Let's get the hardware discussions going
If you take the time after reading one of our articles and feel like leaving a comment in the discussion, we’d love to hear from you – you’ll be supporting us more than you might think. Of course, the bottom line is “having something to say” and the main goal is to increase the value of the articles through your comments. Feel free to oppose our claims or you can confirm them from your experience.
Given the above-average experience that the typical HWCooling reader possesses, we would like to appeal to you to express yourselves more, if possible, in the discussions below articles. By doing so, we can achieve the following:
- Even more complete information, where, for example in tests, you can confront our findings with other conclusions. Or fill in things that are missing in the articles. Although we try to look at things from as many perspectives as possible, we still don’t see “everything”. And it will be guaranteed to be beneficial if you deepen the level of knowledge from your own experience.
- Higher value of advertising units, allowing HWCooling to invest in resources that can enhance the quality of articles in a variety of ways. Whether it is a reduction in typos (by paying a proofreader), a clearer composition of the style (there will be more time for it once the work is spread out among more people), more articles in English with more accurate translation, especially terminologically (we want to push this forward prospectively, but again… we need more resources), or the expansion of the test portfolio itself. One of the long-term goals is to build a detailed comparison of the properties of thermal pastes and pads, but we currently do not have the funds for this. More lively discussions are usually linked to higher traffic (readers who have already read the article return to the comments…), which would help us a lot in this direction.
However, calling for activity does not mean commenting on something at all costs, that would be counterproductive – we’d rather have no discussions than those that lead nowhere. And also you don’t have to come to praise us, instead you should complain and ideally at us. From time to time we notice various objections to certain methods or to the way data is evaluated, and it is frustrating that we can’t respond to them. If you have a factual objection to something, the discussions under the articles on HWCooling are the place where we discuss everything in detail.
It is appropriate and deserving to thank all those who share our articles as a source of information in various places on the Internet. This is tremendous help, and we can still function because of it. We appreciate it very much. When this or that isn’t being discussed here, it’s great to see that we don’t do a lot of things unnecessarily and they are being discussed elsewhere where there is already a community. But we will always strive to build a healthy community based on the pillars of the knowledge society here at HWC. It would be absolutely perfect if you could mostly discuss among yourselves, and in the meantime we will always serve you something new and informatively attractive. 🙂
English translation and edit by Jozef Dudáš
Personally, it is hard to find something wrong from HWCooling’s test for me. (espeacialy for fans) Because I already know how difficult this type of work is and how much time, effort and money it must have taken. My inclination not to lightly evaluate the results of considerable effort and quality also plays a part.
But from now, I will try to look at articles from a different perspective as possible. Of coures, it will be hard for me because the main opinion between you and me about one topic usually are not opposed to each other..
I basically don’t pay to much attention to the method test or the test itself, if i go to check a fan model i go directly to the good or bad, features, conclusions, price, etc. take that info and compare my own or others just to double check before an purchase or future purchase, the english is very good, but from my PoV i prefer you focus more in analysis or review in more models, brands, aios, new tech etc, not the popular ones, or wait to have it in your hand, i know is hard make a professional review without proper base acknowledge but anyway there is not to much info about fans or a site similar like this so if you want more feedback try to provide something like what to look or need for a certain product, comments will come as far as possible.
Thanks for the feedback, we appreciate it very much. Do you have any specific tips for those lesser known cooling products that you would be interested in? You’re right – the fewer tests on anything attractive that deserves attention, the more useful they are.
mmm, types/differences of thermal pads/paste, fan/blades/pump/bearings, what is the cheap/quality over each brand like have a bar for measuring the “looking for” 👍
Just wanted to say that the thorough way in which you test is not wasted time. Even if many jump straight to the conclusion others, including me, go patiently from page to page.
Keep up the good work.
This makes us very happy. I know that we have more readers who take it this seriously, but I don’t blame anyone for the fact that the majority goes straight to the conclusion. Sometimes I come across the remark that ‘nobody can do that many chapters in a row because it is indigestible’, and I completely agree with that. At the same time I always point out that this type of tests is not read from the first to the last character, but it is the “last chapter system” (and I believe maybe a little bit of a first chapter one, where we try to get across attractive information and not describe what the reader sees in the picture :)) and the rest is for those who are interested in a particular test or instead of our interpretation from the text of the conclusion, they look at the results themselves and evaluate the data according to their own reasoning.
In my opinion, professional tests must be complex, the conclusion should reflect as many possible points of view as possible, and this is really hard to create without a proper database. 🙂
The only reason I’m a regular reader here is because of your thorough and thoughtful methodology. You are doing what all tech reviewers should be doing. Even though times have moved on to a place where real information is rare and all you typically get is marketing disguised as journalism – that doesn’t mean the future has to be that way. I think more people every day grow tired of the modern trend of hype and no substance. If you continue your solid approach you can only increase your viewer base.
We will definitely keep doing the work that we are doing. We founded HWCooling some time ago with quite good experience from a medium which also preferred to disappear rather than to change its “ideology” under the influence of entertainment and tabloid trends. I believe that we have learned from some things to such an extent that it is possible to carry out professional tests in the present. As long as one does not desire to acquire great wealth, fame or power, moral credit can be preserved. It is extremely difficult, people like you are in the tiny minority, but in the end it is like that with all things that are done honestly.
There are some extra information that I would like to see in the fan reviews:
1. More detailed analysis of the physical side of the fan, such as measuring the exact dimensions and weight (vs. what is claimed by the specs), fan and hub diameter, thickness of the blades, how much the blades are recessed, etc. While these properties do not really matter in terms of actual performance, they could help explain why the fan behaves like it does.
2. More destructively, teardown of the fans. It would allow for a better look at components like the motors, bearings and electronics that could not be identified from the exterior. The analysis of such could also help explain some observations of the fan, and could also infer long-term reliability. This, as far as I know, is extremely rare on the internet. Though obviously this would, in most cases, destroy the fan (though you always test at least two samples so this might be feasible…)
3. Adding an analysis of noise level vs. PWM%. This is a check to see if the fan has any peaks in noise levels in specific RPM ranges. (as a side note, I assume the noise-normalized speeds are scanned from fastest to slowest?)
4. More ways to visualize the data, airflow vs. dBA graph for example. Though I think this is something that will eventually be done via the fan encyclopedia?
5. Perhaps also list the speed data, not only for no obstructions environment but also for other obstacles. Maybe include them in the airflow charts as a value. The intention is for readers to be able to quickly understand what the noise-level means for their ear, if they have one of the tested fans on hand. It’s also academically interesting to see, for example, if one fan is better than the other on a specific obstacle because of superior airflow per rotation, or is it mainly because of lower acoustic penalty when placed on obstacles.
Lastly, a question on push vs pull. The airflow tests, if I understand correctly, are only tested in the “pull” orientation (as in air goes from tunnel > obstacle > fan). I think that the “push” orientation (tunnel > fan > obstacle) would probably change the ranking of the fans due to different noise behavior of push vs pull, so why is that only the pull orientation is tested for airflow? Do the static pressure tests, which are done in push orientation (fan > obstacle > sealed tunnel, if I understand it correctly), better represent the ranking of fans when used for pushing?
bro, you want a complete thesis for every model? really? there is no benefit in have that info rather of the already manufactor provide. the only point i agree with you is the last one i think that is a good approach test the enviroment, air direction, comparison, rank, etc.
Actually, the fan manufacturer does not provide much of this information.
Anyway, here I would distinguish between something like “user’s point of view” and “love of knowledge”. When something happens, you need to know why it happens and the better you know the sub-properties, the easier it is. 🙂
Thank you for the list of very useful insights. I agree with everything you write. There are many things (even beyond your suggestions) that I would like to incorporate into the tests. But one thing is the “final form” of the results and another is “what are the realistic possibilities”. And for the latter thing we are absolutely going to all limits and maybe we are beyond them (fortunately I am obsessed with these HW tests and especially fans and I can devote 18 hours a day 7 days a week to them, although it’s a bit unhealthy, but… :)). I don’t want to elaborate on this, so it doesn’t sound like crying, but if you could see the laughable financial resources we are working with… the earnings of professional tests (made in the spirit of the knowledge society) cannot be compared to consumer ones with tabloid characteristics. There are always more people around entertainment and sensations, advertising has a similar value regardless of expertise.
1.) The original plan was to put only the basic things in the parameter table, so that it would be clear and not discouraging especially in the later phase. But I admit that a precise thickness that takes into account the widest point (typically defined by protruding corners/antivibration pads) should be captured in the table. It is true that we only have it given as an indication, based on the thickness of the frame. This is also due to the fact that the frame is a “fixed” component and the anti-vibration pads, which reduce the thickness, are “movable” – they can always be compressed to some extent during mounting. But it could certainly be stated as a range of fixed and movable thicknesses (in the uncompressed state), that’s true.
We have been measuring and recording the exact weight of the fans in charts since the beginning. However, as part of data reduction in the overview table, you can only find this data in the pop-up window in the graphs after hovering the mouse cursor over the corresponding fan. It is then even possible to sort the fans by weight – from lightest to heaviest (or vice versa).
And, of course, the diameter of the hub, both outer and inner cross-section. These are very important data, which I point out in passing, if necessary, when it helps to understand some of the results. For example, to make it clear why Arctic BioniX P120 or P12 A-RGB fans often achieve lower airflow per dBA than the P12 even though they don’t have such pronounced tonal peaks… because they have a smaller outer cross section/blade surface area (the BioniX P120 and the P12 A-RGB).
The thickness of the blades is rather difficult to measure, because the cross-section of a blade varies. And also the thickness at the tip of the blade is different on different sides. I guess it would be possible to give a minimum and maximum thickness and I have thought about it several times, but haven’t justified it. For this data it would be necessary to often damage a fan, which seems to me personally a pity and inadequate considering that the exact measurements in this respect do not tell us so much…
2.) I guess you mean mainly finding the maximum voltage at which the fan still works? Such a crash test also seems a bit useless to me, considering that the fan always hits the limit at which it is destroyed. I understand the experimental angle, but you know, when you really like these things, you suffer with them. In my eyes, it’s like torturing some pet, haha. Still, I’ve been thinking about dissections with bearing analysis for a long time now. I’ve had that on to-do list for a while now, but it’s a bit of a problem to get around to it.
3.) The dependence of noise on PWM duty cycle is a perfect idea. When we can increase our reach and make more money from advertising, I believe that I will be able to delegate some of the administrative stuff to someone else and there will be room to add such measurements instead.
4.) Also a good idea, which we will consider. We don’t plan anything like that in the announced encyclopedia, its concept is built a bit differently. So that it would be good material also for users for whom the tests we publish seem unnecessarily complicated. Don’t look forward to the encyclopedia too much though, so that it won’t be a disappointment for you. There will be as much simplification as possible, with relative percentage results from partial tests, which will eventually result in some relative overall score of a fan.
5.) Sure, I have thought about indicating the speed on obstacles several times, but it ended up in the category “it would have to be at the expense of something else, in my opinion more important”. It may not seem like it, but even a thorough recording of speeds across six modes is at least an hour’s work. Remember, we work with multiple fan samples to be as accurate as possible, especially the values for maximum and minimum speeds.
6.) I will comment on the pushed vs. pulled airflow as an extra point. 🙂
I don’t know if you managed to register my reply on ThermalLeft’s comment before writing yours… is probably at the time of its publication you had already started writing your comment. It may contain answers to some of your questions. Apart from the acoustic impact, the anemometer sensor is in front of the fan (and not behind it) for turbulence suppression, thanks to which the impact of tunnel resistance can be reduced. This can also shape the results of the fans somewhat unfairly depending on whether they have a higher or lower static pressure. When testing in a scheme where the airflow is measured behind a fan (pushed airflow) it is necessary to have a very long tunnel so that the results are not distorted at the level of the different turbulences of the different fans. To counteract this effect, it is customary to use a flow straightener, which is one of the elements that increases the resistance of the rotor to a certain extent with pushed airflow, together with which the airflow is reduced and to a slightly different extent for each fan. We also use a flow straightener, but it works on a slightly different, less restrictive principle, which is also concisely but clearly visualised in this diagram. Our tunnel is unsuitable for “pushed airflow” due to its short length. Of course we can flip the fan, but the data at the level of the anemometer will be irrelevant, because the air streams at its level are characterized by excessive turbulence. As part of the development of the methodology, we have built several tunnel designs for both pushed and pulled airflow and finally, after a lot of tinkering, we came to the conclusion that the tunnel we use is the best one for a fair comparison of all the designs of computer fans. The knowledge from the internal research cannot be interpreted in this way in the form of a commentary, but one day the time will come when we will properly process everything in the framework of detailed analyses, which will be interspersed with measurements in various contexts. But we have to do it gradually and teach enough people the basic things about the functioning of fans. It would be ineffective now. We would certainly chat with ThermalLeft about it, but otherwise nobody would read it and it would be very time consuming to put it together in a reasonable form.
Anyway, the relationship between the position of the anemometer and a fan has nothing to do with the position of the fan and an obstacle (push/pull). What is important is the pressure difference at its level, which, of course, varies depending on the orientation of the fan. And it is clear that, in the scheme tunnel > obstacle > fan and tunnel > fan > obstacle, there are differences at several levels. These topics, supported by measurements, will be dealt with in separate thematic articles, which we’ve had marked for a long time in the outline of principal analysis articles, which we started in July with simpler topics. Gradually we will work our way up to more and more complex topics, which will build on each other.
PS to the last question: the results chapter “Static pressure, efficiency by orientation” compares two situations without obstacles. In one the rotor orientation to the closed tunnel is push and in the other pull. These results show in which position the fan tends to perform less well. Of course, this cannot be interpreted as meaning that at a coefficient of 5.000, the orientation will result in several times lower/higher airflow. Here again it must be remembered that the static pressure is measured at zero airflow, i.e. at an enormous environmental resistance, which is incompatible with any practice. In this case, even the motor itself has a different/higher output than normal. This is just an indication of a trend that can only be achieved at the level of single percentages, depending on the resistance of the obstacle. We also plan to address these things in separate articles at a level that will be well understood by the average user.
I’m sure the site will gain more views as times goes on, as long as the quality remains high. Aside from high quality, there are also many unique tests, like testing different variants of 12400/13400 for example, which I am sure many can appreciate.
I wonder if there are feasible ways to improve the broad appeal without compromising on quality. For example, I noticed that many of the reviews here are published several days after NDA lifts, which I am sure will hurt views by a lot. It’s obvious that you are already prioritizing publishing such reviews, but if there’s some way to publish them even earlier (with the same quality of course), it would be perfect.
On the extra things I want to see in fan reviews, I totally understand that those involves a lot of additional work, so just take them as things to consider instead of a complaint. I am glad (and, honestly, not surprised) that you have already considered most of these things.
On the “teardown” of fans I only mean taking them apart, and for most fans it involves destroying the fan, hence “destructive”. Physical stress testing does not seem to have much value, haha. And yes I totally understand not wanting to destroy the fans, but for the sake of knowledge it might be needed in some cases. I’ve been reading some Chinese discussion from Chiphell mentioning from which factory each brand sources their fan/components from, which is quite interesting.
The encyclopedia might not be that useful to me, but it sure will be great for sharing the data to others.
On the push vs pull point, I totally understand putting the anemometer in front of the fan (yes I read your exchange with ThermalLeft thoroughly and learnt a lot! Perhaps an article can be written using the mirrored Thermalright fans to illustrate this point…), so my intention was to ask only for the relative position of the fan and the obstacle. I look forward to the articles you have mentioned.
Yes, we are also trying in other areas than fans. 🙂
Publishing something with an ending NDA is very challenging, especially because we usually get samples late and in combination with time-consuming methodologies (especially for graphics cards) it is often impossible to release tests on the first day. For example, we only got to the RX 7800 XT on the 4th of September and the RX 7700 XT didn’t arrive until the 6th of September.
Even if we have samples in some sort of advance, we are missing drivers and such for a change. Of course it makes me angry, but we can’t do anything about it. Even though we publish articles in English with a reach abroad, companies index us as “local medium in a small market with low purchasing power”, so we often end up on the tail of priorities. But it doesn’t matter, if we are not among the first, we at least try to find useful topics that others don’t address (like the comparison of different steppings of the same CPU model, as you mentioned). We will continue to do this in the future, and I believe it will become more and more intense as time goes on.
It would be nice to have a basic Quick Rundown style of fan reviews for idiots like me WHILE keeping the scientific complexity of your fan reviews of course. I can’t always convert this noise normalized measurement to real world usage. Just straight up tell me how hot and how loud a fan gets while playing a game and also while idle. And compare them to other brands of course. An “idiot -proof version” of the review should be part of the more complex reviews, just a single page or less. a child’s corner so to speak.
I think you will find the upcoming encyclopedia interesting. Its presentation will be very simple, but the scores (partial and overall) will be based on the data from the in-depth tests.
The encyclopedia will be a separate publication from the standard tests, but you will find it online, regularly updated.
how could I have missed this article. I can’t wait for the launch, thanks!
I have a few suggestions on how the fan reviews can be more digestible and informative. Some of these points have already been mentioned but I will reiterate them.
1. I think some people may be intimidated by the number of pages in each review. The methodology and manufacturer specs is mostly repeated info that can be collected at a single place. In fact there is already a complete methodology post from 2021.
2. Due to the way the airflow and static pressure graphs are created, the test result pages themselves are also very long. By using a two axis line graph with sound pressure on one axis and airflow or static pressure on the other, the graphs can be condensed. Perhaps the concern is that many lines on the same graph will be hard to read, and I can think of two ways to make it easier to read. The first is to use only a select number of fans of similar price, form factor, or performance to create the line graphs, and include the one axis bar graphs separately. The second is to create an interactive graph where lines can be turned on and off.
3. Pictures of the hub and bearing disassembled would be interesting. To see the circuit board design and number of poles. Many fans now claim to have some kind of fluid dynamic or hydraulic pressure bearing. It would be interesting to compare them.
4. This is just out of personal interest. Thermalright fans have recently become popular outside of China and while they have received good reviews, Chinese users for whom these fans have already been available for a number of years seem to have a low opinion of them. A HWCooling review of these fans would be insightful.
Thanks for the useful feedback.
1.) Agreed. For users who have the desire to know in a minute which fan is the “most suitable”, the scope is really intimidating, but for them the upcoming encyclopedia will be useful for a quick overview.
I consider it appropriate to leave the details of the methodology in each test. Of course, you are right that these are repetitive, unchanging chapters that do not interest the constant reader. But when someone comes across one of our tests for the first time, it is useful for them. Just a hyperlink from the text to another article with the complete methodology is not enough. It’s not effective. Few people click through somewhere, but if it’s already included in the article they’re reading, they’ll go through it. Sure, not everything, but at least something…
2.) Line graphs to reduce visual data, sure… we can discuss it, but I can’t imagine it very well, with respect to maintaining good clarity. Cutting the article into very many chapters still seems to me a more reasonable and digestible option. However, if there are more voices supporting your suggestion, saying that the current presentation seems “worse”, then of course we will look into the possibility of a different way of visualization. For example, using 2D line graphs as you suggest. The only thing I will always distance myself from are P/Q curves and gradually we will discuss in extreme detail at just how many levels they are extremely misleading.
3.) We will probably add this sooner or later. I admit that for the target group of readers of detailed fan tests these things may be missing.
4.) We were not in contact with Thermalright until 7/2023, but that has changed. TR fans will definitely appear in HWCooling tests in the future. What models are you most interested in?
Note that the low opinion of Chinese users on Thermalright fans is mainly on the reliability and quality control side. Things like fans failing after a few months, unreliable when mounted at the top of cases, some samples being balanced poorly.
Somewhat related, while Thermalright was mainly Taiwan-based, it seems that most models (fans and coolers) released in recent years are developed by the mainland China division instead.
And yet those fans often use dual ball bearings, which should ensure longer service life. But the bottleneck will probably be elsewhere else. I admit that I have practically no experience with the newer Thermalright fans. I remember them from ten years ago, where they were often criticized for higher minimum speeds, for which in very quiet operation they could not compete with AAA fans from Noctua, etc., thus decreasing the attractiveness of otherwise perfect CPU coolers. This was also the case with the Silver Arrow SB-E’s excellent heatsink, whose design I still consider to be unbeaten. At a significantly lower weight than the NH-D15’s heatsink, it was no worse even with the same, below-average fans. I think with the arrival of the Silver Arrow ITX came fans with lower minimum speeds, but the heatsink itself wasn’t as efficient. A nice nostalgia is also the IFX-14’s heatsink or the unique heatsinks on the VRM of graphics cards.
Soon our tests of SSD coolers will include the HR-10 2280. Thermalright fans will then follow it, so we’ll see…
Based on Thermalright’s website most of their current fan lineup uses what they call S-FDB. Double ball bearing is only on a few models like the TL-B12 extreme and TL-E12 extreme. Disassembly of S-FDB fans have shown that it is just rifle bearing.
The fan models I’m most interested in are the TL-C14, which has a similar rotor to the NF-A14, the TL-C12C, their most affordable and widely available 12cm fan, and the TL-B12 and its ball bearing and LCP variants, to compare with other Gentle Typhoon style and LCP fans. I would also be interested in seeing the differences between the 12cm dual tower coolers Peerless Assassin 120 (6 heatpipes crimped fins), Phantom Spirit 120 (7 heatpipes crimped fins), and Frost Tower 120 (6 heatpipes soldered fins).
Based on your comment, I tried to find more information on “S-FDB”, and wow is this (Chinese) article a good read: https://post.smzdm.com/p/avxpo9wm/
I haven’t gone through them yet, but the articles by the same author (WittmanARC) also seem to be very informative (you have to click the “original” tab): https://zhiyou.smzdm.com/member/4171119816/
It’s lucky that we have good translation software these days.
You’re right, I was just looking at some TL-B12 models + the combination of ball bearings with TR from the past, when they had a larger share in the offer (compared to the then widely used competition with sleeve bearings).
I’m writing down the TL-C14 and the TL-B12, including the Extrem LCP variant. And Thermalright CPU coolers will be added over time as well, and we would also give priority to the models you mention. 🙂