Was faster than expected, less text that I had anticipated! Enjoy:
In order to benchmark the different fans included in this review, we have used 2 sort of metrics:
1. Airflow, measuring that of the different models, was as an open fan, then on a Hardware Lab SR1 radiator [Hasdrubal's comment: 120.1 for 120 fans, 140.1 for 140 fans]. We used aTSI 5725 anemometer, placed at the end of 1m long/20 cm diameter PVC tube. Foams are rigged within a foam block at the other end of the tube.
Data on radiator tests are gathered using the same protocol, the fan being affixed to the rad and encased at the exit of the PVC tube. In order to minimize airleaks between the fan and radiator, we used a silicon gasket between the radiator and fan.
To ensure data consistency, airflow results are the average of 5 successive 5 seconds measurements. Fans were started 10 minutes before data was gathered.
2. Noise levels are measured using a Cirrus Optimus CR152A Class 2 sonometer. It can detect sound levels as low as 20 dB(a), at sound level that can be achieved in the room used for the benchmark.
Fans are encased within a foam block in order to minimize vibrations, and the sonometer is placed 25 cm away from the fan in order to avoid impact from the air turbulance generated bu the fan. Setting a small distance was deliberate in order to best differentiate fans throughout the benchmark.
Data on radiator tests are gathered using the same protocol, the fan being affixed to the radiator, then encased within a foam block. In order to minimize potential vibrations, we used, whenever possible (depending on the form factor of the fan frame), a slilicon gasket between the fan and the radiator. Whenever not possible, silicon spacers where used to the same effect.
A noise level in the 20 to 25 dB(A) range can be deemed silent when used within a case, but they effectively because inaudible on ly when reaching the bottom of the range. Up the 28 dB(A), cooling is very quiet, and will be qualified quiet between 28 an 31 dB(A). A dB(A) level up to 37 is considered standard for a computer, beyond cooling is noisy. Beyond 45 dB(A) the sound level becomes very noisy, and can bceome difficult to cope with unless drowned by music, altough this threshold is, as the other ones, very subjective and wil be impacted by several factors such as the auditor's sensitivity to noise, as well as the smoothness and regularity of the noise or the environment in which the computer is operated.
Fans have been fed from 2.5 to 12 V, with 0.5 V increments, using a ELC AL942 stabilised PSU, that allows for a continuous tension calibration, and delivers up to 2A, which hopefully was never required!
RPM measurements as well as PWM speeds have been gathered using a Zalman ZM-MFC3 fan controller, fed froma passive PSU in order not to add up unwanted noise.
We still gathered data using PWM fans using 3 different set speeds: 600, 900, 1200 RPM, in oredre to get an eaiser comparison, even though these 3 speeds couldn't be reached by all fans of this review when in PWM mode.
FYI, hardware.fr can be deemed one of the reference french-speaking computer fans. Their reviews and test articles aren't the most numerous ones, but certainly among the most thorough and documented ones.