Allgemein

Interactive case fan database – Test methods and equipment

Our fan measurement chamber and claim

Since there is currently no source that offers realistic and usable data even in direct comparison, we invested a lot of time and money and simply developed our own fan measurement station with the advice of a large German cooling and air conditioning manufacturer and then calibrated it on site in the laboratory. Our colleague Pascal Mouchel did a great job and the result is quite impressive. The heavy and massive body with two chambers made of thick MDF boards was calculated in cooperation with the manufacturer’s engineering, planned in the implementation and of course also soundproofed.

Of course, what we can measure from now on doesn’t meet any ISO standard, but the differences to the million-dollar original chamber of the supporting manufacturer are in a range where the manufacturing tolerances of the fans already start. Thus, the ideal can be approached quite well, especially since the installed measuring devices were individually selected from a larger quantity and then also professionally calibrated at great expense. This makes such an investment, including hardware and software, almost worth a five-digit amout, but you don’t have to spend a six-digit amount to be able to objectively evaluate a fan in the end. That’s because we measure a total of 4 installation scenarios in 100-RPM increments, then create a high-quality long-term audio recording over the entire speed band in each case and offer a frequency analysis. In addition, we provide all DC and PWM controlled results up to the respective power consumption.

Since a lot of know-how, time and also money is involved in the setup, we will only describe the things that are really necessary to understand the tests here. This is not a blueprint for rebuilding. A very simplified version with a long tube instead of the chamber would have been much cheaper, but it is so inaccurate due to the design that the values can only be considered as a rough estimate with many errors. The pressure drop is immense and a serious sound level measurement is simply impossible due to the resonant body thus created. Values below approx. 30 CFM can hardly be evaluated in a meaningful way. There have also been such tubes as a promo action of various fan manufacturers to measure, but more than nice toys, these are all not.

Tests as case fans and on radiators

Currently, the question always arises as to what characteristics a 120 or 140 mm fan really has. Not every model is suitable for all radiator thicknesses, and some supposedly powerful fans lose so much pressure on radiators that they can hardly be considered suitable. The specifications for airflow (“throughput”) and static pressure in the data sheets are of little help when something works well on a slim radiator but completely fails on a 45 mm radiator. For use, both 120 mm and 140 mm use a thin ST30 from Alphacool, which in terms of characteristics closely matches the typical 25 mm radiators of AiO water cooling systems, especially in terms of measurement data, an XT45 for the medium-thick radiators with very tight fins, and a thick UT60 for the heavy models.

The fans are specially decoupled and screwed tightly. Because the stator (i.e. the outer frame), must not move. Nomen est omen. Some laboratories measure fans with so-called wire spiders and a spring tension in the suspension. This is technically wrong, because a possible humming cannot be measured this way, because the resonances are compensated by the springs. We have already provided this proof once, where a fan that was completely inconspicuous in the laboratory then already began to vibrate at certain speeds while lying on the smooth table. By the way, we also do this very simple preliminary test in order to pay special attention to such speeds during testing. Glass plates are particularly suitable for this. Each of the chambers is also appropriately lined with dimpled foam and designed in terms of material so that there are hardly any disturbing influences.

In the picture below we see the middle partition between the two chambers, which carries the fan and also the radiator. Of course, decoupling is again a big issue here and we thankfully had expert help in calculating the volume for the chambers and the positioning. By the way, the “honeycomb” behind the fan was recommended to us by Black Noise and our industry partner and does its job perfectly. Here, too, the shape, honeycomb diameter and depth were optimized in advance in a simulation and the whole thing was then printed out in the 3D printer.

As a result, all coolers are equally well integrated, since each has a different exit angle and precisely this problem can be compensated for with this. The honeycomb reduces the problems with the tear-off edges and the airflow is directed directly to the outlet. You are really measuring what is being delivered by the fan.

Volume flow

We measure the volume flow at the outlet of the second chamber, where the air is blown out. This area is covered relatively accurately by comparative measurement in the measurement setup of the cooling unit partner, so that our measuring device delivers quite reliable results to the electronic measurement data acquisition, which coincide quite well with the reference data of the professional measurement. What is important here is not the price of the equipment, but rather the expedient positioning and accurate calibration with countless series of comparative measurements.

The control is done by an Aquaero from Aqua Computer purchased by us, so we can control and test the fans both by voltage (DC) or by PWM. We need both, because many fans, as some people don’t know, can’t be brought to the lower and upper limits of the speed range with pure voltage control and also show other anomalies, which we will also write about at the appropriate place in the individual tests.

Static pressure

The measurement of the static pressure is done as usual as a differential pressure measurement. For this purpose, the special “cup” is clamped in such a way that it closes airtight. Here too, of course, we used borrowed, professional equipment for remeasurement and time-consuming calibration. For this measurement, we also use  a device we purchased ourselves and also collect the data wirelessly. Since many now try to copy, the angles of the cups are of course secret, because without knowledge of the exact dimensions of the funnel, a replica is virtually pointless and it also only really works quite accurately within a fairly small tolerance range. We have also made comparative measurement series for this. In practice, the whole thing looks like on the picture below.

The installation is problem-free and the self-printed pressure receptacles (see picture above) now close perfectly thanks to a special ultra-soft seal. The rest is just a matter of calibration, which we also had to do. It’s unlikely to be much more accurate with such semi-professional methods. However, especially when measuring the pressure, errors are quickly possible, which are above all not immediately recognizable. Even if the deviations are small, we had to measure everything again at the beginning.

 

Noise emission (“loudness”)

Measuring the noise level is a bit tricky, but works quite well in the evening hours at the measurement location in the isolated and tempered chamber. We deliberately chose dBA because values below one sone can hardly be reliably recorded with still affordable and calibrated equipment and the software conversions of various software programs tend to confuse and become inaccurate in this low range. Then rather dBA, especially since most can start something with it. The measuring distance is 50 cm to the center axis of the fan input.

One of our own purchases is a calibrated measurement microphone with XLR connector and low-noise USB interface. Here, too, several specimens had to be tested against each other to find the microphone with the best signal-to-noise ratio. The measurements are made in the evening and night hours in rural areas with a hard low cut at 50 Hz, so that one can already be quite satisfied with a basic level of under 25 dB(A). Since everything will be above this anyway during the measurement setup with a distance of 50 cm, this should therefore not be a problem. The noise that can be heard in the normalized audio recordings is low enough not to distort the measurements.

We also took the feedback from the community and evaluated the frequency band for each measurement, so you not only get the SPL values (sound pressure) in dB(A), but also a nice frequency analysis that helps to describe the sound character perfectly. Bearing or motor noise, vibrations or the tearing noise on the rotor – everything becomes mercilessly visible with this.

However, we will not go into details and solutions here, because there is still some foreign know-how in this setup and some things would probably be too far for the normal consumer. If you are interested and would like to build something like this, please contact us. This also applies to all those who would like to contribute suggestions and hints, because we are still at the very beginning and can also still correct or expand.

209 Antworten

Kommentar

Lade neue Kommentare

ipat66

Urgestein

1,396 Kommentare 1,406 Likes

Wäre heute Morgen fast vom Sofa gerutscht.
Was lange währt wird endlich gut!
Schöne Überraschung.
Werde mich mal durchklicken.

Danke Igor.
Danke Pascal.

Das wird ein wirklich praktisches Werkzeug für den PC-Bau werden.
Einzigartig.

Edit:
Der direkte 1 zu 1 Vergleich führt bei mir zu (beta) – 404 ....
Aber Beta ist ja nur der Anfang :)

Antwort 5 Likes

RX480

Urgestein

1,885 Kommentare 874 Likes

Könnte man noch zusätzlich die W einbauen?
(quasi als Alternative zu 1000rpm, ... da scheinen ja einige Modelle auch noch tiefer zu funzen)

View image at the forums

Antwort Gefällt mir

e
eastcoast_pete

Urgestein

1,642 Kommentare 971 Likes

@Pascal TM-Custom/Igorslab , @Igor Wallossek :
Toll! Da habt Ihr Euch echt reingekniet, die Messungen und daraus resultierende Datenbank können sich sehen lassen! Derartige Informationen sucht man sonst vergebens - bis jetzt, denn hier sind sie, die Lüfter Daten. 👍👍👍 Und ich hab schon mal angefangen, da zu stöbern...

Antwort 1 Like

Igor Wallossek

1

10,408 Kommentare 19,386 Likes

Jetzt nicht mehr, das war die einzige Seite, die noch auf "privat" stand. Muss mur so durchgerutscht sein. Danke! :)

Antwort 2 Likes

Igor Wallossek

1

10,408 Kommentare 19,386 Likes

Dann schau mal in die Detaildarstellung, da sind sogar die Langzeitmessungen über alle Drehzahlen als Kurve drin ;)

Antwort 2 Likes

Rizoma

Veteran

184 Kommentare 156 Likes

Schon mal nettes Teil aber mir fehlt was. In der regel sucht ein Nutzer nicht den leisesten Lüfter oder den mit den meisten Luftstrom sondern den der den besten Kompromiss aus beiden bildet. Von daher wäre es meiner Meinung nach sinnvoller wenn 2 Metriken angezeigt werden sollten Von mir aus Lautstärke als feste Metrik da sie in aller regel die wichtigste ist und dann wahlweise eine 2 der anderen Metriken damit man sich ohne 2 Fenster Methode das gewünschte ergebnis auf einen blick hat. die Farboptionen im Ranking finde ich als extrem unwichtig und sollten dafür raus genommen werden.

Antwort Gefällt mir

DrQuendt

Administrator

20 Kommentare 27 Likes

Pro Lüfter kann man das schon sehen, zumindestens mit 2 Y-Achsen z.B. wenn du bei den Detailansichten runterscrollst.

Oder meinst du X-Y statt mit 2 Y-Achsen?

Antwort Gefällt mir

Klicke zum Ausklappem
Igor Wallossek

1

10,408 Kommentare 19,386 Likes

Solche vergleichenden Metriken kann man NUR in der Einzelansicht machen, das wird wirklich filigraner Overflow, wenn man das ins Vergleichfenster presst.

View image at the forums

Antwort Gefällt mir

Igor Wallossek

1

10,408 Kommentare 19,386 Likes

Hat zufällig jemand Zeit und Lust ein Tutorial zu verfassen? :)

Antwort Gefällt mir

RX480

Urgestein

1,885 Kommentare 874 Likes

Jo, jetzt seh ichs, dort gibts auch 500rpm und das Diagramm.

Antwort Gefällt mir

Martin Gut

Urgestein

7,896 Kommentare 3,660 Likes

Super. Wieder ein riesiger Schritt weiter. Vielen Dank für eure grosse Arbeit. Da versteht man, dass so etwas länger dauert. :cool:

Dem kann ich mich nur anschliessen. Bei welcher Drehzahl ein Lüfter eine Leistung bringt, ist mir relativ egal. Interessant ist, wie laut es bei welchem Luftdurchsatz ist.

Antwort Gefällt mir

Igor Wallossek

1

10,408 Kommentare 19,386 Likes

Geht ja :)

View image at the forums

Antwort Gefällt mir

RaptorTP

Veteran

327 Kommentare 158 Likes

Holy Moly ! ... wie geil ist das denn ...

Ein Kurztest hat gezeigt das zB bei der Gegenüberstellung hier und da noch die Angabe der Größe im Dropdown-Menü fehlt (120/140mm)

Antwort Gefällt mir

Megaone

Urgestein

1,801 Kommentare 1,705 Likes
Opa-Chris

Mitglied

87 Kommentare 122 Likes

Hallo @Igor Wallossek,
hallo @Pascal TM-Custom/Igorslab,

was ihr da geliefert habt, ist einzigartig!
Auch wenn ich hier schon wieder die ersten Kritiken lese, dass man ja doch nicht alles sofort auf einen Blick auf der ersten Seite geliefert bekommt und man bloß nix mehr machen muss....:
Ihr habt etwas geschaffen, dass es in der Form noch nicht gab und schon super Vergleiche ermöglicht! Der 1:1 Vergleich ist einfach nur geil!

Ein dickes Danke dafür!

PS: Es ist ja erst noch die Betaphase, das sollte man auch nicht vergessen!

Antwort 2 Likes

RaptorTP

Veteran

327 Kommentare 158 Likes

In der BETA Phase sollte Kritik erwünscht sein (konstruktive)

Antwort 3 Likes

Martin Gut

Urgestein

7,896 Kommentare 3,660 Likes

Ich will da auch nicht kritisieren, aber ... . Ganz ehrlich, ich weiss wie viel Arbeit dahinter steckt und schätze das sehr hoch ein. Deshalb finde ich es auch wichtig, dass die Datenbank so weit kommt, dass jeder schnell findet was er braucht.

Die Grafik ist so aber wirklich nicht übersichtlich um verschiedene Lüfter zu vergleichen. Damit ich verschiedene Lüfter vergleichen kann, muss ich jetzt verschiedene Lastfälle annehmen (50 m3/h, 80 m3/h, 120 m3/h). Dann muss ich in der Grafik von der rechten Skala nach links bis zur blauen Linie. Dort sehe ich wie schnell der Lüfter dafür dreht, was mich aber nicht interessiert. Von dort muss ich also nach unten/oben zur gelben Linie und dann auf der linken Skala schauen wie laut der Lüfter dabei ist. Das muss ich dann für verschiedene Lastfälle und Lüfter machen bis ich vergleichbare Daten bekomme.

Man programmiert manchmal zu gerne das, was gerade gut machbar ist. Dabei verliert man aber manchmal etwas das Ziel aus den Augen eine Darstellung zu erreichen die wirklich übersichtlich ist. Es ist wirklich nicht böse gemeint, aber so werde ich die Grafiken kaum verwenden. Die Daten sind zwar alle da, aber für normale Nutzer die mal schnell hinein schauen kommt so nicht viel verwertbares heraus.

Antwort Gefällt mir

Klicke zum Ausklappem
echolot

Urgestein

1,069 Kommentare 821 Likes

Dinge die die Welt definitiv braucht. Vielen Dank dafür!
Hab gleich mal was Komisches gefunden. Kann das sein oder sind die so gut?

View image at the forums

Antwort Gefällt mir

Pascal TM-Custom

Urgestein

1,122 Kommentare 1,361 Likes

die sind wirklich so gut der Lian Li ist auch etwas Dicker und das tut manchen Lüftern schon gut.

Du siehst auch bei denn anderen Lüftern das von 500 auf 1000 es sich fast immer verdoppelt an Luftstrom

Antwort Gefällt mir

Danke für die Spende



Du fandest, der Beitrag war interessant und möchtest uns unterstützen? Klasse!

Hier erfährst Du, wie: Hier spenden.

Hier kannst Du per PayPal spenden.

About the author

Igor Wallossek

Editor-in-chief and name-giver of igor'sLAB as the content successor of Tom's Hardware Germany, whose license was returned in June 2019 in order to better meet the qualitative demands of web content and challenges of new media such as YouTube with its own channel.

Computer nerd since 1983, audio freak since 1979 and pretty much open to anything with a plug or battery for over 50 years.

Follow Igor:
YouTube Facebook Instagram Twitter

Werbung

Werbung