Phanteks PH-F120T30 in detail
Does it or does it not clearly beat the Noctua NF-A12x25 PWM? The Phanteks T30 doesn’t like second places, and in the vast majority of cases it doesn’t get to them, outperforming competing fans in its format (120mm) by a good chunk when optimally setup. But that was a bit of a must to justify the extra thickness, which is associated with poorer compatibility after all.
The task for the Phanteks PH-F120T30 fan (hereafter referred to as the Phanteks T30 or just T30) was “to design the most efficient 120 mm fan”. That’s pretty clear, Phanteks doesn’t admit anything else even after two years since the T30 was released, and somehow over time users have adopted this as well. And this is no mistake.
The T30 has an undeniable advantage over the rest of the high-end, but commonly formatted 120mm fans in its 5mm greater thickness. Instead of 25 mm, the Phanteks T30 is 30 mm thick. It’s not just the frame that’s enlarged (or that the anti-vibration corners would protrude this way), but the rotor itself, a key element, is also deeper. Even though you’re in for an exhaustive analysis, you won’t learn from it how beneficial the extra 5 mm is, because it’s not just about them. The Phanteks T30 uses a unique rotor that is, compared to major competitors such as the Noctua NF-A12x25 (PWM), Silent Wings Pro 4, Lian Li UNI Fan P28 or Thermaltake Toughfan 12 Pro, is different mainly in that it has fewer blades, only seven. In this, the T30 is design-wise halfway to the Arctic P12. The latter has even fewer blades, by two.
One of the reasons a rotor with fewer blades is attractive is that it has fewer leading edges that cut through air and create noise. Of course, to avoid a drop in static pressure, the compensation in such designs must be in the larger blade width. These things have been obvious to Arctic for a long time with 5-blade designs, but Phanteks has technically taken it to the next level. As you well know, the biggest drawback of Arctic P12 fans is vibration, because of which, with the chosen design, the fan operation is accompanied by a high noise level at low, rumbling frequencies. Arctic countered these at a later stage by implementing a hoop (first in the P12 PWM PST A-RGB and then in the P12 Max), which compensates for the use of lower stiffness material and reinforces the tips of the blades. Phanteks solved this with the T30 by using LCP, a material that has significantly higher tensile strength than PBT (Arctic) at the same thickness, and lower thermal expansion as well.
In cosmetic detail, however, the T30 is definitely not a masterpiece like the Noctua NF-A12x25. For example, the leading edges are unintentionally jagged, also due to their high sharpness, where it’s harder to achieve a nice smooth edge than with “blunt” fans like the Sterrox Noctua. But it should be more or less nothing that would significantly degrade (especially with such a heavy rotor) the quality of the balance. To maximize accuracy, Phanteks has two rings of different radius inside the hub, in whose segments there is additional, balancing material. The architecture of this solution is captured in the available illustration of the hub cross-section.
The Phanteks T30 is thus a 7-blade fan with a design that is clearly predisposed to deliver higher airflow and static pressure at comparable speeds in addition to lower noise.
The large “active” area of the total cross-section is important for high airflow, which is also achieved because of the smaller number of gaps between the blades, and the hub surface area is also relatively smaller (compared to the Noctua NF-A12x25 PWM by 20.1%). This creates more space for the blades, which are only about 0.5 mm from the frame, or the stator tunnel. It should also be emphasized that the fan itself has a slightly oversized cross-section, so the rotor is really big for a 120 mm fan.
The aforementioned narrow gap between the blade tips and the frame plays a part in the higher static pressure, which is higher compared to other fans but mainly due to the greater thickness of the profile. The blades are really wide here, which also contributes significantly to the reduction of flexibility, which also reduces vibrations. These are then traditionally damped at the end, on the frame, by the softer rubber pads in the corners of the fan.
A 6-pole motor with three-phase torque standing on six coils and magnets is used. The advantage of this design is smoother operation, less force is required to overcome the individual poles. Thus, the minimum speed, vibration and power draw can eventually be lower (than with 4-pole motors). And for fans with a super-robust rotor, like the Phanteks T30, this can be really useful.
An important part of the fan’s features is a three-stage switch that limits the upper limit of PWM duty cycle. Otherwise, at 100%/12.00 V with the range ending at about 2984 rpm (in the “Advanced” position) the operation would be very noisy.
In factory settings, the T30 is set to “Performance” mode, where top speed doesn’t exceed 2000 rpm (we averaged 1993 rpm over three samples). Finally, in the lower “Hybrid” position, the maximum is about 1212 rpm. Oddly, in this mode, the minimum speed is the highest, at about 489 rpm. In the other two modes (Performance and Hybrid), the lower speed limit is at about 294 rpm.
Naturally, the greater fan thickness also means the need for longer screws for mounting on radiators of liquid coolers. One nice kit is also a standard accessory, even in a quantity of 8 pieces (so double-sided fan mounting is expected). However, be warned here that the screws have inch threads (UNC 6-32) and many radiators count with metric ones. These must then be purchased separately.
Beware of this, as it may simply happen that you will not be able to install the fans on your radiator. Sometimes 35mm bolts with M3 threads are supplied with a radiator, but this is not the rule. However, suitable screws (M3x35) can also be conveniently purchased separately, and a number of liquid loop accessory shops offer them.
The cable that is soldered to the fan is shorter (145 mm), but can be extended by the included adapter to the final 665 mm. The T30 has two connectors – one for its own connection and the other for daisy chaining with more fans. The individual wires are otherwise in a shared mesh with a dense weave, which has reliably fitted shrink tubes on the sides.
And one more thing: To navigate through the result graphs as easily as possible, you can sort the bars according to different criteria (via the button on the bottom left). By (non)presence of lighting, profile thickness, brand, bearings, price or value (with the option to change the sorting to descending or ascending). In the default settings, there is a preset “format” criterion that separates 120mm fans from 140mm fans.
- Contents
- Phanteks PH-F120T30 in detail
- Overview of manufacturer specifications
- Basis of the methodology, the wind tunnel
- Mounting and vibration measurement
- Initial warm-up and speed recording
- Base 6 equal noise levels…
- ... and sound color (frequency characteristic)
- Static pressure measurement…
- … and airflow
- Everything changes with obstacles
- How we measure power draw and motor power
- Measuring the intensity (and power draw) of lighting
- Results: Speed
- Results: Airlow w/o obstacles
- Results: Airflow through a nylon filter
- Results: Airflow through a plastic filter
- Results: Airflow through a hexagonal grille
- Results: Airflow through a thinner radiator
- Results: Airflow through a thicker radiator
- Results: Static pressure w/o obstacles
- Results: Static pressure through a nylon filter
- Results: Static pressure through a plastic filter
- Results: Static pressure through a hexagonal grille
- Results: Static pressure through a thinner radiator
- Results: Static pressure through a thicker radiator
- Results: Static pressure, efficiency by orientation
- Reality vs. specifications
- Results: Frequency response of sound w/o obstacles
- Results: Frequency response of sound with a dust filter
- Results: Frequency response of sound with a hexagonal grille
- Results: Frequency response of sound with a radiator
- Results: Vibration, in total (3D vector length)
- Results: Vibration, X-axis
- Results: Vibration, Y-axis
- Results: Vibration, Z-axis
- Results: Power draw (and motor power)
- Results: Cooling performance per watt, airflow
- Results: Cooling performance per watt, static pressure
- Airflow per euro
- Static pressure per euro
- Results: Lighting – LED luminance and power draw
- Results: LED to motor power draw ratio
- Evaluation
it seems at low noise levels A12x25 may be still unbeatable, both by thicker and “wider” fans, though we don’t have tests to confirm that yet
I wonder how much (or how little) improvement is still possible, but at some point the air flow itself becomes noisy so there’s an upper limit of that too
At extremely low noise levels, the Phanteks T30 will definitely have the upper hand. The lead in our quietest mode over the NF-A12x25 is very significant, and I don’t see too much reason why this trend would start to reverse (to the detriment of the T30) after a 150–250rpm reduction. The lower the speed, the smaller the tonal peaks. And even at starting speed via PWM control, the motor is extremely quiet, virtually noiseless.
What I promised you some time ago still holds true, namely that fans that have the potential to perform well at even lower than low speeds will be specifically addressed later at even lower noise levels. We just need to find a reliable way of getting those fans into such conditions. At the lower end of the range of measurement instruments, the percentage error of measurement tends to be the highest, and the last thing I want is for it to determine the order in the charts in designs that are evenly matched. So first we need to sort out these technical aspects so that everything is sufficiently accurate and fair.
depending on the specific use case at 31 placements of these two shift quite significantly
as the speed goes down and the air noise silences other sources surface and become more apparent, it’s entirely possible that the motor and bearing quality are on par with Noctua, but I hope you don’t mind I don’t trust personal and subjective reports anymore and want to rely only on measured numbers comparable to products I had a chance to test myself, too many times I was severely disappointed with “silent” and “inaudible” products
and I understand it’s much harder to get the proper ambient noise for these tests, to get and configure the right tooling, to exclude anomalies and that’s extra work and time most of readers will never appreciate
it would be amazing to have these up and running before the new 140mm from Noctua comes, but for the better or worse that’s probably a lot of time
Sure, but with that we are already talking about psychoacoustic optimizations. There it is important to properly design the end environment, or the obstacle with which the fan will work. Naturally, there are cases where, for example, with a generic plastic filter, the NF-A12x25 starts to catch up to the T30. With some structure that makes the Phanteks fan even more noisy, it will probably outperform it in the same noise level. It’s all about a properly chosen obstacle that will give the selected fan the least amount of tonal peaks.
If the best options are chosen for all solutions (or outright imperfect designs are eliminated, as plastic filters in cases undoubtedly are), the T30 will always have the edge to a similar degree as the results without obstacles suggest.
that’s true, testing things in void isn’t entirely precise, it’s pretty much like testing audio equipment: pairing two “best in class” can yield worse results than getting matching components because the former may be amplifying each others flaws
but that’s the best we can get and allows us to estimate results we’ll get
A12x25 is terrible as a pull fan unless you add a spacer, filters and grills interfere with the airflow and introduce noise but are required to keep the dust and fingers away, T30 is always thicker and may not fit many builds, the ATX standard itself and everything compatible with it make components cooling a real pain unless you aren’t using any expansion cards at all and on top of all that all consumer grade and most of professional grade components are heavily overclocked and overvolted out of the box resulting in severely degraded efficiency and much beefier and more expensive (sub) components being used, again interfering with the airflow for the sake of power you may not want to use, with 7950X you may reduce the PPT by 100W and add some CO to that to keep nearly stock performance with much less heat, but you still keep the giant VRM and heatsinks and you still pay for them
it’s a real mess
UMIK-2 with its lower noise floor when compared to UMIK-1 could be helpful in such measurements. One thing to look out for is that shortly after release there were big measurement differences in the bass region, supposedly it’s fixed but not sure if it requires user action.
Thanks for the tip. Besides UMIK-1 we tried other microphones on a technically higher level during the optimization of the methodology, but none of them were powered via USB, so their use would be quite inconvenient and mainly the shortcomings of the UMIK-1 do not interfere with the relevant frequency band in which the fan sound is.
We can certainly get below the REED R8080’s capabilities with the UMIK-1, although, as you write, the UMIK-2 has a lower noise floor. I didn’t know about big measurement differences in the bass region in the early phase of these mics, but ours is fine in this respect (scales with other, significantly better mics). I consider the most noticeable shortcoming to be the relatively higher (but still quite low) noise floor at frequencies from 5 kHz upwards. A kind of linear slope, where 20 kHz is already quite high. But those are all frequencies that the fan sound does not contain and can be ignored.
When I wrote about the fact that for the quietest fans we will implement a test environment in which noise normalization at extremely low speeds will be possible, I did not think about the use of a microphone for a moment. Here it has to stand on a proper noise meter to filter out electromagnetic noise.
It seems that on thicker radiators, the A12x25 wins not only slightly in airflow but also in static pressure through it. In this situation, given the same 30 mm thickness, A12x25+5 mm spacer will likely lead the T30 by a larger margin, at least when set as pull.
It will be titanic work but hope to see 28/30mm fans vs top 3-5 25mm fans with spacers test in future.
We’ll figure something out. I’ve had the concept of what you might like in my head for a long time. Well, it’s a matter of creating the space for realization. 🙂
So I’m keeping my fingers crossed then!
With such strong performance, it’s curious why there’s a lack of CPU air cooler that utilizes this fan. Is it due to a lack of focus on air coolers by Phanteks, or is it because of the extra thickness that the cooler does not perform as well as, say the PH-TC12DX, while maintaining the same compatibility?
Besides thing you have already said.
They abandoned air cooling long time ago and within two years from releasing T30 fan i often saw shortages and problems with availability or even semi-official note that T30 will be in shops after month or more. Also they are using MagLev bearing which is licensed from Sunon so lesser profit margin than other fans. Adding one to another i think answer is easy.
Maybe it’s a combination of both factors. On the one hand, Phanteks doesn’t innovate air coolers too much, and in my opinion, T30s don’t even fit them. In the context of dual-tower coolers there is the disadvantage of the smaller format, so it would be a relatively smaller heatsink not hitting the limits of the current cases and at the same depth as the high-end models with 140 mm fans, it would also be slimmer. This would result in a significantly smaller surface area of the fins and under such conditions the overall cooling performance would probably not be sufficient to compete with the most efficient coolers at comparable noise levels.
Of course, in the framework of a single-tower solution (as a successor to the PH-TC12DX) it would be a very nice competition for the NH-U12A, but here we are probably already in a very small market, which Phanteks does not care about. Who knows… I can ask Phanteks about it, maybe it’s completely different and some cooler with the T30 is already going through the last prototype phase. 🙂
One of the most important test has come..!
I first checked the test on radiators. I surprised that PH-F120T30 doesn’t have higher efficiency than NF-A12X25 on thicker radiator. Although the thicker radiator has less FPI(12 or 15?) than the thinner radiator(22), the thicker radiator is obviously more restrictive obstacle than the thinner one. So, I expected that PH-F120T30 will make larger airflow than NF-A12X25 especially for the thicker radiator. I wonder what the element made the difference of 2 fans’ relative position between 2 types of radiator.
When it comes to acoustics, I think PH-F120T30 is not good option for some low-noise enthusiast. Despite of its quiet motor noise and absence of resonance, the fan usually has ‘buzzing’ noise which occurs much less for NF-A12X25. I think if the computer is right next to a user (of course it is not common scenario), ‘buzzing’ noise can be annoying.
I have recently replaced a rear fan from TL-C14 to PH-F120T30. It really exhaust hot air as well as previous 140mm fan and doesn’t have any tonal peaks from resonance at specific RPMs. Very satisfied!
Buzzing: I took another good look at the spectrograms and it looks really good for T30 on all levels. Even the peak in the 350–400 Hz band is a bit lower than the NF-A12x25. Based on what I can see (I haven’t relied on my own hearing for a long time, haha) I can’t conclude that at such low speeds the T30 would be a noisier choice for the most demanding users noise-wise.
The thicker radiator is the Alphacool NexXxoS XT45 v2 with 15 FPI. It is often assumed that higher static pressure is important for a thicker radiator, but this may or may not be true. Of course, if the FPI and the rest of the design are the same and the only variable is the thickness, then yes. But if the radiator has less dense fins (with lower FPI), it does not put as much resistance to the fan and yet it loses more air. I imagine it like two test tunnels, where one is perfectly tight with perfectly smooth walls and the other has some flaws in these respects. These are always present in radiators, of course, but on the longer route they have a bigger impact on the pressure difference and the adversity of such a situation obviously does not suit the T30 very well.
Someone can probably also think of a worse acoustic optimization of the radiator profile, which would limit the speeds of selected fans when adjusting to the same noise levels (and thus they would naturally achieve a lower airflow), but this is probably not exactly the case here. The T30’s lower airflow through the thicker NexXxoS XT45 v2 is also achieved at the same speed when noise is ignored.
Maybe the noise I descirbed comes from the ‘flat’ band from 70Hz to 100Hz. PH-F120T30 has the intensity of -88~-87dBu, and NF-A12X25 has the intensity of -91~-88dBu, according to the spectrograms. Could it be that the ‘buzzing’ noise arises from this difference in intensity distribution? However, I also think that PH-F120T30 is not always worse solution for all low-noise users than NF-A12X25. The noise from that low frequency band can be discarded enough at a distance.
Regarding radiators, it seems that testing various types of radiator is needed to certainly derive the reason.
For instance, comparing the results from: EK-CoolStream Classic SE 120(22 FPI, 27T) and Alphacool NexXxoS ST30(15FPI, 30T) If PH-F120T30 has similar or slightly lower noise-airflow efficiency at the latter one, we can assume that the FPI may be one of the main elements. If not, the combination of a deep depth of louvered fins and a low FPI is just not optimized with the fan as you explained.
This is something more complex than I expected. Maybe I need to test with a radiator that has different parameter with previous one.
I overlayed the graphs for better visualization: https://imgur.com/a/RvnH0LK
The graphs used are T30 (blue)/A12x25 (red), no obstacles, 33 and 45 dBA.
Comparing the differences, at 33 dBA, the A12x25 is noticeably noisier at around 45-65 Hz, 150-180 Hz and slightly at 300-400 Hz. The T30 is noisier at 35-40 Hz, 70-150 Hz, 220-280 Hz, and slightly from 400 Hz and above.
At 45 dBA, the A12x25 is noticeably noisier at around 28-38 Hz, ~270 Hz. and slightly from 500 Hz upwards. Meanwhile, the T30 is a lot noisier in the 40-230 Hz region.
It does seem at higher noise levels, the A12x25 have a better noise profile. Probably not at 33 dBA, however.
There’s also the question whether dBA alone can accurately reflect the perception of noise. Could other sound weighing methods be useful?
Thank you for the very useful modified material. This will alow us to compare easily the difference of intensity.
How about comparing them with the ‘area difference’ of low frequency band?
Let’s define: (Area difference) = (Frequency area of PH-F120T30) – (Frequency area of NF-A12X25)
According to the spectrograms, it seems that the area of PH-F120T30 is larger than that of NF-A12X25 for 33dBA, under 200Hz. Note that the scale of a frequency is not constant. The area difference of the range 40~70Hz is actually not large as the area difference of the range 70~120Hz.
The above explanation is just my interpretation about the spectrograms. We need Lubomir’s opinion about this.
I admit that human’s ear is not always accurate, but I can see some reviews about the low frequency noise of PH-F120T30. I also hear the slightly emphasized low sound at a low rotational speed. Of course, I don’t think that this is a critical problem.
I have not used other weighing methods(i.e. dBB, dBC). So I’m not sure about them.
The visualization with the overlaid graphs looks good. I immediately thought that it could probably be made interactive, where only the fans that the user selects would be displayed. I don’t want to promise to implement it right away, but I’ve been thinking about it for a while. 🙂
If the noise intensity in dBu (or even dBA) is captured and interpreted at the level of a spectrogram with a resolution of 1/24 octave, there are probably not many more accurate methods for those who can read the data. I recognize that it can be difficult to make sense of it. In a forthcoming series of articles on this very subject, we will also confront spectrograms with audio recordings, which will make it obvious what the known sounds look like in the spectrogram’s columns. I’m still thinking how exactly to do this to make it as effective as possible. That is, not to make it “useless” work that most of the readers won’t digest. But by “other sound weighing methods” did you mean something specific with a justification as well? By the way, we have spectrograms in dBu.
In the framework of popularizing these topics around the sound of fans, I am of course open to possibilities that we have not considered yet. The key is of course that the measured data can be grasped by as many people as possible, but in such a way that there is no significant distortion at any level (as it happens in the case of sound recordings, which we have already discussed, I think :)).
I agree that the data can be presented in interactive graphs, and instead of histograms, perhaps they can be represented as lines.
On weighing methods, there’s ITU-R 468 for example: https://en.wikipedia.org/wiki/ITU-R_468_noise_weighting
I have heard of it before but have pretty much no knowledge of it, so I have no real idea whether it’s suitable for this application. Based on reading the description, it seems useful, however. Perhaps you can offer your opinion on it.
You surely know more about the ITU-R 468 than I do. It’s completely new to me, this is the first time I’m reading about it. So I won’t tell you my opinion, but I promise to look at it in my spare time and consider whether there really is not something more suitable for interpreting noise measurements. Personally, I find the “audiophile” interpretation of dBu in spectrograms quite faithful, but that doesn’t mean that there isn’t something else that people can relate to better. But then with lesser known standards and quantities there is always the risk that they will not be accepted because they don’t have their own “marketing” or tradition, I don’t know how to put it exactly, that people often look to things they know, things they’ve heard of. This is also why, for example, we use units of H2O to express static pressure measurements, even though there are many other, more reasonable options. But again, people know H2O from fan specs and trying some alternatives could be suicide. 🙂
You are right that there are differences in the 70–100 Hz band and in it the T30 is a noisier fan than the NF-A12x25. Especially at 80 Hz, where it is at -4,2 dBu. For a better overview I have created a comparison table for these frequencies:
level that one would have to have extremely good hearing to be able to somehow distinguish between -86 dBu and -90 dBu, with the technique we use to measure the noise (i.e. short distance, amplifier…). Something else is the Arctic P12, which at 80 Hz goes up to -72.4 dBu in the same mode (33 dBA). Only a deaf person would not hear that. With the T30 I rather wonder if there are not more revisions, although we probably won’t find out easily. Anyway, our samples should be among the relatively newer ones, Phanteks sent them to us on 9/2022 and even before we had three pieces from a store in the testlab (8/2022), which didn’t behave any differently. The manufacture date will, of course, be older, and perhaps what you are looking at may be a newer or even possibly older revision/batch? Too bad Phanteks doesn’t have at least as much clarity on this as Arctic. But maybe the 6-digit number on the sticker on the bottom of the motor (and the higher the number, the newer the fan could be) could tell you something? But if the geometry and material composition of the fan is not different, I wouldn’t be able to explain the difference in noise at these frequencies very well, even with the Arctic P12. Why should rev. 4 not rumble/reach those high tonal peaks like rev. 1, when the shape of the blades is still the same and the material from which they are made has not changed. Or has it changed? I am still waiting for an e-mail from Arctic, in which all the differences of the different revisions of the P12 PWM PST should be listed.
It’s more complicated with the radiators. A simple division into thinner and thicker can probably be quite misleading, unless you know that there is also a difference in FPI. Nevertheless, even equal thickness and equal FPI do not guarantee that the pressure in the radiator will be lost in the same way (there are also differences, for example, in the level of tightness of the frame and so on). This is different from model to model and this is rather a question for radiator tests. No matter how hard we both try in fan tests to make all results relevant for everyone, we can only ever get close to that to some extent. Therefore, I consider it important not to draw clear conclusions about a more or less suitable fan for a given use case if the results are close to each other. In another, though a similar situation, the situation can quickly turn around and the statement from a particular test may no longer be valid. I have come to terms with this and therefore I do not really consider the minimal differences in the context of the evaluation. And now I remember your planned tests with different grilles. Have you made any progress here? 🙂
Perhaps, what I told about ‘buzzing’ usually would not be heard from 33dBA test. I would like to know how fast the PH-F120T30 operates under 33dBA test conditions. Maybe, is it much lower than I expected? (600~1000RPM)
My samples of PH-F120T30 have the number as below:
24268J, 24268W, 2426F2 – 02/2023
13623F – 12/2021
I guess that there are no considerable changes or revisions for PH-F120T30. The older one and the newer one have exactly same aerodynamic noise. (at least from what I hear)
Regarding Arctic’s revision, I think I need to check more samples by myself and other users’ reviews about the revisions, because it seems that the tonal peak is not completely resolved for every batches and environments. Arctic told me that the resonance ‘is optimized’ and ‘achieved improvement’ for the recent revision. But, they didn’t explain me how they improved P series without changing overall design.
I also sent the email to get information about the entire changes between each revisions, but Arctic said that they don’t want to open full-list of revisions to the public.
I understand your opinion about the test for radiators. I’ll just try thicker radiator myself in the future!
When it comes to the test for various grilles, I postponed it because I am recently testing 6 types of obstacle of standardized tests (radiator push, radiator pull, filter+grill intake, filter push, radiator pull+filter push and no obstacle) for various sizes of fans. It is really time-consuming work! I hope I can deal with grilles in this winter.
33dBA here corresponds to 977 RPM, you can check on page 13 (Results: Speed).
Oh, I didn’t check that. Thank you. I need to read the entire article more thoroghly.
The speed of the T30 in “33 dBA” mode is about 977 rpm, as noted before. I would say that what you are describing only occurs at higher speeds. For example, in our “39 dBA” mode (~1273 rpm) there are already several peaks across the entire fan sound band. If this is indeed the case, the claim that the T30 is suitable for demanding users in silent setups (which, of course, implies a significant reduction in speed) could be quite accurate and well repeatable across all samples. Here, of course, we are talking about speeds around 800 rpm (31 dBA mode), where there are practically no tonal peaks and at the same time the Phanteks fan benefits from its greatest strengths, i.e. the large outer cross-section and above-standard thickness.
Thanks for the info on the T30 numbering and also on the Arctic revisions. This is maybe the reason why we are not getting an answer. I wanted as much detail as possible in this regard to make more sense of the claim that the newer revisions have lower tonal peaks.
So I’m a bit confused by these charts and some of the other comments, so it seems that the T30 performs quite bad on thick radiators even with a matched pwm?.
What other fan would be the best, I recently bought the new Galahad II Performance which is 32MM thick and has a 19FPI, what would be a good fan for that then, if I want max static pressure and airflow through a restrictive thick radiator?
Regardless of the obstacle used, the T30 is always a top-notch fan, especially at lower speeds. What we’re addressing here is the relative position to the NF-12×25 PWM and the degree to which the T30 will justify the greater thickness and eventual worse compatibility.
You can’t go wrong with the aforementioned Noctua fan (NF-12×25 PWM), just like you can’t go wrong with the Phanteks T30, as long as you don’t have a VRM heatsink on the motherboard pushing you from the top that could interfere with the cooler. Both of these will be very evenly matched fans, performance-wise, in your conditions, at comparable noise levels.
Thanks for your response, I really appreciate your fan reviews I’ve been following them for a while, really nice.
And I’m also interested in fan spacer testing like another person mentioned, I’m personally planning on putting 30mm spacers with between the fans and AIO soon.
There will be tests with spacers. The question is not “IF” but “WHEN”. In all sincerity, I have to say that it is very difficult to find space for such more marginal topics, given the constant lack of finances on which the magazine depends. But it is on the list and you will see it all one day. 🙂
For some time now, we know from the results that current, the best 120 mm fans can beat the best 140 mm ones when put against high resistance, such as thicker radiators.
Now I wonder, will these 120 mm fans gain even more advantage when put on a 140 mm radiator with an adapter? For example, Noctua has the NA-SFMA1 adapter for using the A12x25 on 140/280 mm radiators. Since there is significantly more (~36%) area for air to pass through, it would make sense that the impedence caused by the radiator will be lower, further increasing airflow at equivalent noise levels. Could it be that the A12x25 with NA-SFMA1 actually beats all other current 140 mm fans on a 140 mm radiator, even if it’s a thinner radiator? I think it might be an interesting test to prepare for the next gen 14 cm Noctua fan.
Tests with adapters have been in the plan for a long time, but somehow I can’t get to them. Overall, the idea was that we would focus more on fan topics by this time, but the reality is different after all. So the plan for the end of the year has changed to “measure mainly the key 140 mm models that will be relevant in comparison to the 140 mm LCP Noctua fans”.
It’ll be worth testing eventually, maybe only a mini test on radiators, and I think it’s pretty relevant to the plan. In an interview during Computex this year, one of the stated design challenges of the next gen 14 cm fan is to make it outperform the A12x25 (with the 140 mm adapter on radiators), so it’ll be interesting to see whether the goal has actually been achieved in the end product.
We’ll have to figure it out somehow… I have had the NA-IS1 here for testing for a long time, but somehow I can’t get to it. :/
I don’t like mini-tests that take little time and I follow the motto “do it properly or not at all”.
Do you measure noise without obstacle and record the rpm at that noise level. Then perform obstacle tests with the recorded rpm?
Or do you noise normalize at each of obstacle tests? i.e. tune rpm to get wanted dB(A) reading with each obstacle.
p.s. I did not read whole article. Info is there most likely
Noise normalization is done independently for each obstacle, the details of which can be found in page 10.
Thank you for your question. The noise level is normalized for each obstacle separately. In other words: The noise level in dBA on obstacles is always the same, but the tonal peaks are of course always different (that is why the measurements are supplemented by a spectral sound analysis).
We only record rpm in a no-obstacle mode. I do not consider the collection of rpm data in other modes to be very important, just because in practice there are countless obstacles with different profile structure and each one forms tonal peaks differently. It is unlikely that you will have exactly the same grille or filter as the dust filter we use in our tests. These are model situations of how the fan will probably behave in similar conditions, but as far as exact speeds are concerned, according to our results you would not set the fan on your grille or filter appropriately, because the optimum speeds for them will be slightly different for a given fan. It is true that the diversity is smaller with radiators, but not recording the speed in other modes is also to save some time. In short, this parameter has already gone beyond what we can manage with our editorial capacities in terms of time.
In free space, without an obstacle, the speed measurement makes the most sense. This is because in this mode there is no longer any influence of secondary noise (obstacles) and the speeds we report will already correspond quite faithfully to your situation. That is, if you also use the fan in a free space where its sound will not be significantly affected by anything. Even under such conditions, there are many external influences (apart from the different pressures in the various cases, which also shape the final speed and ultimately the vibrations on the blades) that will cause it not to scale exactly, but here we are getting quite close to what you can reproduce quite faithfully.