How will we test?
Probably as early as next month, in September, we will release an analysis of the first case that will go through our new testing methodology. We already have a rough outline of it and we would like to present to you what we are going to do and how and why we are going to do it. This is also to give you a chance to comment on things you might think are missing that we could easily incorporate into the standard concept.
The computer case tests will be partly about giving you a good idea of what a particular model allows you to do in terms of assembly (of PC components) or itself (in the context of situating itself in a space), but also about things that can be compared in charts. Such things that answer the question of what kind of case is (un)suitable for whom in terms of cooling or noise. Cooling performance may be sufficient, but in terms of acoustic properties a given case may already be inadequate. That is, depending on who it is for, everyone can decide for themselves based on the results of the measurements.
„Cooling performance“…
Something we could call cooling performance also tells us whether a given case is more or less suitable for the selected components. Specifically, the air circulation that increases or decreases the cooling performance (of coolers of CPUs, graphics cards, motherboards or SSDs). It depends not only on the use of specific fans in a specific number, but also on the specific layout of the case itself. And we will examine this at different noise levels, as we have done so far. There are a number of fixed noise levels, ranging from very low, through low, medium, higher, to high, which may be of interest to users looking for the highest possible cooling performance, regardless of noise level.
The measurements will be done on a Mini-ITX motherboard. In larger cases this will often look comical, but it’s an opportunity to get a comparison across a wide spectrum of different formats. And that’s the whole point. How much “better” is a large full tower at cooling than a small SFF case? And couldn’t it actually be the other way around, that a smaller case means lower temperatures? It’s said to be nonsense and… well, we’ll see.
In the Asus ROG Strix B760-I Gaming WiFi motherboard, the Intel Core i9-13900K processor will end up set to maintain stable performance under all circumstances (i.e., even with the worst cooling). Only in this way will the temperature results be comparable to each other. In addition to this, of course, what is also required is a fixed setting of the fans speed on the CPU cooler. Or rather coolers, because we will be using two of them. The primary Noctua NH-D12L doesn’t fit into small cases, no matter how “low” it is, and in such situations we reach for the top-flow cooler Noctua NH-L12Sx77. For a while we wondered if we could use it for all the cases, but that probably wouldn’t make much sense. Firstly because of the fan orientation, which would be unnatural for larger tower cases in terms of airflow, and secondly because of its relatively low cooling performance. There will be different levels in this regard as well, a lower CPU performance will be combined with the smaller cooler (NH-L12Sx77) than with the larger one (NH-D12L). Thus, all cooling results will not always be comparable.
To test with the same fans? I don’t think it’s advisable, because you could get very misleading results from such comparisons. This is because while for one case a chosen fan may be optimal, for another it may not be, and with a different fan or multiple fans chosen it could be the other way around.
… and acoustics in detail
The reason we will stick with factory fans only is because we respect that some manufacturers have gone to the trouble of implementing psychoacoustic optimizations. Thus, with the fans used, the sound expression will be the most pleasant. If such fans were to be replaced with other, reference, non-optimized fans, there could be unwanted tonal peaks and always (depending on the choice of specific fans) different ones. Thus, we will only test with the fans that are supplied with the cases. Other situations, with the addition of “custom” fans, will only occur if a case is not supplied with fans, but their mounting is being counted on. So you just have a free hand in the choice.
The PC case tests will also focus on detailed monitoring of the acoustic profile. This will be interpreted in spectrograms and our task will be to eliminate the most extreme tonal peaks. Either by moving a fan further away from an obstacle or by “detuning” them to different speeds. It will be important to get the information to know what you need to do to “improve” the acoustic profile over the factory setting, or rather over controlling all system fans with the same PWM signal.
One of the unusual measurements will be vibration analysis. We will record this with a three-axis vibrometer at various locations and point to the five most intense, potentially weakest points. We may also somehow incorporate measurements with and without an HDD (with seeking heads) fitted. In such a situation (with an HDD), we will mainly investigate the degree to which the case can damp the noise of such a source. We assume that due to the declining interest of mainstream consumers in HDDs, case manufacturers will not be too concerned with such things, but for someone, it may also be one of the added values. Most importantly, it could be a relatively valuable measurement, even if only for a few users who, for whatever reasons, still use HDDs and will continue to use them.
For completeness of the case tests, we also consider it important to have an overview of the USB port speeds. These will be monitored similarly to what we do with motherboards. It is a good idea to check that the claimed bandwidth is being achieved and to know what throughput some connectors are achieving in general. This is not always officially stated.
So, now you know our plan roughly. Is there anything missing in it that would be good to focus on? We can, as long as it doesn’t increase the time commitment significantly. It’s one thing to design a technically perfect methodology, and it’s another thing to be able to implement it. We are looking for some consensus after which a reasonable basis for the evaluation of PC cases can be constructed in a reasonable time.
English translation and edit by Jozef Dudáš
- Contents
- How will we test?
On whether coolers should be as standardized as possible – I am conflicted. Using the same cooler across cases can indeed give an idea on how well the airflow of the case is. However, cooler clearance is also an important metric of the case; using a smaller cooler than what’s supported is being unfair to the case itself. Thus, there’s also an argument for using the best cooler that can fit in the case for tests.
Anyway, the cooler selection needs to be expanded – it cannot account for tests of SFF cases. L12Sx77 may only fit in a very limited number of SFF cases, for instance. Many cases also expect you to use an AIO liquid cooler. Say, Dan A4-H2O for example – it can fit 55 mm air coolers, but also a 240 mm AIO liquid cooler.
It might be a good idea to have a larger selection of standard coolers on hand – 240/280/360 mm AIO liquid coolers, air coolers ranging from 37 to 168 mm. I can give you a list if you wish – I know of popular coolers with 37, 47, 53, 67, 70, 77, 100, 110, 115 and 135 mm height, some downdraft, some towers.
It would also be a good idea to find and test the “optimal” airflow configuration of the case, but that can be difficult to tell at first glance and would surely take too much time. Perhaps you should ask the manufacturer for advice on the recommended layout before testing the case. If it’s not actually optimal – that’s their responsibility.
It’s a complex subject, and I added a reminder as well. I guess Ľubo wasn’t too thrilled, ha! That’s a good point about those AIOs. I tend to prefer using 240mm as the most widely used. Even with that small change, testing gets very complicated.
Yes, it’s a very complex topic and I can imagine a really complex treatment of it, but I couldn’t do anything else in such a case, haha, just PC case tests.
With the use of liquid coolers, the situation changes completely, because they (unlike tower coolers) participate very actively in the system cooling. It can be said that with each liquid cooler different results will be achieved. Of course, this is a very important factor in practice, but in the case tests we want to eliminate it at least a little bit. Nevertheless, (with one or even several models of coolers) it will be very difficult to draw any general valid conclusions about the cooling. A case supplied with four fans (three inlet + one outlet) will be completely different when a 360 mm liquid cooler is used (the final scheme will then often be with negative air pressure) than in the case of a tower cooler, when there will be positive air pressure in the case. These are also the reasons why I am a bit discouraged by testing cases in general and it will be a rather marginal topic on HWC (for the thematic completeness of the content), which we don’t want to waste too much time with. And instead of cooling stuff, which will always be controversial, things around fan acoustics will get more space. This will be very interesting, because different cases have different obstacles, so there will definitely be a lot to analyse. 🙂
Thank you for the detailed insight. Of course I understand what you are saying. We have to be economical and the tests of PC cases will have to be less time-consuming. It’s one thing to make a perfect test methodology and another thing to make it so that we can actually do the tests, haha.
The time effort has to be relatively low in this case to be tolerable, because besides case tests I am responsible for quite a lot of other testing methodologies and there is not much space left for extremely detailed things. So yes, I agree with the high variability of CPU coolers, it would be great to evaluate the same cases with different coolers. Unfortunately, this is unrealistic from a time point of view and we will rather do it by focusing mainly on cases for which the NH-D12L cooler will be suitable. We will have the smaller NH-L12Sx77 ready for some of the smaller cases to be “at least in some way” tested. We can only do what is within our capabilities. I hope that in the future the time possibilities will improve and the work can be distributed among more authors. We will work on this looking forward as well. For example, now we have the very experienced Adam Vagner coming over from another editorial office, who will be testing graphics cards instead of me, so there will be more space for cooling topics. But these are already taken up by the upcoming case tests, which will be joined by CPU cooler tests in October. In neither of these cases can we take it to extremes as is the case with fans, for example. With our capacity it would not work out time-wise, I can’t go beyond 18 hours a day. 🙂