Details of the Noctua NF-A14x25r G2 PWM fan + minitest

Noctua NF-A14x25r G2 PWM

Close up of the new Noctua NF-A14x25r G2 PWM fan and its first tests. However, consider these as material for a “rough idea”, since as you already know, the pilot 140 mm LCP fans have atypical mounting holes for their format, so their mounting to the wind tunnel was necessarily non-standard. However, a few things can be concluded from these partial measurements as well.

The new 140-millimeter Noctua NF-A14x25r G2 PWM fan is aimed mainly at CPU coolers, primarily for cooling the new Noctua NH-D15 G2 (LBC/HBC) CPU coolers. It can also be suitable as a replacement for older, less efficient fans, for example on the NH-D15 G1’s heatsink (instead of NF-A15 PWM fans), but also other non-Noctua fans whose clips support 105 × 105 mm spacing for 140 mm fans (that is the standard spacing for 120 mm fans). This is the main reason why they are virtually incompatible with liquid cooler radiators and also with many cases.

Those cases that support 140 mm fans may not have the required mounting hole spacing for the Noctua NF-A14x25r G2 PWM and again for models that do, the impeller of these fans will be too large and even if it did fit, some part of the case bracket structure may interfere with it, which can be expected to have a negative effect on the acoustic properties. On the contrary, Noctua NF-A14x25r G2 PWM fans are suitable for the heatsinks of tower coolers, including the older Noctua NH-D15 (G1) cooler. These fans are compatible with the latter in terms of mounting. However, the sound profile may not be as pleasant as on the new heatsink, as psychoacoustic optimizations of the new fans on the old heatsink were naturally not Noctua’s focus.

And now on to the design details, which are in many ways the same as they will be later for the square-framed fans (Noctua NF-A14x25 G2 PWM).

Models with a circular frame (Noctua NF-A14x25r G2 PWM), which have already been released, are also not suitable for radiators due to leaky corners. Even if you manage to install them in a makeshift way, due to the leaks between the fan frame and the radiator frame, the air pressure/flow will drop and thus naturally the cooling performance will also drop. The latter probably won’t be downright bad, but expectations of peak performance may quickly dissipate. We saw what the difference between tight and leaky corners in a radiator application can be with the BeQuiet Silent Wings Pro 4 fans, which come with different types of corners. And a similar design to the one on the Noctua NF-A14x25r G2 PWM, with the highest leakage rate, delivers the worst results on radiators, naturally.

The NF-A14x25r G2 PWM fans, like all “second generation” LCP models, have humped blades (Progressive-Bend impeller) to achieve a higher pressure differential at points close to the impeller hub.

To achieve the best possible results, it is then articulated, with protrusions up to 0.5 mm in height. These are to bring the air streams in front of the impeller blades to achieve higher airflow. Only cosmetically, of course. Centrifugal Turbulator, as Noctua calls this innovation, is one of the new micro-optimisations.

The “old” micro-optimisations remain, although for example ISM (Inner Surface Microstructures) have a slightly different shape and depth again. The shape of this structure on the inside of the fan frame is based on the size of the gap that is between the blades and the tunnel. The smaller it is, the shallower the ISM is as well, since such a large microturbulent layer is not required to achieve optimum results.

In the previous generation NF-A14 PWM fans, the ISM is quite deep due to the several times larger gap between the frame and the blade tips. In the case of the NF-A14x25r G2 PWM, this gap is only around 0.7 mm.

While making this gap very small is beneficial for achieving the highest possible static pressure (and therefore airflow through an obstacle), it has also caused the most recent delay in the release of these fans. We also have a separate article on this topic – Noctua on NF-A14x25 G2 frame deformations (interview).

The “old” aerodynamic micro-optimisations (Stepped Inlet Design or Flow Acceleration Channels) have been joined by
winglets to reduce unwanted microturbulence at the blade tips.

In regards to the winglets, Noctua talks about a very good synergy with its AAO (Advanced Acoustic Optimization) frames and the overall design of the Progressive-Bend impeller. The latter, by the way, in this case has 9 blades.

The gaps between the individual blades appear to be quite large, but for high flow even through obstacles, there must be a combination of several elements. This will be dealt with, however, in the detailed analysis of the square-framed fans. For now, just the basics, which include information on the amount of static pressure (up to 2,56 mm H2O) and airflow (up to 155,6 m3/h). The values given in brackets refer to the maximum speed, which is specified and achieved at 100% PWM duty cycle or at linear (DC) 12 V, which is about 1500 rpm. With the slowing adapter to approx. 1250 rpm, the static pressure is supposed to be 1,3 mm H2O and the airflow then 127,1 m3/h.

And a short test at the end

Note: Aside from some mounting differences noted in the text below, measurements were taken according to our standard fan test methodology.

Although the NF-A14x25r G2 PWM fan is incompatible with our wind tunnel mounts, we still tried a “makeshift” solution with atypical mounting. Namely, using a hot glue gun.

The above mounting procedure is of course completely outside of our strict criteria, which includes constant mounting pressure, but it’s at least “something” until it’s time to fully test the square frame models (NF-A14x25 G2 PWM) with the standard spacing for 140mm fans in September. In addition, the measurements of the NF-A14x25r G2 PWM in the tests below are based on only one sample (instead of three samples, as is standard), so take them with a grain of salt.

For our demonstration, we selected the “36 dBA” mode without an obstacle, which corresponds to a speed of about 810 rpm at zero static pressure. This is the third quietest and also the fourth noisiest setting in which we test fans based on standardized noise levels. Downwards, but also upwards, the relative ranking may be different, or the ratios between the compared fans may be greater. The ranking of the NF-A14x25r G2 PWM fan at the very top may still mean nothing for your particular situation. Especially in a mode where the fan faces no obstacle.

The NF-A14x25r G2 PWM’s relatively lower static pressure also doesn’t indicate that the airflow through obstacles, such as radiators, needs to be lower compared to competing fans with higher static pressures (think Thermaltake Toughfan 14 Pro…). The important thing to remember here is that the static pressure is measured at zero airflow, i.e. at a significantly higher resistance than in practice on a radiator. And then the acoustic response also plays a role on obstacles. So no jumping to conclusions yet, we have to wait for a complete all-round analysis of the square-framed model, the NF-A14x25 G2 PWM.

Please note: “Higher is better” (“static pressure results in graphs”) only applies to comparing fans of the same formats.
What the quantity “static pressure” means and how to understand it is discussed in this article.

English translation and edit by Jozef Dudáš

Flattr this!

Thermaltake: Rotate a fan? Nah, replace the rotor!

Just when it seems that the pointlessness of some things must have gone through the roof, along comes SWAFAN 12/14 RGB. We can’t put it any other way, Thermaltake is taking advantage of the fact that casual users don’t know how a fan works. After all, paying for two rotors is a better “bargain” than having to orient one as you need it. At least if they didn’t have the same performance characteristics… That’s not how elite companies are built. Read more “Thermaltake: Rotate a fan? Nah, replace the rotor!” »


Comments (19) Add comment

  1. Does Noctua provide any numerical information on how much impact these “micro-designs” on the blades and frames have on performance and noise?

    1. Now we have the answer, I’ll try to paraphrase. Noctua can’t comment on which micro-optimization improves what and by how much. This is because it depends on the specific circumstances under which the fan is operated and this includes the nature of the operating environment. Another thing is that individual adjustments cannot be evaluated in isolation – all of the micro-optimisations are designed to work together.

      1. We have Noctua’s statement, which I will try to rewrite in my own words, but as accurately as possible.

        The NF-A14x25(r) G2 fans use single phase torque motors, but with the new NE-FD6 controller. Noctua did not see any technical advantages when switching to a three-phase torque motor. In combination with the NE-FD6 controller, even at very low speeds. On the contrary, in the final form (i.e. with a “single-phase” motor) higher power efficiency and even smoother commutation performance are to be achieved.

        Three-phase motors may reappear in the future in iPPC models, where they make more sense from a technical point of view. On a final note, these are all Noctua’s claims, which are not based on our knowledge, but on their internal tests.

    1. We should soon have a technical statement from Noctua on this topic, which we will post as a reply to the original post. We can certainly also speculate on the practical benefits of three-phase torque over single-phase torque, but Noctua’s statement will be better. It is true that single-phase torque is not always used in this class and there will be some explanation as to why this is the case with Noctua fans, which I don’t want to speculate on. 🙂

  2. NF-A14X25 G2 style impeller is not something new, but still it is the first time to introduce the design into computer cooling fans. Looking forward to future results!

    1. I would say that Endorfy Fluctus/Stratus are kind of similar, with the root of the blades sweeping slightly backwards and gradually bend forwards to a Gently-typhoon like outer region. Obviously the two are quite different in the exact implementation.

      1. “with the root of the blades sweeping slightly backwards and gradually bend forwards to a Gently-typhoon like outer region.”
        I have not think about Endorfy’s fans, but that’s the point! The leading/trailing edge starts from the hub while those are slightly sweeped towards to the opposite direction of rotation and bend progressively. This design concept can be found from axial fans in some automotive applications. For example, N003243-104228 and 252311R390 for replacement parts of Hyundai automobiles can be found.

        Also, there are some researches about axial fans in automobiles as below. You don’t need to read the full text, but check the appearance of the axial fan which is the research subject of a paper.
        S. Wang, X. Yu, L. Shen, A. Yang, E. Chen, J. Fieldhouse, D. Barton, and S. Kosarieh, “Noise reduction of automobile
        cooling fan based on bio-inspired design,” Proc. of the Institution of Mech. Engineers, Part D: J. of Automobile Eng., Vol. 235, No. 2-3, pp. 465-478, Oct. 2021, doi: 10.1177/0954407020959892. [Online]. Available:
        JO. Mo and JH. Choi, “Numerical Investigation of Unsteady Flow and Aerodynamic Noise Characteristics of an Automotive Axial Cooling Fan,” Applied Sci., Vol. 10, No. 16: 5432, Aug. 2020, doi: 10.3390/app10165432. [Online]. Available:

        Those fans usually have far larger diameter of impeller than computer fans. I think that the design concept (to make the aerodynamic load of the blade uniform throughout impeller surface) coincides with NF-A14X25 G2, resulting in a similar LE/TE curvature style.

    2. A similar impeller design is also found on the coolers of the newer Sapphire graphics cards, for example. It is not exactly the same, the hump of Noctua fan blades is less sharp, rounded, but these modifications are supposed to lead the fans to the same goal.

      We registered humped blades in graphics cards for the first time in our tests with the RX 6650 XT Nitro+ model, which has been around for more than two years… and they are also in the lower ranges, the RX 7600 XT Pulse cooler has them too.

      PS: Are you also preparing tests of NF-A14x25r G2 PWM or will you wait until September (NF-A14x25 G2 PWM)?

      1. Looking at fans of Sapphire graphics cards, the leading/trailing edge of blade is sweeped towards to the direction of rotation unlike NF-A14X25 G2 and automobile fans, so of course they have different design. But, we may be able to assume the similar engineering approach.

        I have almost stopped my fan tests since March due to several reasons. I may start the test after the (regional) release of NF-A14X25. (September or October) Until then, I’m going to slowly prepare the some enhancements of my setups.

        1. This, pretty much, the few other fans that have bends that look similar, might not actually have any benefit, or not have as much benefit as Noctua’s,

          Similar to how many, many fans over the last 20-25 years had Squama geometry, or as Noctua calls it Inner Surface MIcrostructures, and for most of the fans it doesn’t do anything, or matches or does somewhat worse compared to Noctua. Silverstone’s new Shark Force finish/geometry is the first in many years, that seems to have good benefits, and be quiet!’s was always good as well.

          It’s not just enough to simulate, problematic frequency peaks don’t show in simulation, and sometimes companies might put it just for looks

          1. I agree that Noctua has the top expertise on designing and fine-tuning small, low speed and low static pressure axial fans, but I suspect that they are actually good at others, such as fans with low speed and large diameter, for example. They have quite limited experience on fans with other characteristics. Combining with their small manpower of development, this is obvious shortcomings of the company. I don’t want to over-evaluate them too much.

            The discrete frequency can be predicted by analyzing pressure fluctuation on a blade surface. This sometimes does not match with test results of a prototype in real-world, but still CFD is very useful tool to look impeller closely.

            Company like Noctua shall utilize CFD much more due to expensive cost to make prototypes and prepare test setups and their limited development power.

            1. I agree Noctua doesn’t have as tuned high speed, nor big fans. What they do seem to have is top-notch methodology, guidelines and workflow to achieve that.

              I don’t wanna to over-evaluate Noctua, but my ears keep disagreeing with me xD, to a point, I still like the pitch of some other fans more than Noctua at medium speeds.

              That’s strange, Noctua specifically said CFD hits limitations for fine-tuning frequency issues, especially when impedance/obstructions come into play, and that’s not the first time I heard that (from the computed interviews), I heard similar things elsewhere.

              Now I assumed that applies to various microstructures on the surface of blade and frame, but they didn’t explicitly say that.

              1. — „I still like the pitch of some other fans more than Noctua at medium speeds.“

                Also compared to the NF-A12x25 fan? I certainly don’t want to say that it can’t be. On the contrary, it can. But if Noctua fans achieve a worse or less pleasant acoustic profile compared to competing solutions, it will probably be mainly in models based on older aerodynamic designs than the NF-A12x25. Of course, even compared to this fan (NF-A12x25) there are many models that can be considered acoustically “better” optimized. But this will be based on subjective criteria and which sound frequencies you are sensitive to. And, well, that’s a bit different for everyone.

                1. Yes, compared to NF-A12x25, I was a bit disappointed I could notice them >850RPM. From fans I heard in person, I liked how Cooler Master Silencio FP sounded more, though the performance wasn’t there for the Silencio +they didn’t last long. I’m waiting until I can get my hands on a the new revision P12 MAX with FDB/Rifle bearing to compare.

                  If I count what I heard over recordings in the internet, I’m guessing it’s possible i’d like Phanteks T30-120, or be quiet! Silent Wings 4 PRO more, but can’t say for sure.

                  But on the hand NF-A12x25 is really good when close to a case panel, it’s hard to beat it for turbulence noise, when in small cases xD

          2. You are right, none of these micro-optimisations will necessarily lead to the desired outcome unless they are implemented “correctly”. Even Noctua talks about the need for mutual alignment, only then does any… synergy work, let’s say.

            And when you mention ISM, it has to respond appropriately to the impeller. If you notice, ISM, in terms of shape and depth, differs from fan to fan even within Noctua models. If one does not approach these elements well enough, with a clear focus, then, naturally, they can probably also lead to a deterioration of the performance or acoustic properties of the fan.

        2. Of course, the aerodynamic design of the fan blades on Sapphire coolers is different than the NF-A14x25(r) G2… I was just talking about the effort to load the blades more evenly by bending them with a view to achieving better performance characteristics. And this has been done before in the context of PC fans here, on Sapphire graphics cards and probably before that somewhere else, on other PC fans.

          Thanks for the update and fingers crossed that the fan test methodology innovation will work out in the best possible way! 🙂

          1. Yeah, from just a blade loading standpoint and reducing the deadzone effect, any of those other kinked blades, might do really good, I’m just a bit skeptical on how the overall noise profile will turn out in the end xD

            And how much CFD time, and frequency analysis time on physical printed prototypes those other companies do, well did, and how soon did they decide, oh well, good enough.

            And yes, most of the increased time Noctua spends is for extreme longevity, and that’s not as relevant, for just noise performance, or raw dBA performance

Leave a Reply

Your email address will not be published. Required fields are marked *