You are here

Apack ZeroTherm BTF80 & BTF90 CPU Heatsink/Fans

August 5, 2006 by Devon

Apack ZeroTherm BTF80 / BTF90

Socket 775 / K8 CPU Heatsink
Market Price

Apack's ZeroTherm BTF80 and BTF90 were hard to take seriously when
they first showed up at the SPCR sound lab. A heatsink shaped like a butterfly?
It's a joke, right? They look too frivolous, they can't possibly be any good.

No joke. Silly as they might look to some, these CPU coolers are serious efforts. Just
think of the shape as a means of differentiating the heatsinks for the dozens
of other high-end heatsinks out there. If ASUS can sell a
heatsink that resembles a jet engine
, why can't Apack sell one that looks
like a butterfly?

Viewed from the top, their profile looks distinctly like a butterfly.

There's little question that the effect is intended.

As far as we know, this is Apack's first foray into the the retail heatsink
market. The Korean company has a background in making heatpipes and CPU heatsinks
for the OEM market. It has business relationships with several major players,
including Samsung, LG, and Intel, and appears to have money behind it. Unfortunately,
distribution in the English speaking world seems practically nonexistent at
the moment, but hopefully that should change with time.

There are two versions: An all-copper model for those who feel they need a
little extra performance, and an aluminum-finned model for those worried about
weight. According to Apack's specifications, the copper version is slightly
more efficient than the aluminum version, raising the cooling capacity from 140W
to 150W.

If the butterfly shape wasn't enough, the slick retail packages should tell
you that this heatsink isn't destined for the OEM market.

Neatly and safely packaged in a plastic blister pack.

Brackets for Socket 775 and K8 systems (including AM2) are included.

Apack ZeroTherm BTF80 / BTF90: Specifications
108 (L) x 81 (W) x 128 (H) mm


BTF80: Aluminum / BTF90: Copper


Copper (Sintered type)
Heat Dissipation Area
4,404 cm² (682.6 inch²)
Cooling Capacity
BTF80: Over 140W

: Over 150W
Fan Size
92 x 25 mm (3.62 x 0.98 inches)
Fan Speed
750 ~ 2,500 RPM ± 10%
Acoustical Noise
Under 27.0 dB ± 10%
4-pin (PWM)
Operating Voltage
12 VDC
Power Consumption
Max. 1.56W
Max. 42.8 CFM
BTF80: 458 g (without optional components)

: 678 g (without optional components)


The butterfly shape lends itself surprisingly well to its role
as a heatsink. Two stacks of fins form the wings and the base of the fan forms
the body. Decorative as they are, the shape of the wings is also quite functional:
Each wing has a total of four evenly spaced heatpipes that run the full vertical
length of the heatsink. The heatpipes are grouped in pairs, each pair connected
at the top of the heatsink so that, even though there are eight vertical pipes
in total, only four long pipes need be manufactured. Doubling up heatpipes in
this way is not uncommon.

Front and back.

The pairs of heatpipes form a V that is roughly followed by the shape of the
wings themselves. In addition to giving the heatsink its distinctive shape,
the contours of the wings cut away fin material that is a long way from the
heatpipes, reducing impedance where heat transfer is likely to be low.

The decorative ridges, the holes on the tips of the wings, and the trailing
"tail" on each wing may also have useful purposes that go beyond aesthetic
appeal. The ridges are likely intended to direct airflow, and they are arranged
to channel the air as close to the heatpipes as possible. The holes in the wingtips
are located where air pressure is likely to be weakest, and should reduce impedance
to some extent. The "tail" fins serve a double purpose: They prevent
stray cables from fouling the fan, and take advantage of the lateral airflow
the comes off the tips of the fan blades.

The fins themselves are thin and loosely spaced — both good signs for
when the fan is turned down. Our only concern is that, with all of the contouring,
there is less surface area compared to many tower heatsinks. The larger the surface area, the better the ability of the fins to transfer the heat from the CPU into the air.

The fins are fixed at the base.

Like the fins, the red plastic body also looks as though it is functional.
The "tail" of the body forms a wedge shape that splits the airflow,
forcing it out sideways instead of allowing it to blow straight through. This
forces the air towards the heatpipes where it is most useful. The wedge shape
continues through the length of the body, with the two antennae pushing the
air outwards as it leaves the fins.

Does this butterfly have propeller-assisted flight?

Although it is necessary for the design of the heatsink, the rear-mounted fan
looks a bit odd attached to the back of a butterfly. With the fan direction
reversed, it looks as though it could be the butterfly equivalent of a wheelchair;
perhaps it is unable to fly without the fan. It even has a taillight: The fan
glows red whenever it is spinning thanks to a red LED embedded in the hub.


Getting back to the technical details, the copper base on both of our samples
was slightly tarnished, but otherwise flat and smooth.

Slight discolorations on the base are unlikely to affect performance.


Although the specifications list the fan dimensions as 92 x 25 mm, a quick
look at the fan shows that this is not the case; it is not possible to simply
drop in a 92mm fan as a replacement. Although the rotor itself may be the same
size and thickness of a more conventional fan, the highly unusual frame
is one of a kind.

The wedge shaped base fits snugly between the two wings, where it is secured
with two machine screws. It can be easily removed by undoing the screws and
pulling the frame up out of the heatsink.

The fan is removed by sliding it up between the wings of the butterfly..

The fan itself is branded with Apack's ZeroTherm logo, and lists the model
number SFC9225LU-12P, the words "Silence Innovations", and a link
to, which yields an "Invalid Hostname" error when
typed into a browser. No information about either the model number or "Silence
Innovations" could be found, so we are left with APack's specifications
for information: 750~2,500 RPM, 43 CFM, and 27 dBA. We find it unlikely that
the fan is 27 dBA (at what distance?) at 2,500 RPM, but you can see our own SPL measurements later in this review. In any case, the fan is not expected
to be run at full speed; it comes with a four pin fan connector that assumes
the motherboard will control fan speed.

The unusual frame is specifically designed to fit the BTF heatsinks.


Socket 478 is rapidly fading into computing history, so it is no surprise
that it is not supported. K8 sockets, including AM2, are supported using AMD's
standard retention bracket, while Socket 775 systems use a backplate and four
screws through the motherboard. A separate bracket for each mounting system
is included, and the appropriate one must be screwed on to the bottom of the
heatsink using four screws. Both mounting clips can be rotated 90 degrees(in fact, 360 degrees in 90 degree increments), which means that the heatsink can be mounted with the fan blowing towards any direction, physical impingements from motherboard components notwithstanding. The ideal setup of the fan blowing towards the back panel (so that the heated air from the CPU can be evacuated by the back panel exhaust fan) can be achieved with just about any motherboard.

Both mounting systems have their problems, but neither is difficult to figure
out. The K8 bracket suffers from an issue that is common with clip-based heatsinks:
The tension is too high for easy removal. Installation is not a problem, since
it's really just a matter of pressing down on the clip and letting it click
into place. However, removal is difficult since the clip must be unhooked from
the retention module while simultaneously holding the clip under pressure. We
found that a pair of pliers came in handy for this purpose. Keep in mind most users do not often install or uninstall their CPU heatsink.

Socket 775 has a different problem: Two of the mountings screws are partially
blocked by the fan blades. There are two ways of getting around this. One is
simply to hold the screwdriver at an angle, avoiding the fan. This is the lazy-man's
solution, but it does make it more difficult to tighten the screws fully. A
better solution is to simply remove the fan during installation; it's only a
couple of screws.

We were pleased to note that the mounting screws on the Socket 775 bracket
are spring-loaded and permanently attached to the bracket. This means no worries about correct tension. The ease of installation cannot be better: Screws only needed to be tightened until they stop turning. Spring-loaded bolts were first implemented years ago by Alpha Novatech (we reviewed their S-PAL8952 a couple years ago), but still rarely seen, and captive spring-loaded bolts goes one step better. In a perfect world, all heatsinks would use this type of tensioning system.


Our socket 775 test bench was used to test the ZeroTherm heatsinks. Details of the system are outlined
below. So far, only six heatsinks have been tested on this system: The
Arctic Cooling Freezer 7 Pro
, the
Spire Verticool II
, the
Arctic Cooling Alpine 7
, the Scythe Ninja, the Thermalright XP-120,
and, most recently, the Zalman
. No other SPCR heatsink test results are directly comparable with
the results of this review. The Pentium 520 used in this test is cooler than
most of Intel's soon-to-be-forgotten Prescott and Presler chips, but it is still
15-20W hotter than the P4-2.8 Northwood used in our socket 478 HS testing platform.
On the other hand, it is also hotter than almost about every AMD processor on
the market, not to mention Intel's new Core 2 Duo chips, many of which draw
much less power than the Intel 520.

On the test bench...

Test Platform

Measurement & Analysis Tools

Noise measurements were made with the fan powered from the lab DC power supply
with everything else turned off to ensure minimal ambient noise. Airflow
measurements for this heatsink were not made due to the difficulty of measuring
the stock fan accurately

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 ambient conditions during testing were 16 dBA and 24°C. This is warmer
than the usual lab conditions, and reflect a recent spate of hot weather. Please
keep this in mind when comparing results.


APack ZeroTherm BTF80 (Aluminum) with Stock
Fan Voltage
°C Rise
APack ZeroTherm BTF90 (Copper) with Stock fan
Load Temp: CPUBurn for ~20 mins.

°C Rise: Temperature rise above ambient at load.

°C/W MP / TDP: Temperature rise over ambient per Watt of CPU heat,
based on CPU's Maximum Power (100W) or Thermal Design Power (84W) rating
(lower is better)

Noise: SPL measured in [email protected] distance with high accuracy B &

The two heatsink models produced results that are similar enough that they can be described
together. Both come with identical fans, so comments about noise apply to both
unless otherwise noted. As a general rule, the copper version outperformed the
aluminum version by about 2°C.

Fan @ 12V: At full speed, the fan was too noisy for use in a quiet
system. However, even users who don't care about noise are unlikely to run
the fan at full speed very often as long as they use the four-pin header to
control fan speed. Almost all newer Intel-chip motherboards and many AMD AM2 boards feature some kind of built-in 4-pin PWM fan speed control. The performance at full tilt is good enough that it is
hard to imagine many people needing the cooling performance of these heatsink/fans at the full 12V.

The quality of noise left quite a bit to be desired. The sound consisted
mostly of a noisy whine, although this was far from the only component in
the noise. Several other pitches could also be heard in the noise, most of
which sounded higher than the primary tone. The multiple tones seemed to come
from the fins, which tended to amplify vibrations from the fan. The poor noise
character was more marked on the aluminum version, although both versions
showed signs of resonance. Both models also exhibited a rapid metallic clicking,
as well as a small amount of turbulence noise.

Fan @ 9V: The noise dropped to an acceptable background
hum, and performance suffered by only 2°C — a barely
noteworthy drop. Unfortunately, the noise character continued to disappoint,
with the metallic clicking growing more prominent as the whine receded into
a low hum. The noise also exhibited an odd throbbing that is clearly audible
in the near-field (one foot) recording at this level. The effect was even
more irritating in person, and continued to irritate even from a greater distance

Fan @ 7V: 21 [email protected] is normally very quiet, but our perception was that the noise from these fans is still more noticeable than we
are used to when we measure so low. Although the primary motor noise had faded
away to almost nothing, a veritable orchestra of noises remained. The mechanical
clicking now dominated the noise character and it seemed to be clicking in
time with the throbbing from before. A low, clattery growl could also be heard.
Cupping a hand around the fins did a good job of making this noise smoother
and more muted, so it seems that much of the noise came from the fins, not
just directly from the fan. There is obviously some condution of vibration from the fan to the fins which is causing some of the noise.

Performance continued to be excellent. The aluminum model managed to equal the copper at this level and no other.
At a comparable level of measured noise, the only heatsink we've tested that
can beat either butterfly is the
Scythe Ninja

Fan @ 5V: The excellent performance continued even at 5V. We would
have no objection to leaving the fan at this level permanently, even with
our burning hot Prescott processor; the heatsink could handle it, even inside
a system. What this means is that only overclockers — and those unfortunate
enough to own a Pentium D or a Prescott-based P4 — should ever have any
need to run the fan above 5V.

At 5V, the fan is very quiet, if not inaudible. The noise character was similar to the noise at 7V, but muted and with a
slower rhythm in comparison. The throbbing was still present, but it finally
dropped to a level that it would probably be hidden by other sources of noise. It still sounded
louder than the measured SPL of 19 [email protected], but it was quiet enough
that it would likely be masked by whatever other sources of noise there are
in the system. It was not quite good enough for a silent system, but those
are few and far between.


We would not recommend running the stock fan above 5V, so it makes sense to
compare the BTF80 / BTF90 at this noise level. Due to the extreme limitations
on airflow that this noise level imposes, many heatsinks do not do well at this
level, as can be seen in the table below. Please note that the comparison is
approximate; data is collected from past reviews, which may have been performed
under slightly different thermal or acoustic conditions. For this reason, not
all of the noise levels are identical, but they are close enough for comparison.

Heatsinks Compared at ~20 [email protected]

(SPL - [email protected])
Fan Voltage
°C Rise
APack ZeroTherm BTF80
APack ZeroTherm BTF90
Arctic Cooling Alpine 64


Arctic Cooling Freezer 7 Pro*
Spire Verticool II SP601B3


Scythe Ninja, Nexus 120


Thermalright XP-120, Nexus 120



The ZeroTherm heatsinks are outperformed only by the Scythe Ninja, and they
come in at roughly the same level as the Thermalright XP-120. That's not bad
considering that these heatsinks achieved the 20 [email protected] noise level using our
reference Nexus 120mm fan, which is both bigger and quieter than the stock fans
on the ZeroTherm heatsinks. Every other heatsink that we've tested on our Socket
775 test bed is beaten decisively at this noise level.


APack ZeroTherm BTF80 & BTF90: 5V-7V-9V-12V, 5s Ambient between levels: One
, One Foot


Zalman CNPS8000: 5V-7V-9V-12V, 5s Ambient between levels: One
, One Foot

Arctic Cooling Alpine 64: 5V-7V-9V-12V, 5s Ambient between levels: One Meter,
One Foot

Scythe Mine w/ stock fan: 5V-7V-9V-12V, 5s Ambient between levels: One
, One Foot

Nexus 92mm fan: 5V-7V-9V-12V, 5s Ambient between levels: One Meter,
One Foot


These recordings were made
with a high resolution, studio quality, digital recording system and are
intended to 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
, 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


Call my manhood into question if you must, but I can't help but be charmed
by these butterflies. I'm not usually one to fall for looks (at least, not where
computer hardware is concerned), but the fact that the butterfly profile appears
to work with the design instead of hindering them is impressive. The excellent
performance shows that, even if the butterfly design doesn't contribute to performance,
at the very least it doesn't seem to hurt.

Either of these ZeroTherm BTF heatsinks is capable
of adequately cooling almost any processor with the fan at 5 volts. The "almost"
in that sentence can be eliminated if you're willing to deal with more
fan noise than we are. We are confident that they can handle even Intel's hot Prescott processors
with under 30 [email protected] of noise. Only heavy overclockers need look for more.

Unfortunately, the proprietary fan is not good enough for use in a system that
is silent — and it will not be easy to replace it with one that is. Good
as the heatsinks are, they are better suited to a quiet gaming system that is
likely to have some residual noise, not a no-holds-barred silent machine.

However, noise and performance are only relevant if you can manage to get your
hands on one of these — which could be a problem at the moment. They do not
seem to be available for purchase anywhere just yet. We
hope that APack manages to get wide distribution of these heatsinks, since they deserve a strong recommendation.


* Excellent performance

* Do you like butterflies?

* Simple installation procedure

* Good low airflow characteristics

* Performs well enough to use fan at 5V

* Uses stock K8 retention bracket

* Fan can be easily "aimed"

* Irritating fan noise

* Fan difficult to replace

* Fan must be removed for socket 775 mounting

* Poor / nonexistent distribution

Much thanks to Apack
for the ZeroTherm BTF80 and BTF90 samples.

* * *

Articles of Related Interest

Recommended Heatsinks

Arctic Cooling Freezer
7 Pro

Scythe SCNJ-1000 Ninja

Thermalright XP-120:
1st 120mm fan CPU Heatsink

* * *

POSTSCRIPT, Aug 7, 2006: The fan just slides out.

It turns out that removal of the fan is much simpler than we originally thought. The fan is actually mounted on a sub-assembly that is plastic parts are tongue and grooved to fit into the main red plastic piece that's the anchor for the fan. Just push up from the bottom or pull up, and it slides out easily as shown in the photos below. It slides and clicks back in securely. There's no need to undo screws and remove the whole fan assembly. Our thanks to Max Page, editor of, for pointing this out.

on this article in our Forums



www SPCR