The range of 140mm fans is now expanded by the Noctua NF-A14x25 G2 PWM. The wait for this model was very long, but finally, after various hardships, everything was brought to a successful end. This, by the way, is also indicated by our in-depth tests, from which the NF-A14x25 G2 fan takes away the “best” results across many usage scenarios. This is also due to the significant innovations in the aerodynamic design.
Basis of the methodology, the wind tunnel
Before you start reading the methodology with all the details, take a look at the test tunnel as a whole. This is the heart of the whole system, to which other arteries are connected (manometer, vibrometer, powermeter, …). The only solid part of the tunnel from the measuring instruments is the anemometer.
The shape of the wind tunnel is inspired by the Venturi tube, which has long been used to measure the flow of liquids and gasses. The Venturi effect for wind speed measuring is also known from the aerospace industry. However, the design for measuring computer fans has its own specificities, which this proposal of ours reflects.
The individual parameters of the HWC wind tunnel for fan tests are the result of physical simulations and practical debugging. All the details (folds, material or finish used) have a rationale behind them and are designed this way for a specific reason. We will discuss the individual design details in turn in the description of the sub-variable measurements.
Now we will briefly elaborate on some things that do not fit thematically into the text of the following chapters. Namely, for example, that the skeleton of the wind tunnel is the work of a 3D printer (PLA). The rough print was, of course, then thoroughly machined by grinding, fusing, polishing and varnishing. Especially important is the smooth finish of the interior walls.
When joining the individual parts, the emphasis was on making sure that they fit together flawlessly, that they were sealed flawlessly (we will come back to this when we describe the test procedures for pressure measurement), but also that the joints were not loosened by use. Everything is disassemblable for servicing purposes, but it is ensured that the properties are maintained during use and, for example, even under the stress of vibration. The threads are secured with either lock nuts or thread-locking fluid. It depends on which is more suitable in which place.
When the wind tunnel is not in use, it is enclosed in a dust-tight chamber. In addition to the technical equipment and its correct storage, it is also important for objective outputs that all measuring instruments are calibrated according to the standard. Without this, it would be impossible to stand behind your results and rely on the manufacturers’ specifications. Calibration protocols are therefore an important part of the methodology. Testing is carried out at an ambient air temperature of 21–21.3 °C, humidity is approximately 45 % (± 2 %).
Fans come to us for testing in at least two pieces of the same model. If the deviations of any of the measured values are greater than 5 %, we also work with a third or fourth sample and the average value is formed by the results of the fans that came out the most similar and the differences between them fit under 5 %.
Wow really a launch day review, is this a first for Noctua? 🙂
It will be the first time, yes. 🙂
Minor errors:
LS-PWM version is up to 800 RPM, not 1200.
Value’s missing in “Noise level – No obstacle – max speed” chart.
Question:
I remember Noctua saying that the protrusions on the gasket can compress enough to form a tight seal (based on my question)… guess not? Or is there not enough mounting pressure?
Thanks for the heads up, it’s corrected! 🙂
Yes, definitely, at different mounting pressures these protrusions in the anti-vibration gasket will compress differently, and in some cases they are compressed completely. As is the case in our situation on the thicker radiator. But there is still a gap on it. Caused by the way the walls of the radiator are rounded and the gasket is not perfectly straight and in places it twists a bit (you can see it well in this photo). It’s still just a soft liner, so it’s a natural thing. Don’t expect dramatic differences at this level though – the airflow through the radiator would be comparable even with an older system without the gasket material around the fan walls. Hmm… it’s maybe a good idea for a specific test looking at the dependence of the airflow on the presence of the gasket. Well, we’ll see…
I’ll test and see how well the gasket compresses once mine arrives. I’m guessing it might need pretty high mounting force to compress completely.
While the imperfect seal might only change the airflow by a few m3h, it seems enough to knock the fan from #1 to #2 at least in some cases. Case in point: no obstacle @ 36 dBA, Round = 810 RPM, 73.9 m3h; Square = 825 RPM, 72.1 m3h. The very slight difference in RPM should be due to the more substantial frame attenuating noise slightly more; meanwhile, the deterioration in airflow despite having higher RPM could only be attributed to increased leakage from the sides.
We will be glad if you share the results after your own tests. 🙂
I don’t have the proper equipment to test the exact force needed, but it seems to require pretty high mounting force indeed. I couldn’t compress it at all with my bare fingers, and in order to compress it on an acrylic panel, I needed unconformably high force (and still couldn’t get the sides to fully seal). For my application, foregoing antivirbation pads and gaskets altogether resulted in a much better seal. Fortunately, I have the intake side against the panel, and according to the manual that side does not need load relief.
Thanks for the info. The degree of deformation of these pads (and the size of the gap between the fan and the radiator) will be strongly dependent on the mounting pressure of the screws, which can vary even when fully tightened.
Such an achievement! Don’t forget, fellows.
Here is a selected comparison to the P14, dust filter and hexa grille.
Nice comparison with the fact that with your frame even better results are achieved. The offset from the obstacle is obviously useful. Microturbulence is suppressed, which reduces noise at the same time as the pressure/airflow is increased. Great, good work. 🙂
In a corsair 5000D where a 7800x3d is cooled by a noctua dh-15, which set of fans or fan configuration do you recommend me for optimal slightly positive pressure? Should i go for 2 front 1400mm, 1 rear 120mm and rear top 120mm? Thanks
One of the good options are Noctua NF-A14x25 G2 PWM fans. Then there are several alternatives that have a suitable aerodynamic design, but you need to know about the things that could possibly bother you. With the BeQuiet Silent Wings Pro 4 (BL099) it could be a whiny coil in PWM control, with the Arctic P14 Max it could be the minimum speed that’s too high, and with the Thermaltake Toughfan 14 Pro it could be the lower build quality (if you come across a clicking piece or multiple such pieces). Of the cheaper fans, the BeQuiet Pure Wings 3 (BL108) is still attractive, but it also has some weak points – check them out in our analysis and consider what is (in)essential for you.
Why did it fail so bad against a plastic filter compared to the previous gen? Are they not good for case intakes?
It is hard to say why there has been such a deterioration. The aerodynamic design of the two fans is significantly different. Some improvement could be achieved with a spacer, as indicated by the results of the NF-A14x25r G2 fan with the DIY frame, but in that case below average results are achieved. In cases with fans more significantly offset from the filters or with a thicker front panel structure, the Noctua NF-A14x25 G2 might already be suitable/top-notch, as can be seen in the results with the hexagonal grille. These are already significantly better than with the first generation fans (NF-A14 PWM).
Thanks for the test ! Time to try those on the North XL. <3
Planning to put 3 on the front and limit them at 1.000 rpm, hopefully they won't resonate with the front wood 😀
The vibration of the NF-A14x25 G2 PWM that reaches the frame is negligible at 1000 rpm. They will almost certainly not cause secondary noise on the front panel of the FD North XL. How the fans will interact with the structure of the brackets or the wood panel is, of course, another matter. Due to these elements, the acoustic profile can change even at lower speeds. Be sure to let us know how this combination of the case (FD North XL) and the fans (NF-A14x25 G2 PWM) behaves. 🙂
I was hoping to replace my 140mm radiator fans with these but, having seen your results, I don’t know that I would benefit from doing so. I have been using the Toughfan ex14 Pro and it seems to beat the NF-A14x25 G2 pretty handily in most of the radiator airflow tests. Am I missing something?
As far as cooling efficiency on radiators is concerned, it will be very similar for both fans. Nevertheless, it must be stated that the Toughfan Pro EX14 should not be lagging behind in this regard and rather will be the one that is more efficient (i.e. achieving a hair higher cooling performance at comparable noise levels). However, the premise that you don’t get to clicky pieces with the Toughfan Pro EX14 has to be met. This is something you probably won’t have to deal with with Noctua NF-A14x25 G2 fans, as the manufacturing quality requirements are higher.
I’m confused about those static pressure and airflow measurements. In your charts, the Arctic P14 (which is much cheaper) seems to beat the NF-A14x25 G2 at matched sound level. It produces higher static pressure against resistance, and more or comparable airflow.
So it seems that this premium fan isn’t that great in terms of static pressure, airflow or noise. Am I missing something?
Thank you for your question, I will be happy to explain.
It is not true that higher measured values of static pressure at the same airflow automatically mean a smaller loss of airflow through an obstacle. This may or may not be the case. One reason for this is that static pressure is measured at zero airflow, which represents significantly more resistance (and back pressure) than any obstacle in practice. The resistance of a radiator is also significantly smaller than the situation at which static pressure is measured. At zero airflow, the fan behaves differently than on an obstacle at non-zero airflow. At zero airflow, a different, practically unrealistic motor output is achieved, and so in the real world it is not possible to work responsibly with this value. In other words, a fan that has a lower static pressure at the same airflow (these values are in turn usually based on a zero static pressure environment, which is also out of common practice…) may not have a greater loss of airflow through an obstacle. Again: It can, it doesn’t have to. Meanwhile, how a particular fan interacts with a chosen obstacle also plays a role in tests normalized by noise level. Specifically, to what extent does a given obstacle increase the noise level of a particular fan. This can also vary from case to case. It depends on the combination of the particular aerodynamic design of the fan with the particular structure of the obstacle. While in some combinations there are no significant tonal peaks (compared to a situation without an obstacle), these are dramatic in some/other cases. It is individual. Anyway, to evaluate responsibly the aerodynamic performance of a fan for practical use based on traditional measurements of static pressure and airflow, and hence actually P/Q curves, is impossible. This is because practical measurements on obstacles, which are important, are lacking. For what reasons you already know. Does that answer your questions? If anything is unclear, feel free to bring it up, we can discuss the specifics again. 🙂