Viewing page 1 of 7 pages. 1 2 3 4 5 6 7 NextFan Round-up #6
June 3, 2012 by Mike Chin
Four years have passed since SPCR's last fan roundup. Some will say this is an inordinately long time between reviews in the computer industry where rapid and constant change is the norm. In the DC fan sector, however, this is not true. The technology of DC axial fans has seen very little fundamental change in years. Mostly, what has happened in the past four years is that there are more brands offering consumers a variety of quieter fans, with improved mounting techniques that try to reduce the transfer of vibration from the fan to the chassis or heatsink.
Long time readers of SPCR will remember that a new way of assessing fans was introduced in my article Fan Test System, SPCR 2010. Those who have not read the piece are urged to at least give it a scan, if not a full read, because it is the system used to assess fans here, with one relatively minor change, a new Thermalright heatsink designed for use with a 170mm fan. The original setup employed a 120mm fan heatsink, but with bigger fans being so widely used these days, the change to a larger heatsink that could mount bigger fans was deemed necessary. This will allow us to run tests on 80mm to 150mm diameter fans on the same test platform.
The gist of the 2010 article, for those unwilling to read it:
- Airflow (cited commonly in Cubic Feet per Minute) is a useful specification which fan makers provide mostly for the sake of buyers. It is an oversimplified single parameter that can be used to compare fans. As with horsepower for automobiles, when regarded solely by itself, it does not have a direct or clear relationship to cooling performance. That relationship is non-linear and varies through the power or airflow curve.
- Cooling in a PC depends on several interrelated factors. The relationship between measured airflow and temperature change is non-linear even for a single combination of heatsink + CPU and fan. It never scales up and down in a linear way.
- CFM is extremely difficult to measure, and fan companies use very complex, expensive systems to obtain the CFM value they cite.
- Static pressure is another often cited parameter, again, difficult to measure, and difficult to correlate to cooling performance.
- We shift our focus from fan airflow measurements to temperature change in a tightly controlled environment, so that the cooling effectiveness of a fan at specific sound levels is assessed instead. In other words, we don't really care how much air a fan blows, but what it provides in terms of cooling and at what sonic cost.
- Aside from the de-emphasis on airflow measurement, another big change since
the last fan roundup is the SPCR
Hemi-Anechoic Chamber and associated audio analysis system. The chamber
was created and the new test gear acquired in late summer 2008. They were
not featured in any of the previous fan roundups. Our instrumentation and
the ambient noise in previous tests was limited to about 18 dBA@1m; it was
not possible for us to measure any lower. In the chamber today, the ambient
level is 11 dBA, and our instrumentation is capable of measuring accurately
slightly below that level. Regular readers know that this sound test system
has been employed for all of our heatsink/fan and case reviews since 2009,
and we have measured fans at or near the 11 dBA@1m limit of the system.
Our main goals are to determine the effectiveness of a fan's cooling airflow, the noise it generates, and the nature of its acoustics.
THE TEST PROCEDURE
The gist of the fan test system and procedure is fairly straightforward, but extensive:
- A copper block with the same dimensions as an Intel i7-1366 processor has a small heater coil embedded within, capable of handling 150W.
- A large heatsink is mounted to cool the above CPU simulator.
- The heater coil is powered by a regulated lab power supply to 137W. (Typically,
64.6VDC x 2.1A). This is the maximum power that the heater coil can pull from
the lab power supply, although there is headroom in the coil as well as in
the power supply, which is rated for a maximum of 3A at 64V (192W).
- The fan to be tested is mounted on the heatsink and driven by a regulated 0~12 VDC power supply at standard RPM points.
- A PWM fan controller is used with PWM fans, at standard RPM.
- The SPL is recorded (in dBA@1m) in the anechoic chamber at every RPM.
- The temperature of the CPU block, and that of the air 6" in front of
the fan, is monitored closely using T-type thermocouple wire sensors and a
dual-input digital thermometer.
- A precision anemometer is used to record air velocity (Feet Per Minute) at every speed and SPL.
- The most important test results are Temperature Rise vs SPL.
We refer so often to temperature rise at SPCR that we sometimes forget that not everyone lives and breathes it. Temperature rise refers to the difference between ambient temperature and the temperature of an object under themal load. Better cooling results in lower temperature rise; worse cooling results in higher temperature rise. In this case, the ambient is the temperature of the air 6" in front of the fan, and the thermal load temperature is that of the CPU die simulator.
In planning the new test system, we had a choice to make regarding how the data would be presented, which has a strong bearing on how we conduct the tests. There are three main ways:
1. Use SPL as the reference, and select several targets. Present temperature
rise, RPM and FPM information for each fan at each SPL. For a noise-centric
site like SPCR, this would seem to be the best approach... but SPL is not something
we can simply set or dial in, it has to be obtained by adjusting the fan speed
manually, while monitoring the noise. Trial runs showed that this method took
the longest time.
2. Use Temperature Rise as the reference, and select one or more targets. Present noise, RPM and FPM information for each fan when the target temperature rise points are met. Again, temperature rise cannot be simply dialed in, it has to be arrived at by some experimentation. This process also took a bit more time than the method we chose on the basis of simplicity.
3. Our chosen option: Use RPM as the reference, select several targets, and present noise, temperature rise, and FPM for each fan when the targets are met. This allows us to simply dial in a speed and record the data at each RPM. It's a much quicker process than the others.
Using RPM also has another really important, practical advantage: For
most computer users, RPM is the fan/cooling data that is most readily accessible,
and controllable. Almost every fan in computerland these days offers
RPM data output, and every motherboard has the ability to monitor it. If you
set the speed of your selected fan at one of our test points, you know exactly
what noise level (within a decibel or so) will obtain. There are many ways to
adjust fan speed: Most motherboards are equipped with speed controllers for
their fan headers, and monitor fan speeds for any standard 3-pin fans or 4-pin
PWM fans, and the RPM can be displayed right on the desktop using any number
of fan and/or thermal utilities.
So what reference RPM are the fan test points?
In the past, 12, 9, 7 and 5 volts were used as test points. It made sense for a long time, as these voltages are fairly easy to obtain in any PC (except for 9V). Since we have little reason to change our long-standing reference of the Nexus 120 fan, its speed at 12, 9, 7 and 5 volts are used for standard test points for all 120mm fans: 1080, 880, 720, and 550 RPM. For convenience, those RPM points have been rounded to 1100, 900, 700 and 550. For faster fans, 1500 and 2000 RPM were chosen, somewhat arbitrarily, as well as the top speed of the fan. Again, for convenience, these RPM points are also used for all fans larger than 120mm.
Long experience has shown that neither noise nor cooling is affected by changes in fan speed that are lower than ~50 RPM. I did not sweat to make the targets exactly, but they were always better than 50 RPM within target, as measured by the stroboscope.
At 550 RPM, most good quality fans are inaudible and provide very little cooling, but it's included for the sake of continuity with out past.
Although we take many measurements with machine instruments, we always base our final recommendations on how a fan sounds subjectively.
Typically, there is not enough variance in the objective measurements alone to make clear
distinctions, yet differences can be heard even with fans that measure similarly, even well below 20 dBA. We've always said what really counts is what we hear.
FAN TEST GEAR
It's a long list.
- i7-1366 CPU die simulator with embedded T-type Thermocouple wire
-- A generous contribution from Thermalright. It can handle up to 150W,
but its heat distribution is somewhat more even than a typical CPU. The
main thing is that it gets hot enough, with extreme consistency, and there
are no worries about a CPU or motherboard breaking down.
- Thermalright Archon heatsink -- It's a good performer like most Thermalright
CPU heatsinks, and it can fit very large fans. It is also quite responsive
to the size of fan used due to its big mating surface area for the fan. Given
the same RPM, for example, a 140mm fan always results in lower temperature
than a 120mm fan. For a fan test platform, this is as it should be.
6030D DC Regulated Power supply, 0-64V/3A -- It heats up the
CPU die simulator with power up to 137W.
- For Voltage fan speed control, we use a custom built 0~12 VDC Regulated
Voltage Fan Controller -- The same one used for years and years. It is
sometimes used for PWM fans when the lowest test speed is not achievable on
the PWM fan controller.
- For PWM fan speed control, Fan Xpert 2 utility in Asus P8Z77-V Pro motherboard -- Larry Lee reported on this utility in his review of the board, and it is great to work with to test fans. You'll appreciate the detailed data summary it generates. It also incorporates a voltage regulation circuit for its non-CPU 4-pin headers, which allows 3-pin non-PWM fans to be analyzed using its auto-tune function, and to run the entire test on the fan when appropriate. It has too conservative a definition of "safe starting speed", which prevents many 3-pin fans from running at very low (but still safe) speeds.
- Kanomax 6803 Vane Anemometer
-- ±1% accuracy rating, which is believable. This is by far the most
accurate of the handful that we've acquired over the years. Ironically, it
is used not as a primary tool, however, but a secondary one as we're not concerned
about airflow per se, but its thermal effects in a cooling system.
- Mannix DT8852 Dual Input Thermometer (K, J or T Thermocouple input) --
Supposedly 0.1% accurate. This is to monitor the temperature of the CPU
die and the ambient air ~6" in front of the fan intake
- High accuracy general purpose Multimeter
- Guangzhou Landtek Instruments Scroboscope DT2350P (primary tachometer) -- This is supposed to be accurate to 0.1%.
- Laser digital tachometer by Neiko Tools USA (alternate tachometer) -- This is supposed to have 0.05% accuracy, but I don't trust it as much as the strobe, it requires a reflective tape to be stuck on a blade, often gives false readings (like 9687 RPM when measuring a fan spinning at ~700 RPM)) and doesn't work well with light colored fins.
- SPCR hemi-anechoic chamber
and audio analysis system.
This is the core of our fan testing setup: Clockwise from the left
lower corner, optical infrared tachometer, Mastech power supply for the
CPU simulator, Kanomax anemometer atop the Mastech, Thermalright Archon
heatsink mounted atop with Nexus 120 fan (a reference), multi-channel
voltage controller powered by ATX PSU mounted in an old Shuttle case,
Guangzhou Landtek Instruments Scroboscope DT2350P, and finally, the Mannix
dual-channel thermocouple wire thermometer. In actual use, the lab table
becomes far more cluttered, as you can see below...
Click for larger image
This is a more complete picture of the testing setup in the anechoic
chamber, with the exception of the audio test system, of which only the
mic is shown (placed just so it would show up in the photo). The output
from the mic runs out of the room to the mic preamp and PC running SpectraPLUS
in the adjoining room. The main difference between the top photo and this
one is the inclusion of the Asus P8Z77-V Pro based fanless system
whose 4-pin fan headers are used with the Fan Xpert 2 utility (open
on the monitor screen) for PWM fan speed control and analysis. The Mastech
lab power supply is the only piece of gear on the lower shelf that is
part of the fan test system. Yes, those are floppy disks, used infrequently
for some HDD testing, and there is another open-air fanless PC on that
second shelf. Despite all of the pictured gear, the only noise in the
room comes from the fan(s) being tested.
The i7-1366 CPU simulator, with T-type thermocouple wire embedded. Blue
wires are used to power the heater coil in the copper block. The mounting
block is a piece of phenolic resin (like Bakelite), which has high heat
The mounting system of the Thermalright Archon was incompatible with screw holes in the CPU simulator base, so an alternative system had to be used. I adapted the mounting system from one of the Prolimatech heatsinks, and it works fine. Note multiple fan mounting wire clips for different size fans. An old-style Nexus 120 fan is mounted.
This is the casing of a Shuttle Zen PC, gutted to squeeze in an old Fortron-Source
Power 300W ATX12V PSU and a Sunbeam 4-channel fan controller.
The Sunbeam is unusual in that it allows the voltage to be varied from
maximum to 0V. The 12V output of the PSU has been tweaked up to nearly
13V in order to compensate for the voltage drop through the fan controller.
The fan in the original PSU has been removed; it is not needed for the
tiny loads of 12VDC fans. A separate multimeter is used to set fan voltage.
A Mannix DT8852 Dual Input Thermometer keeps track of temperature at the
top of the simulator CPU die, and of the intake air ~6" in front of
the fan; the difference between these two temperatures is key. A Mastech
6030D DC Regulated Power supply can provide up to 64V at 3A. 137W was
chosen to be the standard load. The stroboscope is one of several
tools used to measure fan RPM.
The Kanomax 6803 Anemometer came with a super low friction calibrated
Pacer 275 vane probe. This is a high precision tool; the tiniest bit of
airflow is enough to get the vane spinning. Airflow is measured in FPM
(feet/minute) directly at both intake and exhaust sides of the fan, in free air. The highest sustained value is recorded for each speed.
The foam harnesses used in earlier fan test are still used for acoustic
and airflow measurements, and recordings. I tried doing this while the fan was mounted
on the heatsink, but rattling or buzzing of the fins on the heatsink can occur. The foam harnesses
minimizes such effects; you can say, generally, that the SPL measurements and the recordings of
the fans represent them at their best or quietest... although in a few
odd exceptions, some fans sound better mounted on the heatsink.
The following fans are reviewed in this roundup:
- Nexus D12SL-12
- Thermalright TR-TY170
- Noiseblocker M12-S1, S2 & P
- Scythe Gentle Typhoon 120 - 12 & 14
- Antec TrueQuiet 120 and TrueQuiet Pro 120
We thank all of the manufacturers for supplying the many fan samples.
|Help support this site, buy from one of our affiliate retailers!|