Howdy all....this is long awaited and long delayed....but I'm very excited to be able to finally put up the finished version today!
First off, I'd like to talk about the market, objectives, methods, and use of these results.
1) The fan market is an interesting one....there are at least a hundred different 120mm fans to choose from, all with different color, noise, CFM, pressure, lighting and radiator characteristics. Theoretically, there should be a different fan for every use....and a different fan for everybody's setup.
There's an issue with it though....manufacturers lie about fan performance ratings. Well, unless you test, you can't confirm that, but let's say that some fans have ratings that just seem 'too good.' How has the community dealt with this? User experience and rationale....there is one rating which is typically very accurate as it can be easily measured by users: RPM.
Through the user experience, the community (well, XS at least) has settled in on a few favorite fans. Is this justified? Well, as you'll see, not all fans are created equally.
2) Well, since some fans are just blatantly mis-rated, and others seem just 'too good,' I became very interested in the true performance of fans, on a consistent and reliable platform. I looked around the internet, and while there had a been a few reviews out there...they were either too focused on silence for my liking, and/or fell short on technique and results.
I hunted down proper hardware to test a fan's CFM output and dB measurements. After gathering the testing equipment, I formed a methodology that would provide extremely consistent and telling results.
3) To measure CFM, I purchased an anemometer that could accurately measure the 120mm equivalent of ~2CFM all the way up to ~350CFM. Rather than just put the anemometer up to the fan and draw a conclusion based on air moved at the hub (which is almost useless), I have shrouded the entire area between the anemometer and the fan. This is to ensure that all airflow is measured, not just the airflow at the hub or another part of the fan. On top of that, because air coming out of a fan, in addition to having forward momentum, has rotational momentum, I have the fan pull air through the measurement setup. Initial results, compared to trusted ratings (Delta, Panaflo, Yate Loon) were EXTREMELY accurate. CFM testing was done at 1V increments between 4V and 12V when possible. Minimum running voltage was also tested. This test is to test the validity of manufacturer ratings and also CFM pushing capabilities as a case fan.
To measure the restricted CFM, I fastened a Swiftech MCR120 radiator between the fan and the measurement setup. This would put a consistent restriction on the fans, and because it was also a commonly used radiator, it allows us to draw conclusions which fans are best for a typical radiator setup for watercoolers, especially those using a shroud to reduce effects of the hub. Restricted CFM was tested at 1V increments between 4V and 12V when possible. Minimum running voltage results were extrapolated from 4V results.
To measure decibel ratings, I fastened the fan on an exhaust hole on a case and positioned my decibel meter 12cm away from center of the fan. It was angled from the side such that it would not suffer from any buffeting. The dB meter is isolated from any external vibration through bubble wrap. dB readings are used throughout the comparison and are measured from 4V to 12V, as well as minimum running voltage.
To measure CPU tempurature, I used Thermaltake Big Typhoon VX on a Celeron D and P5W DH configuration clocked at 3.6GHz and 1.63V (loaded). This setup was used because the TTBT can easily mount ANY 120mm fan, responds well to increased airflow, and because it is a very good cooler also capable of keeping the Celeron stable with very low airflow. The Celeron D was used because it runs extremely hot yet can also run at extremely high temperatures stably (87C loaded was the highest temperature I saw, and it was stable). Speedfan and OCCT were used to load, measure and record temperatures. A UEI DT200 was used to measure ambient tempuratures. WCG was run as an 'idle state' in order to prevent the processor from cooling down too much, allowing me to use a shorter (8 minutes rather than 30) load time to assess performance. If a processor did NOT cool down at the completion of the 8 minutes of load, the test was repeated until it did (always the next try). The maximum temperature readout was used as loaded temp. Ambient temperature was subtracted and gives use a dT, the basis for determining which fan is ideal for CPU cooling. This test was done because restricted CFM alone cannot be used to determine fan performance cooling a CPU due to varying hub sizes and air dispersion patterns. The test was performed at 12V, 9V, 7V and 5V when possible on every fan.
4) Having tested the methodology multiple times and having tested each fan at least twice, I am extremely confident these results are accurate, repeatable and revealing. This allows me to test fans later and obtain results that are directly comparable. This is key.
There may seem to be results that, especially with the use of lines connecting the points, seem out of line and I recognize them as well....they are repeatable results and I'm not 100% why fans sometimes behave like that
There are also some fans that have extremely long 'lines' connecting the points. This is because these fans do not scale linearly with voltage (most fans do). In these cases, the manufacturer supplies a small rheobus to change the fan's supplied voltage. The rheobus is specially designed for the fan and typically only varies the supplied voltage between ~8.5V and 12V. This ensures the voltage cannot go too low to prevent startup. Anyway, long story short....with these fans, at around 9.5V, the RPMs suddenly drop off almost completely.....this provides both a silent and high-CFM fan in one setup (though not simultaneously). Because I test at 1V increments, often times a fan will be powerful at 10V and then at 9V will be spinning slowly and due to the organization of results, that results in very separated scatter points for those fans. Testing at intermediate voltages with a resistor or user-supplied rheobus is EXTEMELY difficult due to the exactness needed in pinpointing voltage levels. It was therefore not done.
I supply a set of graphs with 'adjusted dB' that can be roughly used to judge whether or not a fan manufacturer is honest about their ratings. These are non-scientific....just a simple linear adjustment of measured dB determined from a group of trusted manufacturer ratings (Delta, high-speed Panaflos, Yate Loon). It in no way changes the results, just the scale they are on. I would put their 1meter accuracy at around 95%.
Hopefully they can be used to put perspective on the dB readings in the scale we are all used to dealing with.
n.b., yes, I realize there are some issues with some of the later graph's titles and axis titles....I'll fix them later, they are minor though
EDIT: I would also like to have a big shout out to member Baenwort who provided many of these fans for testing, it would only be a fraction of the 26 if he hadn't lent me part of his collection.