Spire Fourier IV heatsink/fan

It’s an ambitious, heavy, all-copper, quad-heatpipe heatsink with high end aspirations from a brand not usually in the top ranks. Spire’s Fourier IV has most of the features and technology to excite heatsink enthusiasts. How does it fare?

March 6, 2007 by Mike
Chin

Spire is a heatsink brand that’s been around for over 15 years, but it’s
not instantly recognized by every PC enthusiasts. That’s because the brand has
traditionally been more focused on cost-effective cooling solutions of greater
interest to system integrators than to overclocking and performance fanatics.
Such a focus is certainly no reflection on the quality of the products; Spire
has had some very good products that belie their generally low price tags, a
bit like Arctic Cooling.

Spire’s current product lineup includes dozens of heatsinks, along with PC
cases, fans, hard drive enclosures, and even power supplies — but who doesn’t
offer power supplies these days!? The item examined in this review is an all-copper
heatsink with four heatpipes that’s among Spire’s top models, priced accordingly,
definitely not a budget model.

The Fourier IV comes packaged in a retail box made entirely of
clear plastic. We’ve never liked such packaging because of what they mean for
the environment, during both production and disposal. It would be nice if unbleached
recycled cardboard with a minimum of color ink was used instead.

Way too much plastic in the packaging is environmentally unfriendly.

PHYSICAL DETAILS

The Spire Fourier IV is a fairly sizeable unit that
immediately brings to mind the Thermaright SI-97 or its larger brethren, the
SI-128. Unlike those comparatives, the Spire is an all-copper model, complete with a 92mm fan. There are four heatpipes that join the fins
to the copper bottom base. The main cooling fins are formed of very thin copper
pieces press fitted onto the heatpipes. The gap between the bottom of the fins
and the top of the base is used for the fan, which is secured to the underside
of the fins with two very easy to use wire clips.

The Spire Fourier IV is a good looking HSF. Shown here with stock fan and socket 775 mounting hardware. You can see the wave shape profile of the top edges of the cooling fins,

The side view, without fan or mounting hardware.

It is a big heatsink but because of the way the fan is mounted beneath the fins, it could fit fairly easily in many lower profile cases, such as HTPC cases. Additional clearance of perhaps an inch is needed above the fins for proper airflow, and to avoid turbulence noise, but still it looks promising.

One drawback of copper is that it is substantially heavier than aluminum. Weight is not specified by Spire, but a quick manual comparison against the 665g original Scythe Ninja tells us that this HS is a bit heavier without any fan. 700g is probably a reasonable guess. Add 80-90g for the fan, and we’re up to around 800g. This is no lightweight.

The Fins

Removing the fan and looking straight down shows how thin the
fins are, and how closely they are spaced. The left and right edges of the fins
(as seen in the photo below) are actually closed with a little flap of the fin
that’s been bent. As a result, air cannot flow through from any of the sides,
only straight up or down.

Copper is considerably more expensive than aluminum, and generally considered preferable as a heatsink material, as its thermal conductivity is about 50% better. However, copper is only better if there is enough airflow to ensure that the additional heat conducted into the fins can be moved into the air quickly enough. If the bottleneck is airflow, then copper will not show any advantage over aluminum. This usually means a higher airflow, faster, and noisier fan is need to realize the copper fins’ cooling advantage. Whether copper fins are really advantageous at the very low airflow that comes along with low fan noise is questionable.

The fin spacing is quite tight, but the fins themselves are very thin.

The Base and Mounting Clips

The photo of the base below shows it’s not that smooth. There
are many coarse machining lines which can be both seen and felt without much
difficulty. It is quite flat, however. One of the two fan clips can also be seen.
It requires open flange corners on the fan.

A “ledge” formed by a notch on the ends of the fins is used for the fan wire clip.

The mounting hardware for each of the different types of CPU sockets gets screwed
to the four threaded holes in the corners of the base. The photo below shows
the socket 775 hardware in place.

With 775 socket mounting hardware installed.

The stock 92mm fan is made by Fanner Tech (otherwise known as Shen Zhen), model
FD09025B1M. The B in the model name means ball bearing, and M means medium speed.
The combination of ball bearings and clear plastic is not promising. We have
yet to encounter really good acoustics in any clear plastic fan, and few ball
bearing fans have good sound quality.

We’re always leery of clear plastic fans.

The fan controller is a genuine voltage regulator. It’s similar to the circuit found in the well-know Zalman Fanmate. It has a pass-through for fan rpm monitoring. The circuit and control knob are mounted on a PCI slot cover plate which it is meant to replace so that you can control the fan speed by reaching behind the computer. For a DIYer, it would be a simple matter to remove the control from the PCI cover plate and install it in the front of the case for greater convenience.

Fanmate-like fan control circuit.

INSTALLATION

The Fourier IV is small enough that on our test motherboard, it could be mounted in three of the four orientations. The heatpipes interfered with the northbridge heatsink in the fourth position. In the opposition position, the heatpipes hung only about half a centimeter over the edge of the motherboard. With most cases, there is a bigger gap than that between motherboard edge and bottom of PSU. Fit should not be an issue in the vast majority of systems.

Installing the Fourier IV on our test bench socket 775 motherboard was as expected
for a through-the-motherboard bolt mounting system; it’s a bit of a pain. The
bottom backplate is insulated on one side with rubber, and it’s smooth, which
means there’s no way to make it stay in the right position while you maneuver
the heatsink on the top side of the motherboard to get its mounting holes in
the right place. The machine screws go in from the trace side of the motherboard and thread into the
lugs of the heatsink.

The job would be very awkward for one person without some tool to hold at least
one of the items still — preferably the motherboard. I had help: Nick Geraedts’
two hands to hold the board and heatsink, and his eyes to tell me when the heatsink was in
the right place while I positioned the bottom plate and used a magnetic head screwdriver to
tighten up the screws. Captive screws in the backplate would have helped.

It was awkward to hold backplate, motherboard and heatsink in place while trying to insert a screw with screwdriver to go through holes that needed to be lined up in all three items.

Just how tight to do up the screws was also an issue here. There is no positive physical stop or guide. Not enough or uneven pressure means poor cooling. But overtighten a screw
mount heatsink like this one, and you can crack the motherboard. It could be the
smallest of micro cracks, but if some key trace in one of the many layers on
the PCB gets broken, you end up with an intermittent motherboard useless for
a testing platform. It’s the way at least one of SPCR’s previous test
boards was forced into early retirement. (That and the many dozens of times heatsinks and CPUs were installed and uninstalled on the board.)

Such “how tight is tight enough?”
issues for bolt/screw mounting are completely solved by the pre-loaded spring-in-bolt system used on
heatsinks such as the Thermalright Ultra 120, the
Apack ZeroTherm BTF80/90, the S-PAL8952
by Alpha Novatek, who first brought the concept years ago to large CPU heatsinks, and finally the Swiftech MC462, featured in the very first review of heatsinks at SPCR. The MC462 still had four bolts and nuts, but Swiftech did it better than Spire even back in the stone ages of 2002.

Never mind, deal with the here and now. I turned to the manual for illumination. None was to be found. The manual stated the following:

“Fit each of the screws to each post. Do not tighten them fully until all have been tightened equally cross ways. Note: Rule of thumb is, fully tightened is sufficient. Do nut (sic) use a lot of power to fix the screws into the back-plate. Doing so might brake (sic) or even your motherboard which will not be covered by Spire’s warranty policy.”

That’s not helpful at all, is it? Each of the second and third sentences alone is vague, but the two together are downright cryptic. We’re left grasping for the meaning of fully tightened.

The gap between each corner lug of the heatsink and the motherboard.

I examined the way the heatsink sat atop the CPU before being
secured down. The photo above shows a small gap, perhaps 1/16″ or a touch
more, between each mounting lug and the motherboard. I wondered if this gap was small enough to allow each screw to be threaded as far as it could go. The stiffness of each steel
lug was checked by pressing against them. They seemed extremely stiff. If the screws were done up all the way, the pressure against the CPU and the motherboard would be enormous. A hard spacer washer of the right thickness (between motherboard and HS mounting lug) would have been helpful, but I had no way to judge which of the washers in my bin would be the “right” thickness.

In the end, I decided to leave a small, even gap all around the four lugs. With only a visual check, this was not easy to do, especially as the screws go in from the underside of the motherboard. Afterwards, I crossed my fingers when I booted up
the system. Happily it booted up fine, but I am not happy about the installation procedure or the instructions. Both need some serious attention!

Safely installed, but not without a lot of hassles and worries.

TESTING

Testing was done according to our
unique heatsink testing methodology, and the reference fan, a Nexus 92mm, was profiled
using our standard fan testing
methodology. A 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 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 that most
  realistic loads. 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.

Sound pressure level (SPL) measurements were made with the fan powered from the lab variable DC
power supply with no other noise sources in the room. The ambient conditions during testing were 18 dBA and 22°C.

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 fan was tested at four voltages: 5V, 7V,
9V, and 12V, representing a full cross-section of the its airflow and noise
performance.

TEST RESULTS

The Fan: The first portion of the table below shows mostly information provided by Spire. The yellow portion of the table displays the results of our testing.

The SPL measurements should make it very clear that this is not a quiet fan except at the very lowest settings. The “M” for medium speed seems like a bad case of mislabelling. In our books, this is undoubtedly a high speed, high power fan. Both the ball bearings as well as the clear plastic construction had audible consequences. It was only at the 5V or minimum controller settings that the acoustic signature became benign and quiet enough for the fan to be considered worthy of use in any of SPCR’s own computers. As the 2000 rpm or 7V mark was approached, the noise simply became too much to be acceptable. It was not just a matter of the overall level or “quantity” of noise; its quality was plainly annoying. The recording further below will clarify better than words.

Cooling and Noise

Stock Fan @ 12V: The quick-eyed reader will notice the 5 dBA@1m measured
SPL increase in the cooling
/ noise table above compared to the the fan table in the previous section. It is not an error. (It’s there through the entire speed range.) The increased turbulence of the fan
mounted on the heatsink made it much louder than in our standard fan test setup. There was also transfer of vibration into the heatsink, which added its own tonal sounds to the mix. Naturally, it was far too loud. There was also an odd variability
in pitch as if the fan was changing speed, up and down. This was not the result
of interaction with our power source; the voltage across the fan terminals
was steady. The cooling performance was good but not top notch.

Stock Fan @ 9V: The noise level dropped quite a bit. Subjectively,
it seemed only about half as loud as before, but still much too loud. At this speed,
too, the fan was louder operating on the heatsink than when it was in free air. Cooling performance
was barely affected, which suggests that the airflow at 12V is ineffectively
high.

Stock Fan @ 7V: The stock combination finally approached our threshold of quiet, 30 dBA@1m.
The overall sound quality was not good, however, with traces of resonance
and vibration as well as tonal characteristic audible through the turbulence
noise. Cooling performance dropped 2°C, to a 25°C rise above ambient,
but we’d still consider this pretty good performance.

Stock Fan @ 5V + Controller at Min: Finally, the noise level dropped into acceptable territory. It should be quiet enough for most users here, as long as the quality of the sound is acceptable. Since there are still some tonal aspects to the sound, users reactions will probably vary. Cooling performance fell into the barely acceptable range, however. Setting the controller to the minimum setting brought noise down a titch and CPU temperature up a notch.

Reference fan @12V: What a relief to switch to this fan! As we’ve remarked elsewhere, the reference fan may be quiet enough at 12V for most people. It seems very slightly louder than in free air, due to increased turbulence through the fins. The cooling performance was actually the same with the stock fan and its controller at minimum, but it was smoother and quieter.

Unfortunately, the cooling performance with the reference fan at 9V and below fell into the unacceptable category with ourCPU. If we assume case temperature to be 10°C higher, we’d be seeing CPU temperatures of 67°C and more. That’s not recommended. On the other hand, if your processor is one rated for lower power disspation than the test rig’s fairly hot Pentium D950, the Fourier IV could be good enough even with the Nexus at 9V.

COMPARABLES

To find a heastink/fan of comparable size, we have to go back quite a ways. Our recent reviews have all been of large heatsinks. Here’s a possible list:

Zalman 8000: It was tested a year ago on a different platform with a cooler CPU, and we thought it was a stinker, terrible value for money. The only thing better about it is the better installation system. Currently in the market for around the same price as the Fourier IV, which I’d take over the 8000.

nMedia Icetank: Again, tested a year ago on a different platform with a cooler CPU, but it was a decent performer with a decent fan. Guesstimating puts is performance on par with the Fourier IV, and it came with a better fan. Also has a simple clip mounting system. Similarly priced. We think the F-IV loses out here.

Zalman 7700 / 7000: These go back 2-3 years, but they’re still on the market, they are good performers, and they may be viable alternatives with today’s CPUs. I would personally take them over the F-IV on the hassles of the F-IV’s installation system alone.

NOISE RECORDINGS IN MP3 FORMAT

Spire Fourier IV with stock fan: 4.6V-5V-7V-9V-12V,
with 5s ambient between levels: One
Meter (1mb file!)
The first 4.6V level is with the fancontroller set to minimum.
Recordings at one foot were not made; you can hear clearly that it’s too loud with the one meter recording.

Comparatives:
(Our current databse of HSF recordings done the same way is a bit small.)

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

Zalman 8000 w/ stock fan: 5V-7V-9V-12V, 5s Ambient
between levels: One
Meter, One Foot

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

CONCLUSIONS

Some products are difficult to stay neutral about. The Fourier IV looks so promising, with a nice set of features and technology. Yet it caused much frustration over installation and acoustics. While the heatsink cools quite well with the supplied fan, the cost is a level of noise rarely encountered at SPCR. It’s much quieter when the fan controller is turned right down, but then the cooling becomes questionable. It’s obviously not intended for good cooling performance and low noise.

There is the possibility that the heatsink was not making optimal contact with the CPU, but this is a problem that could afflict any user. We were unwilling to find out, because it meant risking the test platform motherboard by possibly overtightening the mounting screws.

The installation issues are of concern to everyone, not just silencers. Poorly designed mounting systems are not uncommon, but it’s a shame to see one even in such a heavy heatsink with high end aspirations. A little bit of ergonomic thinking could have made a big difference here. Finally, the current market pricing puts this model at a disadvantage against many better designed, better performing, and quieter HSF that we’ve already reviewed.

It’s difficult to recommend this HSF to anyone outside the circle of fanatical heatsink collectors who must try everything interesting in the market. It may be OK for the gamer with a tight case who doesn’t care about noise, but for most others, there are better options whether your priority is cooling performance or low noise.

Much thanks to Spire for the Fourier IV sample.

* * *

Articles of Related Interest

Recommended Heatsinks
SPCR’s unique heatsink testing
methodology
SPCR’s standard fan testing
methodology
Zalman 8000
nMedia Icetank
Zalman 7700 / 7000

*
* *

Comment
on this article in our Forums