The Scythe Ninja is an outstanding quiet CPU heatsink. Its simple yet elegant design
takes into account all the factors relevant for quiet cooling: Large surface area, wide fin spacing, multiple heatpipes for even heat distribution through all the fins, and flat, smooth copper base. Of course, not all users can utilize the 150mm tall Ninja or similar tower heatsinks due to the dimensional restrictions of smaller cases. Ergo, the new Ninja Mini.
September 6, 2007 by Lawrence
Lee[with Mike
Chin]
Product | Scythe Ninja Mini |
Model SCMNJ-1000 CPU Heatsink
The Scythe Ninja is one of the most
outstanding CPU heatsinks we’ve
had the pleasure of working with. Its simple yet elegant design
takes into account all the factors that optimize quiet cooling
— large surface area, wide fin spacing, multiple heatpipes for even heat distribution through all the fins, and flat, smooth copper base. Of
course, not all users can utilize the 150mm tall Ninja or similar tower heatsinks
due to the dimensional restrictions of smaller cases.
These were the thoughts going through SPCR editor Mike Chin’s mind as he was working with Antec on the design of the NSK2400/Fusion media PC case. After seeing the first prototype of the case in early 2006, Mike contacted Scythe, asking them to release a short Ninja-style heatsink that would fit within the confines of NSK2400/Fusion. It would have to be no more than ~120mm tall. Mike felt that a high performance heatsink that could be run passively, using the closely-positioned 120mm case fans on the NSK2400/Fusion would sell very well. There were also many other medica PC cases which needed a shorter high performance HS with fins designed for the horizontal (not vertical) airflow best suited for optimized cooling in such cases.
Scythe’s response was lukewarm. A similar product concept had been briefly discussed by key staff, but rejected. They mentioned that the cost savings involved in making a shorter Ninja would be minor, and the smaller size would mean not only smaller fin area but also prevent the use of a 120mm fan. These factors would impact the cooling performance while price would remain pretty close to the original Ninja. Scythe was also unconvinced of the media PC’s popularity among DIY builders, who are naturally the primary market for their products. Even a face-to-face meeting between Mike and Scythe’s product development manager at Computex Taipei in June 2006 failed to get this project started.
Other smaller tower HS with horizonal airflow design have appeared on the market, but none have been ideal for a quiet media PC case. They’ve been a bit too tall, a bit too loud, and/or not quite good enough in cooling performance. So this product niche in CPU coolers remains open.
At the 2007 Computex, some 18 months after Mike first mentioned the idea to Scythe, the company released the Ninja Mini. This product conforms perfectly to Mike’s original requirements: It’s short enough for a 3U height case, uses horizonal airflow configution, and it’s basically a short Ninja. Better late than never. How good a quiet cooler is it? Can it work well without a fan on it in an Antec NSK2400/Fusion case? These are the main questions we’ll seek to answer in this review.
As with other Scythe products, the box tries |
The cooler has universal support for |
Scythe Ninja Mini Feature Highlights (from the product web page) | |
Feature & Brief | Our Comment |
Ready for 80mm or 92mm Fan Mounting The default silent fan (80 x 80 x 25 mm) can be easily exchanged to | Due to its lower profile, a 120mm fan is inappropriate. A 80 or 92mm fan will have to suffice. Scythe’s steel wire fan clips are pretty easy to use. |
Improved Mounting Systems Never stop the improvement of mounting systems, the Ninja Mini CPU | The Ninja Mini’s AMD mounting frame is slimmer and less cumbersome than the original Ninja’s. The LGA775 mounting scheme has also been streamlined from a round frame to a simpler square frame. |
Universal Socket Compatibility with AM2 Ninja Mini CPU Cooler is compatible for socket 478/754/939/940/AM2 | Three sets of easy to install mounting frames provide |
Scythe Ninja Mini Specifications (from the product web page) | |
Model Number | SCMNJ-1000 |
Combined Dimensions | 110 x 110 x 115 mm |
Weight | 580g |
Compatibility | Intel LGA775 (SocketT) Processors Intel Socket 478 Processors AMD Socket AM2 Processors AMD Socket 940 Processors AMD Socket 939 Processors AMD Socket 754 Processors |
Fan Dimensions | 80 x 80 x 25 mm |
Noise Level | 24.4 dBA |
Air Flow | 32.2 CFM |
Speed | 2,300 RPM (±10%) |
Bearing Type | Sleeve |
PHYSICAL DETAILS
Most readers probably want to know what the difference is between the Ninja Mini and the original. The photos below should satisfy your curiosity. (Note: Please ignore the worn condition of the Ninja; it’s our original review sample which has not only survived over two years of abuse in the lab, but also an attempt at height reduction which caused the loss of the end caps of the heatpipes.)
No point for guessing which is which. The height difference is 40mm, or slightly over 1.5", as specified by Scythe. |
The family resemblance obvious, but there is a difference in the way the heatpipes are configured. In the Mini, they’re evenly distributed through each fin. This may help improve heat distribution as well as reduce airflow resistance between the fins. |
Notice the "auxilary fins" just above the base of the
heatsink in the various photos.
The heatsink is not symmetrical so it is possible to place the
fan in the “wrong” position with the fan blowing against the side of these fins
instead of through them (as pictured
below). This may have some minor impact on CPU cooling, and perhaps in the cooling of other nearby components on the motherboard.
When compared to a tower heatsink, it’s obvious why it’s called “Mini.” However the name and picture is somewhat deceptive – it’s the same width as the original Ninja. The Ninja Mini is primarily composed of 17 large, flat, horizontal aluminum fins and 6 copper heatpipes running through them. In imperial measurements, it stands approximately 4.5″ tall and weighs 20 oz. |
We also noticed that heatpipes at the
base of the Mini diverge and separate into a well-spaced
clock-like pattern, unlike on the Ninja where the heatpipes
are
packed tightly together. The Mini design allows more air to
pass
between these pipes, especially at the base of the socket where
motherboard components can suffer thermally without top-down airflow.
BASE & MOUNTING CLIPS
Typical of Scythe’s heatsinks, the base is flat, smooth, and polished. This time around, they’ve applied a nickel plating on the copper base, which will keep the copper from tarnishing over time. It’s not clear whether nickel plating has any functional purpose, however; once the HS is installed with thermal goop, it’s not exposed to the oxidation effects of the air.
Screw mounting frames. This design allows the frames to |
The S478 mounting |
The AMD mounting |
The LGA775 |
FAN DETAILS
We hoped Scythe would ship the Ninja Mini with the Scythe Kama Flow 80mm fan (which uses fluid dynamic bearings), but instead they including a
sleeve bearing fan, unknown to us, dubbed the DFS802512L. Being a 3-pin fan, some motherboards featuring 4-pin PWM headers may not be able to control it.
It’s a slow speed sleeve bearing 80x25mm fan.. |
Stock Fan Specfications: DFS802512L | |||
Brand | Scythe | Power Rating | 0.09A |
Manufacturer | ?? | Airflow Rating | 32.2 CFM |
Model Number | DFS802512L | RPM Rating | 2,300RPM |
Bearing Type | Sleeve | Noise Rating | 24.4 dBA |
Frame Size | 80 x 80 x 25 mm | Header Type | 3-pin |
INSTALLATION
Installation of the Scythe Ninja Mini on our LGA775 test
system
was
straightforward. With the four mounting holes being underneath the
edges of the heatsink, we thought there might be problems putting
enough pressure on the fasteners, but luckily this was not the case and
they went in fairly easily. We highly recommend checking the
back of the motherboard to ensure the inner black pins go all
the
way
through; installation inside a case is definitely not a good idea. In any case, once the cylinder fasteners are firmly engaged, the Ninja Mini is quite secure. (Note: The tall column of empty space at
each
corner of the heatsink provides easy
access for a slot screwdriver to release the fasteners for removal.)
Ninja Mini installed on our 775 socket HS test platform.
A clearance snag.
We ran into a small snag as one of the fan clips
touched the memory
in the first DIMM slot. There
was about a millimeter of clearance when we used the second slot
instead. This may or may not be an issue depending on the
layout of the
motherboard used.
Height clearance over our NB heatsink was very good;
without the fan, RAM slots are no problem either.
TESTING
Testing was done according to our
unique heatsink testing methodology, and the reference fan
was profiled using our standard fan testing methodology.
A quick summary of the components, tools, and procedures follows below.
Key Components in Heatsink Test Platform:
- Intel Pentium D 950
Presler core. Under our test load, it measures 78W including efficiency
losses in the VRMs. - ASUS P5LD2-VM
motherboard. A basic microATX board with integrated graphics and plenty
of room around the CPU socket. - Hitachi
Deskstar 7K80 80GB SATA hard drive. - 1 GB stick of Corsair XMS2
DDR2 memory. - FSP
Zen 300W fanless power supply. - Arctic Silver
Lumière: Special fast-curing thermal interface
material, designed specifically for test labs.
Test Tools
- Seasonic Power Angel
for measuring AC power at the wall to ensure that the heat output
remains consistent. - Custom-built, four-channel variable DC power supply,
used to regulate the fan speed during the test. - Bruel & Kjaer (B&K) model 2203
Sound Level Meter. Used to accurately measure noise down to
20 dBA and below. - Various other tools for testing fans, as documented in our
standard fan testing methodology.
Software Tools
- SpeedFan 4.31,
used to monitor the on-chip thermal sensor. This sensor is not
calibrated, so results are not universally applicable; however, - CPUBurn P6,
used to stress the CPU heavily, generating more heat than most
real applications. Two instances are used to ensure that both cores are
stressed. - Throttlewatch 2.01,
used to monitor the throttling feature of the CPU to determine when
overheating occurs.
Noise measurements were made with the fan powered from the lab
variable DC power supply while the rest of the system was off to ensure
that system noise did not skew the measurements.
Load testing was accomplished using CPUBurn to stress the
processor, and the graph function in SpeedFan was used to make sure
that the load temperature was stable for at least ten minutes. The
stock fan was tested at four voltages: 5V, 7V, 9V, and 12V,
representing a full cross-section of the its airflow and noise
performance. It was also tested with our reference 80mm fan, the Nexus 80.
The ambient conditions during testing were 18 dBA and
24°C.
TEST RESULTS
The Stock Scythe Fan
The stock fan was tested to reveal its noise
characteristics. However, we did not measure
airflow as we
are in the process of further improving this aspect of our fan
testing methodology at the moment. We hope to do a more thorough
analysis at a later date.
Stock Scythe Ninja Mini Fan (DFS802512L) Measurements | ||
Voltage | RPM | SPL |
12V | 2200 | 28 dBA@1m |
9V | 1800 | 22 dBA@1m |
7V | 1460 | 19 dBA@1m |
5V | 1080 | <19 dBA@1m |
2.5V (min start) | 430 | <19 dBA@1m |
At 12V, it simply spins a bit too fast to be quiet by SPCR standards, with turbulent noise, a
small a degree of whine and a fair amount of vibration. At 9V the sound
level decreased dramatically but whine and vibration were still evident
with some tonality. At 7V it’s probably quiet enough for the majority
of users though its overall character is unchanged from 9V. At 5V it’s
quiet and smooth with some vibration, and basically inaudible from greater
than 1.5 feet away. We were surprised to find that the
fan started up with only 2.5V though we can’t envision a situation
where such a low voltage would be used. From 5V to 2.5V, the acoustic
difference is negligible from all but the shortest of
distances. Overall it’s a decently smooth fan, a typical
sleeve bearing variety.
Cooling Results
Scythe Ninja Mini w/ stock 80mm fan | ||||
---|---|---|---|---|
Fan Voltage | Noise @ 1m | Temp | °C Rise | °C/W |
12V | 28 dBA | | 19 | 0.24 |
9V | 22 dBA | | 21 | 0.27 |
7V | 19 dBA | | 24 | 0.31 |
5V | <19 dBA | | 28 | 0.36 |
Scythe Ninja Mini w/ reference 80mm fan | ||||
12V | 20 dBA | 46°C | 22 | 0.28 |
9V | ~19 dBA | 50°C | 26 | 0.33 |
7V | <18 dBA | 55°C | 31 | 0.40 |
Load Temp: CPUBurn for ~20 mins. °C Rise: Temperature rise above °C/W: based on the amount of |
The stock fan was a bit more audible mounted on the Ninja than in free air.
As expected, at 12V, it was unacceptable by our noise standards. It also
made the heatsink “ring” — that is, its fins began to resonate in sympathy
with the fan. Undervolting the fan to 9V brought the noise level down, and the CPU temperature barely rose. Performance was good at all voltage levels with the stock fan.
At 5~7V, the overall balance of cooling and noise seemed about the best: 28°C rise at <19 dBA@1m is pretty darn good when you consider that the CPU is pulling some 70W.
As expected, the Ninja Mini did not do as well with the much quieter reference fan. At
7V °C Rise broke 30°C
and we decided not to proceed any lower. Inside a case where
temperatures are usually significantly higher than out in the open,
it’s unlikely the reference fan could have kept our test CPU
sufficiently cool at 5V. Overall performance was excellent for its size.
With our reference Nexus fan in
this
orientation at 12V, the heatsink’s performance was one
degree worse, so keep
this in mind when placing the fan.
Comparables
Comparisons are limited to the heatsinks tested on our current test platform, which is relatively new. Note that both of the other HS are much larger, and have the benefit of a 120mm Nexus fan which naturally can blow a lot more air for the same level of noise compared to any 80mm fan.
SPL@1m | Ninja Mini (stock fan) | Scythe Ninja (reference fan) | Thermalright SI-128 (reference fan) | ||||||
RPM | °C Rise | °C/W | RPM | °C Rise | °C/W | RPM | °C Rise | °C/W | |
22 dBA | 1800 | 21 | 0.27 | 1080 | 14 | 0.18 | 1080 | 21 | 0.27 |
~19 dBA | 1460 | 24 | 0.31 | 850 | 16 | 0.21 | 850 | 26 | 0.33 |
<19 dBA | 1080 | 28 | 0.36 | 680 | 17 | 0.22 | 680 | 29 | 0.37 |
The Ninja Mini, despite being hindered by a much smaller fan, is not
left in the dust as one would expect when compared to bigger heatsinks
utilizing 120mm fans. At similar
noise levels, the Mini is outclassed by the original Ninja, but keeps pace with the Thermalright SI-128 (admittedly designed for best performance with higher airflow).
NOISE RECORDINGS IN MP3 FORMAT
Each of these recording starts with five seconds of "silence" to let you hear the ambient
sound of the room, followed by 10 seconds of the fan’s noise at 5V, 7V, 9V and 12V. The five seconds of "silence" is inserted between each 10 second stretch of fan noise to help you remember the reference ambient.
- Scythe Ninja Mini Stock fan (DFS802512L) @1m: One
meter, One
foot (30cm) - Nexus 80mm "Real Silent Case Fan" (SP802512L-03) – SPCR reference @1m: One
meter, One
foot (30cm) - Nexus 120mm "Real Silent Case Fan" – SPCR reference @1m: One meter, One
foot (30cm)
HOW TO LISTEN & COMPARE These recordings were made The one meter recording The one foot recording is More details about how we make these recordings can be found in our short article: Audio Recording Methods Revised. |
IN-CASE TEST RESULTS
We were interested to see how the Ninja Mini would fare in a quiet media PC case; the natural choice was an Antec Fusion. The heatsink testbed was installed into the case without any other peripherals (which were not needed for our thermal/noise testing). Our sample Fusion is a new black version with a 430W version of the SU380 that shipped with the original NSK2400. Antec is transitioning most of the cases with supplied PSUs to the EarthWatts power supplies, so our Fusion sample must be among the last of the older stock. The SU430 has the same acoustics as the SU380, especially at the modest power levels of our test system.
The
only fans used during testing were the one inside the power supply (which has no effect on cooling any other components) and
one of the Antec Tricool 120mm case fans on the low
setting. We chose to jump straight into the deep end of the pool (or more apropos, onto the hot part of the coals): No fan was mounted on the Ninja Mini. A 120mm fan with close proximity
to the heatsink may be as good or even better than a 80mm fan mounted
directly on it. You can see in the photos that the exhaust case fan is so close to the Ninja Mini that it might as well be mounted directly on it.
The extra 120mm fan vent was blocked off before testing.
addition of a video card, so we tested the system
with
both the onboard video and a fanless Geforce 7600GT (GV-NX76T256D-RH)
courtesy of Gigabyte. Note the use of the black plastic baffle just behind the heatsink to direct the intake airflow from the back panel and the top panel of the Fusion over the motherboard VRM and the CPU heatsink before being evacuated by the case fan.
Fanless Geforce 7600GT graphics card…
…with cooling fins joined by heatpipes on both sides of the card.
The cooling results were consistently good with just the onboard video or the passive 7600GT.
Test | Video Card | Temperatures | System Power (AC) | ||||
GPU | CPU | HDD | AUX | SYS | |||
CPUBurn | Onboard (GMA950) | N/A | 55°C | 36°C | 46°C | 38°C | 137W |
3DMark2006 | 7600GT | 68°C | 54°C | 36°C | 46°C | 40°C | 140~160W |
CPUBurn + RTHDRIBL | 7600GT | 77°C | 55~57°C | 36°C | 46°C | 42°C | 172W |
Temperatures were taken with SpeedFan and nVidia’s temperature display after running the testing application(s) for 30 minutes. Ambient temperature was 24°C. |
The achieved 54~57°C is very good maximum temperature for an Intel Pentium D950 inside a case, and the
other
temperatures are perfectly benign. (Note: We are not sure
what Speedfan reports as “AUX” and “SYS” temperatures refer to
exactly; suffice it to say they are onboard sensors that Asus considers worth keeping track of). Remember, real media
PC system would never to be pressed this hard for half an hour, nor would
it be powered by a CPU that pulls ~70W on its own. A much cooler mid- to low-end Core 2 Duo or 45W AMD A64X2 are more appropriate CPU choices.
The passively
cooled 7600GT also did quite well inside the case, staying under
80°C during testing. Modern GPUs are known to operate nominally
in
excess of 100°C and nVidia’s throttle point is a
sizzling 115°C. Somewhat surprisingly, adding the 7600GT
only
increased CPU temperature by a marginal amount.
The Antec SU430 power supply was actually the loudest component
in the system, measuring 24~25 dBA @ 1m. The Antec Tricool fan set to low only
produces approximately 19 dBA@1m in free air. The
system
measured 25 dBA@1m directly in front of the case. From the right
side, behind, and directly over it, where the
fans and vents are closer, the noise levels were 1-2 dBA
higher. With a quieter power
supply, a 21-22 dBA system is not out of the question. (Note that the requirements of a media PC are somewhat less demanding than for a home desktop PC, because the sound from TV and movie programming will usually mask any <30 dBA@1m broadband noise from the PC. For more discussion of this issue, see the section ACOUSTICS AROUND A MEDIA PC in
Cases: Basics & Recommendations.
FINAL CONCLUSIONS
The Scythe Ninja Mini performed admirably despite its smaller size and smaller 80mm
fan. Its size allows it to be
mounted in smaller cases where cooling is often a bigger issue than
in a large tower case. It’s dead easy to install on AMD and socket 478 boards, reasonable to install on Intel 775 socket boards, and its modular mounting
system ensures compatibility with almost any current socket motherboard. With its modest price tag, it’s a good value compared to some of the
larger, more dated heatsinks on the market. We could complain that it took too long for Scythe to come up with the Mini, but that’s a throwaway remark. They did more than just reduce the height of the Ninja; we have little doubt that the more even distribution of heatpipes through the fins helps the Mini achieve its high performance.
One
minor disappointment was Scythe’s choice of fan. At full speed the
level of noise was a little too high. We’re used to seeing
much quieter fans included with Scythe heatsinks, so hopefully
this does not become a trend. A 92mm, fluid dynamic
bearing, 4-pin PWM fan compatible with most current motherboard fan controllers would have made us swoon.
The Mini is a very good cooler, even without any consideration of size or fan type used. But there are several situations where a Scythe
Ninja Mini is an excellent choice:
- You’re
uncomfortable with the prospect of having a heavy/tall tower
heatsink inside your system - Your case cannot fit the huge high performance CPU heatsinks
- You
just want a solid budget CPU cooler - You want a perfect passive CPU cooler in an Antec Fusion/NSK2400 or similar media PC case
Pros * Excellent overall cooling * Performs well in a low airflow environment * Can be used in smaller cases * Easy to install * 360° mounting frame design | Cons * Non-PWM stock fan could be quieter * Mounting system for socket 775 could be more secure |
Much thanks to Scythe
USA for the Scythe Ninja Mini sample.
Our thanks also to….
Gigabyte for the excellent Geforce 7600GT GV-NX76T256D-RH graphics card, and to…
Antec for the Black Fusion media PC case.
* * *
Articles of Related Interest
Recommended Heatsinks
SPCR’s unique heatsink testing
methodology
SPCR’s standard fan testing
methodology
Scythe Ninja
Thermalright SI-128
Spire Verticool II