Zalman CNPS9900DF Dual Fan Flower Heatsink

The latest update to the iconic Zalman CNPS9900 series adds more girth and a second fan… in an ultimately futile attempt to get back to the top ranks of CPU coolers.

October 8, 2012 by Lawrence Lee

Product	Zalman CNPS9900DF CPU Cooler
Manufacturer	Zalman
MSRP	US$90

When the Zalman
CNPS9900 LED CPU cooler was released in 2009, we thought it might be
the end of the flower series. We were almost certain of it when they launched
conventional tower coolers in 2010. To our surprise, they revived the 9900 in
2011 with the MAX,
which added nickel-plating and a larger fan. And now, in 2012, they’ve updated
it once again with the CNPS9900DF. It seems there is some life left in iconic
Zalman flower HS line after all. The “DF” stands for dual fans, following
in the footsteps of twin tower coolers like Noctua
NH-D14 and Thermalright
Silver Arrow. Like these premier heatsinks, the DF carries a very high
MSRP, US$90. Zalman’s all-copper heatsinks have always been costly but
this takes it to another level.

The box.

As you can probably guess, it’s larger than its predecessor, sporting an extra
100 grams of bulk and a 26% higher heat dissipation area than the MAX. Instead
of one 140 mm high speed fan, the DF has both a 140 mm and 120 mm model with
more moderate speeds so it shouldn’t be much louder than the MAX, if at all.
We weren’t impressed by the MAX’s performance though so we’re not convinced
this latest iteration is enough to compete with coolers in the same price range.

Package contents.

Like previous versions of the CNPS9000 series, the heatsink is packed in a
sturdy plastic clamshell container. The accessories include an instruction sheet,
thermal compound, and mounting hardware. It’s essentially the same package as
the MAX except for the inclusion of a dual fan adapter and hardware for socket
LGA 2011. We frown to see the same installation scheme as the MAX being recycled.
It didn’t produce enough pressure on the base and used very small hexagonal
screws that were hard to secure.

Zalman CNPS9900DF: Key Features (from the product web page)
Feature & Brief	Our Comment
Cross Bending Fin Technology Front and rear airflow incident angles are designed in different directions. This cross bending fin technology produces cooling performance with minimum noise.	The radial shape of the 9900 pushes exhaust air out in every direction. One could argue that a standard tower design directing air toward the case exhaust fan is more effective.
Triple Composite Heatpipes 3 composite heat pipes and pure copper cross bending fins quickly cools down the heat from the CPU.	Not really different than the heatpipes used by other manufacturers.
Dual Fans Cross Bending ultra quiet powerful 120mm Front Fan and 140mm enter Fan produces top performance.	We’re not crazy about having mismatched fans &151; they sometimes produce nasty acoustic results.
Blue LED FANS Beautiful blue LED dual fan.	A must have feature.
Mirroring Base Zalman’s unique sanding process provides excellent heat transfer ability.	It’s not necessary for a base to have a mirror shine to produce good results.
Fits in Mid Case With height 160mm and width 180mm it fits in a variety of cases including mid case.	We believe they mean that 160 x 180 mm is the limit for most midsize cases as the 9900DF is much smaller than this.
High compatibility Compatible with Intel Socket LGA 2011, AMD Socket FM1 and most Intel, AMD platform.	Expected for most performance heatsinks.
ZM-STG2 High performance ZM-STG2 helps to maximize the heat transfer from CPU to CNPS9900DF base allowing optimal cooling performance.	We’ll have to take their word for it. TIM is one of the controls we employ to more accurately compare heatsink performance.

 

Zalman CNPS9900DF: Specifications (from the product web page)
Dimensions	140(L) x 100(W) x 154(H)mm
Weight	850g
Materials	Copper(base & Fin)
Dissipation Area	6,800cm2
Heatpipe Q’ty	3EA
Fan Speed	Front 120mm: 1,000rpm ±10% (3pin) Center 140mm: 900 ~ 1,400rpm ±10%(4pin)
Noise Level	19 ~ 27dBA ±10%
Bearing Type	Long Life Bearing
Control Method	PWM Control, Auto Restart
Input Voltage	12V

PHYSICAL DETAILS

The Zalman CNPS9900DF sports an all copper construction consisting
of a base 3 x 6 mm thick heatpipes and a multitude of very thin fins, all nickel-plated.
By our measurements, the heatsink is 152 mm tall (6.0 inches) and approximately
870 grams (1.9 lb). By modern standards its size is rather conservative.

The core design of the 9900 series remains, though it’s split into two sections each with its own fan. The fins are arranged radially, tightly packed at the center and looser at the edges while the fans disperse the air in every direction.

The DF isn’t much different from the CNPS9900 MAX, The intake portion of the heatsink has been enlarged to encapsulate a second fan but otherwise it looks mostly the same. Our biggest complaint, the fan being mounted on a pair of metal stilts prone to vibration, remains. There’s too much separation behind the center fan and without a casing, it doesn’t generate much static pressure.

The thin fan support is freestanding, not anchored to any portion of the heatsink, making it prone to vibration.

The heatsink extends around the edges of the front 120 mm fan while the sides of the center 140 mm fan are unimpeded.

Though it looks like a six heatpipe cooler, it actually only has three. Each heatpipe is very long, wrapping all the way around the heatsink. In the image above you can see both the thick and thin ends of two of the heatpipes.

The base is a very nice mirror polish but the shape is actually more important for thermal transfer. The 9900DF’s base is slightly concave; heatsinks with flat and convex surfaces perform better on the vast majoriity of heatsinks, which are usually a tiny bit concave. Two concave surfaces mating againsts each other almost guarantees a gap in the middle. The gap may be small, in the order of fractions of a millimeter, but the difference between a gap of any kind versus firm contact is easily 10°C. The screws on the bottom can be loosened so the metal mounting clips can be inserted between the base and baseplate, and then re-tightened.

BASE & INSTALLATION

The most critical aspect of installation is that the heatsink be securely
mounted. A firm mating results in good contact between the heatsink’s base and
the CPU heatspreader and more efficient heat conduction. Ideally it should
also be a simple procedure with the user having to handle as few pieces of
hardware as possible.

The 9900DF employs the same mounting system as its predecessor, the MAX. For Intel systems, a lug nut is inserted into each corner of the backplate through the appropriate hole depending on the socket. A plastic brace goes over it to lock in place for mounting.

Bolts are then used to secure the lug nuts to the clips but because the heatsink hangs over them, it’s hard to get to. To make it easier, Zalman uses bolts with shallow hexagonal threads and includes an Allen wrench in the package. It’s a real pain as the wrench is impeded by board components like the VRM heatsink, while the long end can only engage the screw head at an angle. Also note how the mounting clips bend downward, putting most of the pressure around the perimeter rather than the center of the CPU.

Typically we tightening the bolts down as much as possible to give every heatsink the best possible chance but we ended up wearing down the hex head threads easily. We then had to find a different Allen wrench to remove them which proved to be a major task as the thread size is very small. Oddly, we didn’t run into this problem with the CNPS9900 MAX despite it having the exact same system. After re-examining the whole mounting system and process, both Mike Chin and I concluded that stripping the head threads of these tiny bolts is more normal than not; we were just lucky that it didn’t happen with the CNPS9900 MAX… unless those bolts had better threaded

One nice thing about the 9900DF is it doesn’t take up that much room. There is about 40 mm of clearance below the bottom fin but the body is fairly narrow so it didn’t interfering with anything.

An adapter is included to run both fans off a single motherboard fan header but the larger center fan is a 4-pin PWM model while the front fan is an old school 3-pin. Using PWM control, the 3-pin fan runs at full speed.

TESTING

Before thermal testing, we took some basic physical measurements.

Approximate Physical Measurements
Weight	870 g
Height	152 mm
Fin count	A lot
Fin thickness	0.26 mm
Fin spacing	Varies
Vertical Clearance*	40 mm
* measured from the motherboard PCB to the bottom fin of the heatsink

 

Large Heatsink Comparison: Average Fin Thickness & Spacing
Heatsink	Fin Thickness	Fin Spacing
SilverStone Heligon HE02	0.52 mm	3.30 mm
Thermalright HR-01 Plus	0.45 mm	3.15 mm
Thermalright HR-02 Macho	0.34 mm	3.12 mm
Scythe Ninja 3	0.39 mm	2.64 mm
Noctua NH-U12P	0.44 mm	2.63 mm
Noctua NH-C12P	0.47 mm	2.54 mm
Noctua NH-D14	0.43 mm	2.33 mm
Thermalright Archon SB-E	0.49 mm	2.33 mm
GELID Tranquillo Rev.2	0.40 mm	2.30 mm
GELID GX-7 Rev.2	0.31 mm	2.25 mm
Phanteks PH-TC14PE	0.40 mm	2.21 mm
be quiet! Dark Rock 2	0.38 mm	2.22 mm
Prolimatech Armageddon	0.51 mm	2.08 mm
Prolimatech Megahalems	0.50 mm	2.00 mm
Zalman CNPS10X Quiet	0.40 mm	2.00 mm
Scythe Kabuto & Zipang 2	0.34 mm	1.94 mm
NZXT Havik 140	0.41 mm	1.91 mm
Scythe Mugen-2	0.31 mm	1.89 mm
Swiftech Polaris 120	0.43 mm	1.85 mm
Thermalright Venomous X	0.53 mm	1.84 mm
Noctua NH-C14	0.38 mm	1.79 mm
Enermax ETS-T40	0.40 mm	1.79 mm
Scythe Yasya	0.32 mm	1.78 mm
Cogage TRUE Spirit 1366	0.40 mm	1.70 mm
Arctic Cooling Freezer Xtreme Rev.2	0.30 mm	1.70 mm
Scythe Grand Kama Cross	0.38 mm	1.66 mm
Reeven Kelveros	0.47 mm	1.61 mm
Zalman CNPS9900 MAX	0.16 mm	1.59 mm
Thermalright Silver Arrow	0.32 mm	1.57 mm
Cooler Master Hyper 212 Plus	0.43 mm	1.54 mm
Thermalright Ultra-120 eXtreme Rev.C	0.56 mm	1.52 mm
Zalman CNPS10X Extreme	0.42 mm	1.50 mm

Testing on larger heatsinks are done on our
LGA1366 heatsink testing platform, while smaller coolers tackle our LGA1155 heatsink testing platform. A summary of the test system
and procedure follows.

Key Components in LGA1366 Heatsink Test Platform:

• Intel Core i7-965 Extreme
  Nehalem core, LGA1366, 3.2GHz, 45nm, 130W TDP.
• Asus
  P6X58D Premium ATX motherboard. X58 chipset.
• Asus
  EAH3450 Silent graphics card.
• Intel
  X25-M 80GB 2.5″ solid-state drive. Chosen for silence.
• 3GB QiMonda
  DDR3 memory. 3 x 1GB DDR3-1066 in triple channel.
• Seasonic X-650 SS-650KM
  650W ATX power supply. This PSU is semi-passively cooled. At the power levels
  of our test platform, its fan does not spin.
• Arctic Silver
  Lumière: Special fast-curing thermal interface material, designed
  specifically for test labs.
• Noctua 140 mm fan (used when possible with heatsinks that fit 140x25mm
  fans)
• Nexus 120 mm fan (used when possible with heatsinks that fit 120x25mm
  fans)
• Nexus 92 mm fan (used when possible with heatsinks that fit 92x25mm
  fans)

The systems are silent under the test conditions, except for the CPU cooling
fan(s).

Normally, our reference fans are used whenever possible, the measured details
of which are shown below.

Reference Noctua 140mm fan Anechoic chamber measurements
Voltage	SPL@1m	Speed
12V	1250 RPM	28~29 dBA
9V	990 RPM	21 dBA
7V	770 RPM	15~16 dBA
6V	660 RPM	13 dBA

 

Reference Nexus 120mm fan Anechoic chamber measurements
Voltage	SPL@1m	Speed
12V	1100 RPM	16 dBA
9V	890 RPM	13 dBA
7V	720 RPM	12 dBA

 

Measurement and Analysis Tools

• Extech 380803 AC power analyzer / data logger for measuring AC system
  power.
• Custom-built, four-channel variable DC power supply, used to regulate
  the fan speed during the test.
• PC-based spectrum analyzer:
  SpectraPlus with ACO Pacific mic and M-Audio digital
  audio interfaces.
• Anechoic chamber
  with ambient level of 11 dBA or lower
• Various other tools for testing fans, as documented in our
  standard fan testing methodology.
• SpeedFan,
  used to monitor the on-chip thermal sensors. The sensors are not calibrated,
  so results are not universally applicable. The hottest core reading is used.
• Prime95,
  used to stress the LGA1366 CPU heavily, generating more heat than most real applications.
  8 instances are used to ensure that all 4 cores (with Hyper-threading) are
  stressed.
• CPU-Z,used to monitor the CPU speed to determine when overheating occurs.
• Thermometers to measure the air temperature around the test platform
  and near the intake of the heatsink fan.

Noise measurements are made with the fans powered from the lab’s 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 Prime95 to stress the processor, and the
graph function in SpeedFan was used to ensure that the load temperature is stable
for at least ten minutes. The temperature recorded is the highest single core
reading. The stock fans were tested at various voltages to represent a good
cross-section of airflow and noise performance.

The ambient conditions during testing were 10~11 dBA and 21~23°C.

Stock Fan Measurements

As we mentioned earlier, the fans are attached using metal stands so they lack
the traditional square frame housing found on most case fans. However, the impeller
and blades are similar in size to standard 120/140 mm models. The CNPS9900 MAX
was equipped with a 1600 RPM fan while the CNPS9900DF uses lower speed variants.

Zalman CNPS9900DF: Stock Fan Measurements
Voltage	Speed (Front)	Speed (Center)	Combined SPL@1m
12V	1060 RPM	1370 RPM	32 dBA
9V	870 RPM	1070 RPM	26 dBA
7V	710 RPM	830 RPM	20 dBA
6V	610 RPM	700 RPM	17 dBA
5V	520 RPM	550 RPM	14 dBA
Measuring mic positioned 1m at diagonal angle from the center of the heatsink. Ambient noise level: 10~11 dBA.

We’re not crazy about different size fans on one heatsink due to possible intermodulation
effects on noise. It’s also inefficient if each fan’s individual noise output
is significantly different, i.e. one fan drowning out the other when set to
the same voltage/PWM setting. The CNPS9900DF wasn’t bad in this regard despite
a 300 RPM difference at full speed. The center 140 mm model was the faster fan
and thus produced more noise, though subjectively, it sounded better.

At full speed the pair generated a pretty loud 32 dBA@1m but if you’re noise
conscious, the fans can be reduced down to about 14 dBA@1m through voltage control;
the starting voltage was 4.8V and 4.0V for the front and center fan, respectively.

The quality of noise emitted by the 9900DF was simply put, bad. It generated
tonal peaks at multiple frequencies in the 150 to 1,000 Hz range. It made an
unpleasant, harsh drone. At lower speeds, there was also an underlying “wobble”
to the sound, as if one of the fans was off-kilter. This effect was audible
in each individual fan, but particularly in the front model, so it wasn’t a
result of interaction between the two. The overall acoustics were very disappointing
as Zalman’s previous model, the CNPS9900 MAX, actually had a fairly nice sounding
fan for a change.

The cooler’s mismatched fans contributed to the 9900DF’s poor acoustics. Both
fans had noticeable tonality but they also had distinctive noise signatures.
The front fan had a harsher, lower pitch profile, and was the main source of
the aforementioned “wobble” effect. The center fan had a smoother
profile with a higher pitch but it produced a dry, undesirable hum.

Test Results

Zalman CNPS9900DF vs. CNPS9900 MAX
Fan Voltage	CNPS9900DF		CNPS9900 MAX
	SPL@1m	°C Rise	°C Rise	SPL@1m
12V	32 dBA	43	41	32~33 dBA
9V	26 dBA	46	42	26~27 dBA
7V	20 dBA	48	45	20 dBA
6V	17 dBA	50	47	16 dBA
5V	14 dBA	53	49	13 dBA
°C Rise: Temperature rise above ambient (22°C) on load.

To our surprise, the CNPS9900DF turned out to be less proficient than its single fan predecessor, the CNPS9900 MAX. Despite its second fan and larger heat dissipation area, the DF performed 2~4°C worse depending on the fan speed.

The main culprit for its disappointing performance was the heatsink’s concave
base. Despite tightening the cooler to the point that the bolt threads were
almost completely stripped, the thermal compound pattern left behind showed
poor contact, with large tendrils of TIM all over the surface, which only happens
when there are significant air caps between CPU and heatsink base. When good
contact and pressure is achieved, most of the compound is squeezed out toward
the sides; that obviously didn’t happen here.

Heatsink Comparison Tables

CPU Coolers (ref. 120mm fan): °C Rise Comparison
Heatsink	Fan voltage / SPL @1m*
	12V	9V	7V
	15~17 dBA	12~13 dBA	11~12 dBA
Thermalright Ultra-120 eXtreme Rev.C	38	40	43
Thermalright Venomous X	38	41	43
Prolimatech Megahalems	38	41	44
Noctua NH-U12P	39	42	44
Scythe Mugen-2	39	42	45
Cogage TRUE Spirit 1366	40	42	45
Prolimatech Armageddon	40	42	46
Zalman CNPS10X Quiet	40	43	46
Noctua NH-C14	39	42	48
Scythe Yasya	41	43	47
Thermalright Ultra-120 eXtreme	40	43	48
Cooler Master Hyper 212 Plus	41	44	48
Thermalright Archon SB-E	42	43	49
Thermalright Ultra-120	42	45	49
Titan Fenrir	43	46	50
Scythe Ninja 3	44	47	49
Enermax ETS-T40	44	46	50
Noctua NH-C12P	43	47	51
Zalman CNPS10X Extreme	43	47	53
Swiftech Polaris 120	46	49	54
GELID GX-7 Rev.2	47	50	52
Zalman CNPS10X Flex	45	50	54
be quiet! Dark Rock 2	48	50	52
Cooler Master V8	46	50	54
GELID Tranquillo Rev.2	48	50	53
Reeven Kelveros	47	51	55
Scythe Grand Kama Cross	45	52	57
Zalman CNPS9900DF (stock fans)	50	53	N/A
	17 dBA	14 dBA
Antec Kühler H20 620 (pump at 7V, almost inaudible)	52	52	53
Arctic Cooling Freezer Xtreme Rev.2	49	52	58
Scythe Kabuto	51	53	60
*Note: there are minor differences in measured SPL due to the variety of fan orientations and mounting methods offered by the compared coolers.

Though it has two fans, given its size, we decided not to throw the CNPS9900DF
against top-tier dual fan towers like the recently reviewed Phanteks
PH-TC14PE. Instead, we mercifully placed in our single 120 mm fan cooler
chart. Even with this lesser field of competition, it came out near the bottom
at similar noise levels to heatsinks paired with our reference Nexus 120 mm
fan.

MP3 SOUND RECORDINGS

These recordings were made with a high
resolution, lab quality, digital recording system inside SPCR’s
own 11 dBA ambient anechoic chamber, 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.

These recordings are intended to give you an idea of how the product sounds
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. 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 recording starts with 5~10 second segments of room ambiance, then the fan
at various levels. For the most realistic results, set the volume so that
the starting ambient level is just barely audible, then don’t change the volume
setting again.

• Zalman CNPS9900DF with stock fans at 1m
  — 5V (14 dBA@1m)
  — 6V (17 dBA@1m)
  — 7V (20 dBA@1m)
  — 9V (26 dBA@1m)
  — 12V (32 dBA@1m)

FINAL THOUGHTS

Zalman’s signature flower design has gone through a lengthy period of evolution,
upwards of a decade. Compared to its immediate predecessor, the
CNPS9900 MAX, the main difference is that the CNPS9900DF is larger
and armed with an extra fan. We weren’t impressed with the MAX, and unfortunately,
in the the DF, none of the issues we pointed out in the earlier model were addressed.

Our samples of the CNPS9900 MAX and LED models both had flat bases. The 9900DF
sample, in comparison, has a slightly concave surface that doesn’t make enough
contact with our CPU heatspreader. Concavity in a HS base is an elementary no-no.
This critical error is exacerbated by the lack of pressure supplied by the mounting
system. We did not like it on the MAX, finding it both insufficient and inconvenient,.
This time around, we had severe, hair-pulling difficulty with it. Upon reflection,
it might be one of the worst mounting systems ever devised. The mounting clips
are thin and bend easily, and the use of tiny hex screws that are so difficult
to reach boggles the mind.

The DF is larger than the MAX with more surface area to take advantage of the
added airflow from the second fan but the two fans are different, in size, speed,
and acoustic character — they don’t even use the same connectors! The fans
sound different enough that they don’t drown out each others’ unpleasant characteristics.
Instead they mix together into a muddy symphony of undesirable tones. Some reprieve
can be found at very low fan speeds but at the sacrifice of any appreciable
cooling performance. Being frameless, propped up precariously on two thin metal
legs with no support on the sides doesn’t help the fans at all. (In Zalman’s
defense it might have seemed like a good idea, as it is the same engineering
principle used successfully for decades by real estate agencies and political
campaigns to put up lawn signs.)

We can understand Zalman’s reluctance to abandon the core radial fin design,
and admittedly, they have one of the most attractive lines of heatsinks ever
produced. But the execution has to be so much better than shown by our 9900DF
sample for any chance of competitive success. Each successive flower heatsink
version has used iterative improvements — more surface area, nickel-plating,
a larger fan, a second fan — to improve performance, but without a sensible
execution ensuring all the basic details are well covered, there’s no way this
historic series can compete. For US$90 MSRP, a CPU heatsink must provide
top performance… and if our sample is representatively, the 9900DF doesn’t
even get close.

Our thanks to Zalman
for the CNPS9900DF CPU cooler sample.

* * *

SPCR Articles of Related Interest:

SilverStone Heligon HE02: Monster Fanless CPU Cooler
Prolimatech Panther CPU Cooler
Phanteks PH-TC14PE Dual Fan CPU Heatsink
GELID GX-7 & Tranquillo Rev.2 CPU Coolers
be quiet! Dark Rock 2 Tower Heatsink
Enermax ETS-T40: Direct-Touch Heatpipe Cooler

* * *

Discuss
this article in the SPCR forums.