Testing Methodology

Although the testing of a cooler appears to be a simple task, that could not be much further from the truth. Proper thermal testing cannot be performed with a cooler mounted on a single chip, for multiple reasons. Some of these reasons include the instability of the thermal load and the inability to fully control and or monitor it, as well as the inaccuracy of the chip-integrated sensors. It is also impossible to compare results taken on different chips, let alone entirely different systems, which is a great problem when testing computer coolers, as the hardware changes every several months. Finally, testing a cooler on a typical system prevents the tester from assessing the most vital characteristic of a cooler, its absolute thermal resistance.

The absolute thermal resistance defines the absolute performance of a heatsink by indicating the temperature rise per unit of power, in our case in degrees Celsius per Watt (°C/W). In layman's terms, if the thermal resistance of a heatsink is known, the user can assess the highest possible temperature rise of a chip over ambient by simply multiplying the maximum thermal design power (TDP) rating of the chip with it. Extracting the absolute thermal resistance of a cooler however is no simple task, as the load has to be perfectly even, steady and variable, as the thermal resistance also varies depending on the magnitude of the thermal load. Therefore, even if it would be possible to assess the thermal resistance of a cooler while it is mounted on a working chip, it would not suffice, as a large change of the thermal load can yield much different results.

Appropriate thermal testing requires the creation of a proper testing station and the use of laboratory-grade equipment. Therefore, we created a thermal testing platform with a fully controllable thermal energy source that may be used to test any kind of cooler, regardless of its design and or compatibility. The thermal cartridge inside the core of our testing station can have its power adjusted between 60 W and 340 W, in 2 W increments (and it never throttles). Furthermore, monitoring and logging of the testing process via software minimizes the possibility of human errors during testing. A multifunction data acquisition module (DAQ) is responsible for the automatic or the manual control of the testing equipment, the acquisition of the ambient and the in-core temperatures via PT100 sensors, the logging of the test results and the mathematical extraction of performance figures.

Finally, as noise measurements are a bit tricky, their measurement is being performed manually. Fans can have significant variations in speed from their rated values, thus their actual speed during the thermal testing is being recorded via a laser tachometer. The fans (and pumps, when applicable) are being powered via an adjustable, fanless desktop DC power supply and noise measurements are being taken 1 meter away from the cooler, in a straight line ahead from its fan engine. At this point we should also note that the Decibel scale is logarithmic, which means that roughly every 3 dB(A) the sound pressure doubles. Therefore, the difference of sound pressure between 30 dB(A) and 60 dB(A) is not "twice as much" but nearly a thousand times greater. The table below should help you cross-reference our test results with real-life situations.

The noise floor of our recording equipment is 30.2-30.4 dB(A), which represents a medium-sized room without any active noise sources. All of our acoustic testing takes place during night hours, minimizing the possibility of external disruptions.

<35dB(A) Virtually inaudible
35-38dB(A) Very quiet (whisper-slight humming)
38-40dB(A) Quiet (relatively comfortable - humming)
40-44dB(A) Normal (humming noise, above comfortable for a large % of users)
44-47dB(A)* Loud* (strong aerodynamic noise)
47-50dB(A) Very loud (strong whining noise)
50-54dB(A) Extremely loud (painfully distracting for the vast majority of users)
>54dB(A) Intolerable for home/office use, special applications only.

*noise levels above this are not suggested for daily use

Introduction & the Liquid Freezer III 280 mm A-RGB White AIO Cooler Testing Results
Comments Locked


View All Comments

  • Maksdampf - Wednesday, March 13, 2024 - link

    Coldplate size (thats the technical term, not „contact plate“) being too little is not a reason for the bad performance. What matters is not the size of the copper chunk, but its active area. The active finned area must be the same size or larger then the DIE underneath the IHS. But the fin area and water channels are similarly sized as the competitors.

    More likely your decades old testing Station has a far thicker baseplate or DIE size compared to a real CPU, so the measurements are not realistic. Time to make a new DIE simulator.

    But i assume there is something else wrong too.
    The baseplate in the Liquid freezer II is the same, but it is performing better, so this test result must be worse due to a bent IHS or a fault with the mounting hardware.

    Since every other media outlet so far has seen improvements compared to the LF II series, i would not have published these results without further research into the issue.
  • E.Fyll - Wednesday, March 13, 2024 - link

    The descriptive term "CPU contact plate" is very frequently used by professionals and amateurs alike. True, the correct technical term for liquid coolers is "cold plate" but I personally choose to use casual language and descriptive terms when appropriate, as I don't expect every single reader to be familiar with the technical terminology.

    The mass and size of the cold plate do matter in several ways. The "active" area does matter but it is not the only thing that does. It actually doesn't even matter the most - the shape is what matters the most as it affects both the movement of the liquid (boundary layer fluid mechanics) and the transfer of energy between the mediums. If the liquid contact area of the cold plate was all that mattered for performance, it would be shaped in the form of a small rectangular finned corridor for the fluid to flow through, similar to how finned radiators are shaped, multiplying the liquid contact area compared to current designs. Still, that design doesn't work quite well despite the much greater liquid contact area, for a number of reasons.

    The testing station does not fully replicate a specific CPU's die, which is a very high thermal load concentrated in a very small area, but spreads the thermal energy almost evenly across the entire contact plate, replicating an ideal 40x40 IHS. It is perfectly valid for measuring the thermal resistance of coolers without relying on case-specific parameters. If I were to make the faux die that is the exact shape and die of any processor, the test results would be completely incomparable not only between generations of processors but between different sockets as well.

    I have learned not to take too much heed to other people's test results as I frequently cannot match mine with those no matter how hard I try. I did double-check this review's test results because I found them very strange myself, which is why the review was late (again). Yes, the cold plate may be similar/identical but not the entirety of the cooler is. There are myriads of factors that affect performance, down to the very paint on the radiator.

    For those that are a bit distrustful, or even cynical, one should also consider that I could very well be "cooking" the results to whatever I want and avoid confrontations such as this one but, as you can tell, I choose to go with the actual readings and deal with the fallout instead.
  • Maksdampf - Thursday, March 14, 2024 - link

    Yeah, i get it, turbulent flow for boundary layer separation is important too. I wasn't saying that the internals don't matter, and you weren't saying anything about internals in your article either. I was just saying that you can't judge a book by its cover or a waterblock by its coldplate size.
    The mass of the coldplate is really not that important. A heavier and thus thicker coldplate performs worse in most cases because even pure copper would be an insulator between the flowing water and CPU, adding thermal resistance. Also when talking about thermal mass, you mean thermal capacity most likely. Thermal capacity of copper is almost negligible compared to water. So if you wan't mass simply for thermal inertia in a waterblock it is much better to have water there instead of copper soaking up the heat. So less copper performs better in many cases.

    I understand that the somewhat bad results of this CLLC are most likely down to the test platform specifics and not your thorough measuring process. But still, how valuable are they then?

    If you knew this and even doublechecked the results, how could you not write something about this discrepancy and its most likely cause in your review? Instead you make something up with coldplate size, which doesn't matter on a real CPU and only the very thermodynamics literate and curious people can summize that the Testing setup may be the reason here why this CLC performs so bad an others so well.

    I get it that you don't wan't to throw away 20 years of comparability between coolers by still using the same platform. But then i don't undertsnd why you are not using any of the Data. Why is the graph not a javascript generated one where i can select my comparision with a drop down menu?

    I also don't get these 12V and 7V measurements instead of noise normalized bar charts which are much easier to undertand for a consumer. 20 years ago people limited their fanspeed by connecting it to the +12V and +5V DC from the PSU directly, but nowadays in the Age of PWM nobody cares.
    All other outlets have switched to noise normalized values, which you could be doing too, since you have the data vor a curve. But you chose not to.

    I myself was developing waterblocks 20 years ago and hat a very similar test platform using a microcontroller and a fet heater. I used a Athlon64 IHS with a small 15x15mm copper block soldered onto the middle where the fet was screwed onto and a probe was drilled inserted in a small drilled hole on the side to measure exactly central to the die.
    Even this had slightly different results to real CPUs. I have stopped doing that back then since i could not be true to my aspirations.

    These Tests are not good for anybody looking for an AIO or Aircooler today. Things like offset mounting for AMD or contact frame play a huge role and you just ignored that.
  • osv - Friday, March 15, 2024 - link

    re: your misplaced claims, for instance, "These Tests are not good for anybody looking for an AIO or Aircooler today"

    what i see is that you didn't actually read the review, and therefore your comments are not helpful, and your criticisms are not valid.

    for example, your claim: "nowadays in the Age of PWM nobody cares"... umm no, as the article repeatedly pointed out, the speeds of the pump vs. cooler fans vs. vrm fans can't be controlled individually, which is a drawback for tuners who want those options, and had it years ago... i appreciate that the author made that point wrt price, while you apparently don't understand what matters to people reading these articles.

    i appreciate that the data posted backs up his conclusion that "This places ARCTIC’s latest 280 mm cooler in direct competition with 360 mm AIO coolers and even close to the 420 mm version of the Liquid Freezer II, which used to perform a little better but at a much higher noise level."

    you claimed: "the Testing setup may be the reason here why this CLC performs so bad"... except that it didn't test badly, it was quieter than the competition, which is a high priority for many readers, and the author got that; you didn't.
  • Lolshutupbro - Tuesday, March 19, 2024 - link

    The pump speed, fan and vrm fan can be controlled individually, there is a cable included….
  • m3city - Thursday, March 14, 2024 - link

    I asked this before, but never got a response, and asking google does not really give an answer as well: why is there no comparison to an air cooler? A stock one would be the best, but any would be great. I still don't know what will I gain in terms of loudness, temps, OC/boost headroom?
  • PeachNCream - Thursday, March 14, 2024 - link

    Here ya go

  • m3city - Friday, March 15, 2024 - link

    Thank you very much! That is exactly what I was looking for.

Log in

Don't have an account? Sign up now