Testing was done with the fan mounted one at a time on a single Swiftech MCR120QP radiator, with the fans controlled using a dedicated fan controller (Aquacomputer Aquaero 6 XT) in PWM or power (DC voltage) mode depending on the fan used. The controller also enabled RPM readout. Linear airflow was measured using an Extech 45158 Thermo-Anemometer 6″ away from the fan such that it measured the airflow in feet per minute through the radiator. Fan noise was measured in an anechoic chamber of size 5′ x 8′ with ambient noise level ~19 dBA and a sound probe held 6″ away to measure the sound volume in dBA accordingly. The fans were individually tested and the results below are averaged with a standard deviation calculated.
First up, I wanted to map out the RPM response curves to see if there was a significant difference between the fans controlled via PWM vs DC voltage control before measuring noise and airflow through the radiator.
Unfortunately, there is a bigger difference than I would have liked to see. The PWM version especially is really weird and does not match the RPM-PWM curve that be quiet! has on the product page as well. Both fans did the same as seen, with minimal sample to sample variation. At 100%/12 V, all 4 fans measured in at 1425 RPM which is very close to the rated 1450 RPM number. The PWM versions, however, drop to 1117 RPM at 99% PWM duty cycle. This is a near 22% drop in fan speed out of a single percentage point, and not really excusable. There is a second big drop, although the slope is not as steep here, going from 99 to 95% where the average fan speed was 902 RPM. It is really only from 90% and below that the RPM response gets mostly linear with a consistent slope, and the fans hit an average of 201 RPM at 50% PWM duty cycle before actually shutting down. I am used to seeing PWM devices maintain a certain minimum speed, so this was different. The fans remained off till I increased the PWM duty cycle back to 50% wherein it went back to ~200 RPM again. I have contacted be quiet! about this and will update this section if I hear back from them.
(Update: September 6. 2016- This looks to be a bug with the PWM controller and 6-pole motors, be quiet! was able to recreate this and are working on a solution. It could be limited to the Aquacomputer Aquaero only, although my Asus x99 Rampage V Extreme displayed a similar behavior. More on this as I get more information.)
(Update 2: September 14, 2016- Aquacomputer is claiming and I quote “This has nothing to do with the 6-pole thing. The problem with the Silent Wings 3 series is that the controlling chip that they use has an internal pull-up logic with 200K which is quite high. This results in a pretty low charging current of only 1.5µA (max) towards the aquaero (or any other fan controller). Since the aquaero has no internal pull-up (which is correct), the flank for the PWM signal ramps up slowly resulting in a vague waveform with triangles instead of rectangles or something close to it. We have informed Be Quiet about this with further details.” There appears to be something to this claim, as Corsair Link Commander Mini and the PWM headers on the Gigabyte x99 Phoenix SLI motherboard showed a similar fan control issue whereas those on the Asus x99 Rampage V Extreme did not. It seems to be a mixed bag at this point so until it is cleared up my stance on the PWM versions remains unchanged.)
Voltage control was more as I expected. Max RPM was 1425 at 12 V and both fans went down to 421 RPM at 4.3 V before shutting down. In fact, the fans shut down at 4.2 V and turned back on at 4.3 V so the claim of these fans having a low re-starting voltage is true. The RPM response is also very linear and only drop down at the last data point. There was also next to no sample to sample variability here (< 1%) which was nice to see.
So while you do get a lower min speed with PWM control, I would still argue you will get more usable fan speed control with the DC fans under voltage control. This being a 1450 RPM fan, you would like to get the 1000-1450 RPM range to use as a radiator fan. As such, I chose the DC version for noise and airflow measurements as well as fan comparison.
With the individual fan measurements done, let’s see how it compared with other 120 mm fans rated in the 1201-1500 RPM range. Please note that I am exhausting the capacity of the software used to plot the results here, and henceforth I will probably have to change to bar graphs at certain % of PWM/DC control instead. If you have any suggestions that will help keep the full data available without it looking very cluttered, please let me know. Thanks again!
The be quiet! Silent Wings 3 remains noise optimized, although they have definitely improved performance here over the Silent Wings 2. In fact, it is slightly louder than the Silent Wings 2 it replaces but mostly because it pushes more air through the radiator and has airflow noise added to it. The performance increase is worth it, as it comes in among the best and trading places with the performance optimized EK Vardar fans. So if you want good airflow/noise but give more priority to noise then these fans will do great. If you want good airflow/noise but give more priority to performance, then consider the EK Vardar. A win-win either way, and be quiet! should be proud of these fans.
To better see how much, if at all, they improved over the previous generation let’s compare just those two:
This is what I was referring to above. Despite the Silent Wings 3 being louder at the fan speed compared to the Silent Wings 2 (both 120 mm fans), the emphasis on performance increase comes in so handy that the Silent Wings 3 performs better almost every where compared to the older Silent Wings 2. So unless you are one that has to have the least audible fan, I would consider these a good step up over the previous generation.