 CFM
Model Number FD1212-A3053E RPM Rating 2,000 RPM
Retail Availability Limited Noise Rating 39 dBA
Bearing Type CeraDyna A Header Type 3-pin
Hub Size 1.99" Starting Voltage 4.5V
Frame Size 120 x 120 x 25 mm Number of Samples 2
Voltage
Noise
RPM
CFM
Power
12V
1430 RPM
57 CFM
0.85W
9V
1050 RPM
42 CFM
0.56W
7V
770 RPM
29 CFM
0.43W
5V
470 RPM
16 CFM
0.31W
@25 CFM (6.4V)
680 RPM
25 CFM
0.39W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1430 RPM
40 CFM
0.85W
9V
1050 RPM
25 CFM
0.56W
7V
770 RPM
19 CFM
0.43W
5V
470 RPM
11 CFM
0.31W
@20 CFM (7.6V)
800 RPM
20 CFM
0.45W



A small thermal diode is all that distinguishes the thermally controlled
version from the regular one.

MEDIUM SPEED, THERMALLY CONTROLLED (FD1212-A2060E)
Brand ACT-RX Power Rating 0.27A
Manufacturer ACT-RX Airflow Rating 87.02 CFM
Model Number FD1212-A2060E RPM Rating 2,500 RPM
Retail Availability Limited Noise Rating 46 dBA
Bearing Type CeraDyna A Header Type Molex pass-through
Hub Size 1.99" Starting Voltage 5.7V
Frame Size 120 x 120 x 25 mm Number of Samples 2
Voltage
Noise
RPM
CFM
Power
12V, Thermistor Short-Circuited
2440 RPM
99 CFM
3.31W
9V, Thermistor Short-circuited
1900 RPM
77 CFM
2.24W
7V, Thermistor Short-circuited
1470 RPM
58 CFM
1.63W
5V, Thermistor Short-circuited
980 RPM
39 CFM
1.06W
@25 CFM (4.6V)
710 RPM
25 CFM
0.74W
@Default Speed (6.1V)
1270 RPM
51 CFM
1.21W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
2440 RPM
79 CFM
3.31W
9V
1900 RPM
57 CFM
2.24W
7V
1470 RPM
41 CFM
1.63W
5V
980 RPM
24 CFM
1.06W
@20 CFM (4.4V)
810 RPM
20 CFM
0.72W
@Default Speed (6.1V)
1270 RPM
34 CFM
1.21W

You've probably never heard of ACT-RX, its parent ARX, or its CeraDyna bearings.
As a Taiwanese OEM that generally sells directly to system integrators, that's
really no surprise, but their
web site
is chock full of interesting (if poorly translated) technical information
about fans. Most of the information relates to CeraDyna, a proprietary ceramic
sleeve bearing, but a significant part of that information is specifically about
fan noise. In fact, along with reliability, noise is consistently mentioned
as one of the primary advantages of the CeraDyna bearings.

Naturally, this piqued our interest, so we were happy to take a look when we
acquired some samples. Unfortunately, our happiness evaporated as soon as we
started them up. Reliable they may be (we can only assume), but quiet? Sorry.
These ARX fans may well be the worst sounding fans we've ever tested.

We looked at two models: A "low speed" model, and a "medium
speed" model with thermal control. Although it is the slowest fan they
offer, the "low speed" model is hardly low speed by SPCR standards;
it's rated for 2,000 RPM and 39 dBA (uh oh). Strangely enough, both of our low
speed samples seemed to run at ~1,500 RPM, making us wonder whether our fans
were tweaked in any way. Slower or not, the clatter it produced did not disappear
even at 5V.

The thermally controlled version was even worse. In our 21°C test lab,
it spun at ~1,300 RPM by default. That's 30% more than the Nexus blows at full
tilt, and the 30 [email protected] noise level was completely unacceptable. Even when the
thermistor was short-circuited and the input voltage was turned down to match
our 25 CFM target, the noise level never dropped below 25 [email protected] Our expectation
is that a good fan should be nearly inaudible at this level, and this fan clearly
failed to live up to it.

There's not really much more to say. We could go one and mention the poor noise
character and describe it in detail, but why make a bad reputation worse? If
you really want to know what the fans sound like, feel free to check the recordings
below. Otherwise, pretend you've still never heard about them and continue
reading on the next page...

Noise Recordings

JUMP TO:

EBM PAPST 4412 SERIES

Ambient noise at the time of testing was 19 dBA.

Low Speed: 4412 F/2GLL
Brand ebmPapst Power Rating 0.72W (0.06A)
Manufacturer ebmPapst Airflow Rating 70m³/h (41.17cfm)
Model Number 4412 F/2GLL RPM Rating 1,300 RPM
Retail Availability Yes Noise Rating 18 dBA
Bearing Type Sintec Sleeve Header Type 3-pin
Hub Size 1.86" Starting Voltage 3.5V
Frame Size 120 x 120 x 25 mm Number of Samples 2

Our thanks to Craig in NJ and Ichbinleise for supplying these samples.
Voltage
Noise
RPM
CFM
Power
12V
1290 RPM
54 CFM
0.64W
9V
980 RPM
40 CFM
0.48W
7V
750 RPM
30 CFM
0.38W
5V
490 RPM
18 CFM
0.30W
@25 CFM (6.2V)
660 RPM
25 CFM
0.35W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1290 RPM
34 CFM
0.64W
9V
980 RPM
23 CFM
0.48W
7V
750 RPM
18 CFM
0.38W
5V
490 RPM
11 CFM
0.30W
@20 CFM (7.4V)
780 RPM
20 CFM
0.41W



Papst has evolved into ebmPapst, but the 4412 series hasn't changed.

Medium Speed: 4412 FGL
Brand ebmPapst Power Rating 1.25W (0.105A)
Manufacturer ebmPapst Airflow Rating 94m³/h (~55 CFM)
Model Number 4412 FGL RPM Rating 1,600 RPM
Retail Availability Yes Noise Rating 26 dBA
Bearing Type Sintec Sleeve Header Type Molex pass-through
Hub Size 1.86" Starting Voltage 3.4V
Frame Size 120 x 120 x 25 mm Number of Samples 1

Our thanks to EndPCNoise for supplying this sample.
Voltage
Noise
RPM
CFM
Power
12V
1610 RPM
69 CFM
1.19W
9V
1260 RPM
53 CFM
0.86W
7V
980 RPM
40 CFM
0.66W
5V
670 RPM
26 CFM
0.48W
@25 CFM (4.9V)
660 RPM
25 CFM
0.48W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1610 RPM
45 CFM
1.19W
9V
1260 RPM
32 CFM
0.86W
7V
980 RPM
23 CFM
0.66W
5V
670 RPM
16 CFM
0.48W
@20 CFM (6.1V)
810 RPM
20 CFM
0.57W

Papst fared poorly in
our last fan roundup
, despite great expectations and many recommendations
via word of mouth. In fact, our experience was so contrary to expectations that
we had to wonder whether we'd looked at the wrong fan. And, when German web
shop Ichbinleise offered
to send us a pair of low speed Papsts, we jumped at the chance to let Papst
redeem itself.

The 4412 F/2GLL model is a special low-speed version of the 4412 series. It
is not listed on any of Papst's numerous web sites, but a quick Google search
revealed numerous web shops that stock it — mostly silence oriented shops
located in and around Germany, where Papst is headquartered. We also threw a
faster 4412 FGL into the mix, which is why two sets of results appear above.
Although it bears the older Papst logo without the "ebm" prefix (marking
it as a pre-2003 model), it looks identical to the slower versions and sounds
similar enough that we feel comfortable mixing it in as an example of how the
4412 series performs one the whole.

A quick examination leaves the impression that this is a very serious fan.
It's heavier than most 120mm fans, and it has a metal axle embedded in the center
of the hub. It also spins very smoothly; when the power is cut, the rotor glides
easily to a halt rather than bouncing to a sudden stop when one of the magnets
in the motor assembly catches it. Perhaps this is why the starting voltage is
so low; a smoother rotor probably requires less force to put in motion than
one that has to overcome a static magnetic field.

Surprisingly, it appeared to blow slightly less air per rotation than the Nexus,
but the difference was not enough to make much difference to noise. The two
were more or less the same volume at a given speed, so it came down to noise
character to decide between them. And, while the Papsts were smooth (much smoother
than the 80mm model we listened to), the Nexus was smoother still. The Papsts
had a slight chatter or ticking that was slightly more pronounced in our two
low speed samples. From a distance and at higher speeds, the roughness in the
noise seemed to smooth out a little, but it never achieved the barely-there
hum that has made the Nexus so popular.

Regardless, the Papsts still managed to be inaudible at 25 CFM, so they are
quite usable. The low starting voltage is also an asset, as it allows for a
larger range of voltage adjustment. We are sure there are many users who will
be happy to trade a tiny bit of noise for Papst's reputation for quality and
longevity — or perhaps just on the basis of Papst's superior Sintec Sleeve
bearing.

Noise Recordings

ANTEC TRI-COOL 120

Ambient noise at the time of testing was 19 dBA.

Brand Antec Power Rating ?
Manufacturer Dynatron Airflow Rating 39 / 56 / 79 CFM
Model Number TriCool 120mm RPM Rating 1,200 / 1,600 / 2,000 RPM
Retail Availability Yes Noise Rating 25 / 28 / 30 dBA
Bearing Type Ball? Header Type Molex / 3-pin Combo
Hub Size 1.77" Starting Voltage 3.2V @ High
Frame Size 120 x 120 x 25 mm Number of Samples 8
Voltage
Noise
RPM
CFM
Power
12V (High)
1930 RPM
73 CFM
2.73W
9V
1530 RPM
58 CFM
2.00W
Medium
1430 RPM
53 CFM
1.90W
7V
1220 RPM
45 CFM
1.54W
5V (Low)
870 RPM
30 CFM
1.18W
@25 CFM (4.5V)
740 RPM
25 CFM
1.00W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V (High)
1930 RPM
53 CFM
2.73W
9V
1530 RPM
41 CFM
2.00W
Medium
1430 RPM
36 CFM
1.90W
7V
1220 RPM
30 CFM
0.54W
5V (Low)
870 RPM
21CFM
1.18W
@20 CFM (4.9V)
850 RPM
20 CFM
1.18W

Antec's TriCool fans are fairly well known, though not specifically because
they are quiet. Rather, it is because nearly every case that Antec has released
since the P180 has included
them as stock fans. As stock fans go, they're fairly quiet thanks to a three-speed
switch that allows the fan speed to be reduced to low noise levels. The switch
allows users to choose between noise and cooling, but we don't recommend using
anything except Low unless undervolting. It's good for Antec as well: It allows
them to cater to two different market segments — overclockers and silencers
— with a single product.

In addition to showing up in Antec's cases, TriCools are also available packaged
individually. Noise, speed and airflow are all identical in the retail version,
but there is one trait that sets them apart: The retail fans are made from clear
plastic — a nod to yet another market segment: Bling. We tested three samples
with black frames that came from various cases that we had lying around, plus
another five clear ones, and came to the conclusion that there was no audible
variation between the black and the clear versions. However, there are two variations
that we did not test: TriCool LED, and TriCool DBB (Double Ball Bearing). The
LED version should not sound any different from our samples, but this is not
true of the DBB version, as bearing design contributes significantly to a fan's
noise signature.



TriCool fans get their name from this three-speed switch.

Antec is a well known retail brand, so it's no surprise that the TriCool is
widely available. However, some may be surprised that it is not actually made
by Antec. Like many products in the tech industry, it is built to Antec's specifications
by a Taiwanese company that is only too happy to supply the kind of volume that
Antec can move. In this case, the company is Dynatron,
although the fan itself does not show up on Dynatron's product list. As far
as we know, Dynatron does not produce the TriCool for anyone but Antec, so it's
no surprise that they don't list it. However, it does make collecting specifications
a bit more difficult, since Antec's specifications lack details such as bearing
type and power rating.

At High speed, the TriCool was about as noisy as you'd expect of a fan spinning
at 2,000 RPM. It moved a lot of air, and it was definitely not quiet. The Low
and Medium speeds are of more interest, especially to a casual user who appreciates
the convenience of being able to make his fans quieter without purchasing a
separate fan controller or modifying his system in other ways. At Medium, the
fan was quieter, though still a long way off of inaudible, and most users will
probably choose to stick with the Low setting.

The Low setting was quite impressive; it deserves mention for having one of
the quietest "stock" speeds of any fan we've tested, although the
airflow drops off proportionally. Nevertheless, 20 [email protected] is quiet enough that
some users may find it inaudible, especially in an office environment where
the ambient noise tends to be fairly high.

Noise quality was also fairly good. Like the Papst 4412, the TriCool was smooth
at high speeds, but developed some roughness when slowed down. In this case,
the roughness took the form of a chuffing noise that was clearly audible in
the Low speed setting and even when the fan was undervolted for the test at
25 CFM.

Unfortunately, this acoustic artifact at low speeds is enough to put the TriCool
behind the Nexus and the Papst; it is audible no matter what speed the fan is
operating at, which makes the TriCool unsuitable for users who want complete
silence. But that should not take away from its good points: It is cheap, widely
available, and is quiet enough for many casual users.

Noise Recordings

JUMP TO:

NOCTUA NF-S12 SERIES

Ambient noise at the time of testing was 19 dBA.



The retail package comes with lots of goodies.

Fast Version: NF-S12-1200
Brand Noctua Power Rating 0.12A / 1.44W
Manufacturer ? Airflow Rating 81 m³/h (48 CFM)
Model Number NF-S12-1200 RPM Rating 1,200 RPM
Retail Availability Yes Noise Rating 17 dBA
Bearing Type SSO Bearing Header Type 3-pin
Hub Size 1.63" Starting Voltage 2.9V
Frame Size 120 x 120 x 25 mm Number of Samples 2
Voltage
Noise
RPM
CFM
Power
12V
1250 RPM
63 CFM
1.32W
9V
1020 RPM
52 CFM
1.00W
7V
840 RPM
42 CFM
0.79W
5V
640 RPM
30 CFM
0.60W
@25 CFM (4.3V)
570 RPM
25 CFM
0.53W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1250 RPM
38 CFM
1.32W
9V
1020 RPM
28 CFM
1.00W
7V
840 RPM
23 CFM
0.79W
5V
640 RPM
18 CFM
0.60W
@20 CFM (5.8V)
750 RPM
20 CFM
0.68W


Slow Version: NF-S12-800
Brand Noctua Power Rating 0.05A / 0.60W
Manufacturer ? Airflow Rating 59 m³/h (35 CFM)
Model Number NF-S12-800 RPM Rating 800 RPM
Retail Availability Yes Noise Rating 8 dBA
Bearing Type SSO Bearing Header Type 3-pin
Hub Size 1.63" Starting Voltage 3.0V
Frame Size 120 x 120 x 25 mm Number of Samples 1
Voltage
Noise
RPM
CFM
Power
12V
820 RPM
41 CFM
0.59W
9V
660 RPM
31 CFM
0.48W
7V
530 RPM
24 CFM
0.40W
5V
390 RPM
16 CFM
0.34W
@25 CFM (7.1V)
540 RPM
25 CFM
0.41W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
820 RPM
22 CFM
0.59W
9V
660 RPM
18 CFM
0.48W
7V
530 RPM
12 CFM
0.40W
5V
390 RPM
8 CFM
0.34W
@20 CFM (10.5V)
750 RPM
20 CFM
0.53W

May 5, 2008

Noctua's unusual fin design did not fare well in our new, more accurate
test procedure. Please evaluate the new results carefully.

Noctua's unusual NF-S12 fans have been subject to a lot of hype. People have
been asking about these fans since before they were available anywhere, and
their odd color, unusual blade geometry and Noctua's canny PR have all done
their job to build expectations. The fact that they were developed by "the
Austrian Institute for Heat-Transmission and Fan Technology" and are rated
for absurdly low decibel counts probably helps as well. When reacting to hype
like this, a little cynicism is definitely in order, but there is room for some
optimism as well. The unique design of the fan makes it difficult to predict
what it will sound like without hearing it, but one thing is clear: Noctua is
doing more than just paying lip service to the concept of quiet computing. The
fan is like no other on the market, which means that somebody, somewhere must
have put some engineering into it. The tough question is this: Was the engineering
effective?

Signs are hopeful. The retail package doesn't skimp on the kind of extras that
appeal to silencers: Four firm silicone grommets for soft-mounting, and an adapter
("U.L.N.A.") with some kind of resistor or diode that can be installed
to drop fan speed to approximately half of stock.

The two available models spin at 1,200 and 800 RPM, and, with the help of the
U.L.N.A. adapter, that can be dropped to 600 or 500 RPM respectively. Our actual
tests showed the faster fan dropping from 1,250 to 640 RPM (~4.7V), while the
slower fan dropped from 820 to 550 RPM (~6.9V).



Accessories include silicone grommets, a Molex to 3-pin adapter, and U.L.N.A.:
"Ultra Low Nois
e Adapter".

Initially, we were quite skeptical that the unusually shaped fan blades could
live up to their claim to provide better than usual airflow at low speeds. A
similar blade design on an
80mm Mechatronics fan
ended up being significantly less efficient
when we compared it to more conventional designs. Additionally, the unusually
wide clearance between the blade tip and the frame — intended to reduce
turbulence noise — made us wonder whether the blades had enough surface
area to move a significant amount of airflow.

Happily, the numbers proved Noctua's claims correct.
A close analysis of the data showed that, rotation for rotation, the Noctua
fans pushed about 20% more air than the rest of the fans in the test. Even better,
the Noctua fans did not lose efficiency until the rotation speed dropped to
~400 RPM, while other fans typically dropped off around ~500 RPM.

So much for airflow, but what about noise? Certainly, the Noctua fans had the
advantage of spinning more slowly to produce the same airflow, but we've long
since learned not to judge noise on the basis of rotation speed alone. Luckily,
the noise character seemed to consist mostly of the quiet, broadband whoosh
of airflow, making it very easy to ignore. Close listening revealed a very slight
buzz hidden underneath that gave the noise a slight edge that is absent in the
Nexus fan. In contrast, the Nexus sounded more tonal, with more emphasis in
the deeper frequencies.

Best of all, the better airflow efficiency meant that the Noctua was perceptibly
quieter at equivalent airflow than the Nexus. The difference wasn't large, and
at low speeds both were quiet enough to disappear entirely. Nevertheless, there
was enough difference around ~40 CFM for this reviewer to give the Noctua the
nod.

One final characteristic of the noise deserves some special mention. The Noctua
became inaudible relatively early in the test, meaning that it was impossible
to distinguish it from the background while it was running. But... the fan seemed
to have a subconscious "presence" that hung over the room even when
the fan was inaudible. The "presence" was noticeable as a change in
the background noise when the fan was turned on or off. The change was felt
more than heard; it was impossible to put a finger on just what changed, but
there was no doubt that something had. In short, Noctua did a very
good job of finding a noise character that blends easily into the background.
The subjective "quietness" of the fan can be attributed as much to
the fact that it is difficult to localize as to the amount of noise that
it generates.

Noise Recordings

JUMP TO:

SCYTHE S-FLEX & KAMA FLOW SERIES

Ambient noise at the time of testing was 19 dBA.

Kama Flow SA1225FDB12L (1,000 RPM)
Brand Scythe Power Rating 0.13A
Manufacturer ? (Sony bearing) Airflow Rating 38.9 CFM
Model Number Kama Flow SA1225FDB12L RPM Rating 1,000±10% RPM
Retail Availability Japan Only Noise Rating 15.8 dBA
Bearing Type FDB Header Type 3-pin
Hub Size 1.89" Starting Voltage 6.6V
Frame Size 120 x 120 x 25 mm Number of Samples 3
Voltage
Noise
RPM
CFM
Power
12V
1040 RPM
43 CFM
0.61W
9V
800 RPM
32 CFM
0.38W
7V
610 RPM
23 CFM
0.27W
5V
410 RPM
14 CFM
0.19W
@25 CFM (7.4V)
650 RPM
25 CFM
0.30W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1040 RPM
29 CFM
0.61W
9V
800 RPM
21 CFM
0.38W
7V
610 RPM
16 CFM
0.27W
5V
410 RPM
11 CFM
0.19W
@20 CFM (8.4V)
740 RPM
20 CFM
0.35W

We discovered the Scythe
Kama Flow in our roundup of 80mm fans
, where it was the only fan to equal
the Nexus 80 for quietness and noise quality. Its only downside? A lack of availability
outside of Japan.

Scythe also sells a 120mm version of the Kama Flow in Japan, and it promises
to be just as good as the 80mm version. However, international users stand a
better chance of getting their hands on the bigger version — or at least
one that is very, very similar. Scythe's S-Flex fans look — and sound
— almost identical to the Kama Flow, and these are available internationally.
Both fans use Sony's S-FDB bearings, both have the same blade design, and both
are characterized by three tiny pinholes in the center of the hub. In fact,
the only discernible differences that we could find were the design of the frame
(S-Flex is double-ribbed and seems slightly heavier), and the speed ratings.

It seems likely that there are other "unofficial" models in this
line as well — most of Scythe's recent heatsinks have featured quiet, low
speed fans that sound very similar to the S-Flex / Kama Flow, including the
odd 100mm fan found in the Scythe
Infinity
. The common thread appears to be the S-FDB bearing, which Scythe
has licensed from Sony.

Slow Version: S-Flex SFF21D (800 RPM)
Brand Scythe Power Rating 0.10A
Manufacturer ? (Sony bearing) Airflow Rating 33.5 CFM
Model Number S-Flex SFF21D RPM Rating 800 RPM
Retail Availability Yes Noise Rating 8.7 dBA
Bearing Type FDB Header Type 3-pin
Hub Size 1.89" Starting Voltage 8.9V
Frame Size 120 x 120 x 25 mm Number of Samples 1
Voltage
Noise
RPM
CFM
Power
12V
850 RPM
32 CFM
0.34W
9V
620 RPM
23 CFM
0.27W
7V
430 RPM
14 CFM
0.24W
5V
Stall
Stall
Stall
Stall
@25 CFM (9.5V)
650 RPM
25 CFM
0.28W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
850 RPM
23 CFM
0.34W
9V
620 RPM
16 CFM
0.27W
7V
430 RPM
11 CFM
0.24W
5V
Stall
Stall
Stall
Stall
@20 CFM (10.8V)
750 RPM
20 CFM
0.31W

The S-Flex is available in three different speed ratings: 800 RPM (SFF21D),
1,200 RPM (SFF21E), and 1,600 RPM (SFF21F). Most users will probably be interested
in the -D or -E variants which are around the 1,000 RPM mark that is generally
the fastest a 120mm fan can be run quietly. The Kama Flow fits right in between
these models at 1,000 RPM.



The S-Flex package includes a Molex to 3-pin adapter with an RPM monitor
cable.

The Kama Flow's noise character was almost identical to its smaller brother.
It was quiet, very smooth and had just a touch of whine at full speed. It also
sounded almost identical to the Nexus and measured identically as well. The
only real difference was that the Kama Flow sounded slightly more tonal, which
unfortunately made it a bit easier to hear. The noise had other similarities
as well; both fans produced predominantly low frequency noise, and both vibrated
enough that they may be prone to resonance if hard-mounted.

And, as with the 80mm versions, it was too close to call a clear winner between
the Kama Flow and the Nexus. Both were very, very quiet, so the deciding factor
has to be something other than noise. The superior quality of the S-FDB bearing
gives the Kama Flow an edge here and the wide distribution of the S-Flex takes
away the advantage that the Nexus had in the 80mm form factor.

As mentioned, the S-Flex fans sounded more or less identical to the Kama Flow
at the same speeds. For practical purposes, they can probably be treated as
one and the same, but we did notice two minor variations.

  1. The S-Flex models vibrated less than the Kama Flow, perhaps because of their
    more rigid frame.
  2. The slowest S-Flex model had a slight ticking that was absent in all of
    the other samples we tested. It's not clear whether this can be attributed
    to the lower commutator frequency or if it's just sample variance making an
    appearance.

The Kama Flow is probably the most usable fan of the bunch. It's 1,000 RPM
speed rating offers a lot of flexibility. At full speed, it is probably fast
enough for use in very hot systems while still being quiet enough for users
who don't seek absolute silence. It also undervolts well, and is inaudible at
7V. Its only disadvantage is its starting voltage, and it should not be used
at 5V.

The Low speed S-Flex is probably the best choice for those who just want a
drop in solution. It is very close to inaudible at full speed, and should disappear
entirely if there are any other significant sources of noise in the system.
The fact that it is so quiet at 12V makes it attractive to users who don't want
a fan that just works without needing to tinker with it. However, modders will
probably want to avoid it, as its extremely high starting voltage makes it useless
for undervolting.

The Medium speed S-Flex is the most suitable choice for those who don't have
access to the Kama Flow. It's only slightly faster, and it starts quietly and
reliably at 5V — an important feature for those who don't have variable
speed fan controllers. The high speed S-Flex is also an option. It's definitely
too noisy at full speed, but it is very usable when undervolted and can push
more air than the medium speed version if necessary.

Medium Version: S-Flex SFF21E (1,200 RPM)
Brand Scythe Power Rating 0.15A
Manufacturer ? (Sony bearing) Airflow Rating 49.0 CFM
Model Number S-Flex SFF21E RPM Rating 1,200 RPM
Retail Availability Yes Noise Rating 20.1 dBA
Bearing Type FDB Header Type 3-pin
Hub Size 1.89" Starting Voltage 4.8V
Frame Size 120 x 120 x 25 mm Number of Samples 1
Voltage
Noise
RPM
CFM
Power
12V
1230 RPM
51 CFM
0.62W
9V
910 RPM
36 CFM
0.47W
7V
670 RPM
26 CFM
0.37W
5V
400 RPM
13 CFM
0.29W
@25 CFM (6.9V)
650 RPM
25 CFM
0.36W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1230 RPM
34 CFM
0.62W
9V
910 RPM
23 CFM
0.47W
7V
670 RPM
18 CFM
0.37W
5V
400 RPM
10 CFM
0.29W
@20 CFM (7.8V)
750 RPM
20 CFM
0.41W
Fast Version: S-Flex SFF21F (1,600 RPM)
Brand Scythe Power Rating 0.20A
Manufacturer ? (Sony bearing) Airflow Rating 63.7 CFM
Model Number S-Flex SFF21F RPM Rating 1,600 RPM
Retail Availability Yes Noise Rating 28 dBA
Bearing Type FDB Header Type 3-pin
Hub Size 1.89" Starting Voltage 4.0V
Frame Size 120 x 120 x 25 mm Number of Samples 1
Voltage
Noise
RPM
CFM
Power
12V
1590 RPM
70 CFM
1.05W
9V
1190 RPM
53 CFM
0.75W
7V
890 RPM
39 CFM
0.57W
5V
550 RPM
23 CFM
0.41W
@25 CFM (5.2V)
590 RPM
25 CFM
0.43W
May 5, 2008

The updated airflow results here are the result of improvements
in our testing procedures. They are more accurate than the original results
above, but they are not directly comparable. Please compare these only with
fan reviews published after May 5, 2008 — or ones that have updated
results published in a box like this one.
12V
1590 RPM
47 CFM
1.05W
9V
1190 RPM
32 CFM
0.75W
7V
890 RPM
23 CFM
0.57W
5V
550 RPM
14 CFM
0.41W
@20 CFM (6.3V)
770 RPM
20 CFM
0.51W

Noise Recordings

JUMP TO:

CONCLUSIONS

There appears to be quite a bit of competition for the title of
quietest 120mm fan. Of the six fan series that we tested, three are good enough
for an SPCR recommendation, and the only real stinkers were the ARX fans.

In fact, there was enough competition that our longstanding favorite
— the Nexus — was only the third best fan that we encountered, although
the three top fans were so close that a 1-2-3 ranking isn't really fair.

The best fan of the bunch was the Noctua, which overcame our skepticism
and lived up to its hype by producing more airflow at lower speeds. Its airflow-to-noise
ratio is its best feature, which gives it enough of an edge to make up for its
slightly rougher noise character. The Noctua has seen recommendations around
the web, and we have no qualms about adding our own to the list.

Ranked in terms of noise, the Nexus and the Kama Flow / S-Flex
should probably tie for second, but the better quality S-FDB bearings in Scythe's
fans and the specter of sample variance for the Nexus are enough to secure second
place for Scythe fans. We'll call the Nexus a close third.

Although the ebmPapst and Antec fans are not of the same noise
caliber as the top three picks, they are close enough that some users will probably
find other reasons to buy them. The Antec TriCool is a very good choice for
casual users, as it is very widely distributed and its "Low" setting
is good enough for systems that don't need to be inaudible. EbmPapst's 4412
series sounds good enough to redeem the company after the noisy 80mm fans that
we looked at, and their reputation for quality and reliability is unparalleled.
If security and reliability are requirements on par with noise and cost is of
no object, the 4412 is your fan. Don't bother shelling out for the low speed
4412 F/2GLL though... the faster FGL model is just as quiet when undervolted,
and sounds smoother to boot.

With so many good quality fans to choose from, we are sure that
all will develop followings of their own. Although we have ranked them here,
the rankings are based mostly on judgment calls and subjective impressions,
and we have no doubt that some people will disagree with our ranking. Such was
the nature of this comparison; almost all of the fans were "good enough",
and it was only fine differences that allowed choices between the fans to be
made.

Many thanks to all the readers, contributors and manufacturers
who donated fans so this project could happen.

*

SPCR Articles of Related Interest:

SPCR's 80mm Fan Roundup #1

SPCR's Fan Testing Methodology

Anatomy of the Silent Fan

SPCR's Recommended Fans

Simple Fan Controllers from Zalman

Get 5V, 7V, or 12V for your Fans

* * *

Discuss
this article in the SPCR Forums
.

* * *

JUMP TO:

Sections: 

Google

www SPCR