Articles: Cooling

Bookmark and Share


Table of Contents

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 ]

Most contemporary cooling systems for CPUs that are worth considering as possible replacements for default (read “boxed”) coolers come equipped with 120x120x25 mm fans. It is becoming more common lately to equip them with even larger fans measuring 140x140x25 mm. Moreover, some cooling solutions makers do not even bother to bundle their heatsinks with any fans at all making this important choice users’ responsibility. It is even more acute to find powerful and quiet fans for the system cases, because very often case makers (even eminent ones) rig them up with whatever they can find. That is why we feel time has come for a detailed study of available 120/140 mm fans.

For that purpose we collected 57 fans of 25 different series from 15 manufacturers. Despite a large number of fans, this roundup doesn’t cover the entire existing variety available in the today’s market and we hope that from now on we will be able to offer you new fan roundups on a more regular basis. Since our first roundup called 11 Fans for Two Super-Coolers and One System Case we have developed a completely new testing methodology. We do not claim that it is the only correct approach to fans testing, but we are going to improve it in our upcoming reviews and hopefully with the help of your feedback as well. So, let’s start with the description of our testing methodology and equipment.

Fans Testing Methodology and Equipment

The primary fan testing was performed outside the system case without any cooling heatsinks involved. We decided to do it this way for two reasons. First, we had to exclude the effect of the noise generated by the system case on the results of our acoustic measurements. No matter how quite your system case and its components are, you can’t underestimate the effect they have on the ambient acoustics. Second, testing fans on top of heatsinks does have certain practical value and may be an important addition to the main test session, but you won’t be able to get exact results in this case. Especially, since some results obtained during our test session turned out pretty close to one another and there is no way to catch this difference using the temperature of the central processor hard drive or liquid circulating inside the cooling system contour.

During our test session we used a controller of our own proprietary design. This controller works independently and doesn’t need to be connected to a PC. It doesn’t generate any noise, and most importantly, allows controlling fans of all types – with three-pin as well as four-pin connectors.


Three-pin fans are regulated through voltage that can vary from 0 to 12 V with 0.3 V increment obtained with a liner stabilizer. Four-pin fans receive nominal voltage of 12 V and their rotation speed is regulated with a separate PWM signal with standard frequency of 25 kHz. In the latter case fan rotation speed is changed from 0 to 100% with 2.5% increment.

The fans are connected via low-resistance shunt, which allows us to measure the current. The maximum fan current that our device can handle is 1 A with 1 mA increment. The device was calibrated before the tests using reference load, so the measuring error for the voltage and current doesn’t exceed 1%. The unit is based on Atmel ATmega168 microcontroller. For current measurements it uses its own ADC together with AD8605ARTZ operational amplifier that amplifies the shunt signal. For fan rotation speed control via voltage it uses Analog Devices AD5245BRJ10 digital potentiometer and voltage regulator built with National Semiconductor LM7301IM5 operational amplifier and IRF IRL3502 field-effect transistor. In case of four-pin fans the PWM-signal is generated using built-in timer of the microcontroller.

Moreover, this device also measures fans rotation speed using the signal from its own diode that generates two impulses for each fan rotation. The operational mode (voltage or PWM control), voltage setting or relative PWM duration, as well as the measured current and rotation speed are displayed on a small LCD screen in real time.

During the tests the voltage of each fan changed from 3 V to 12 V with 0.9-1.2 V increment. In our opinion, it didn’t make sense to lower the voltage below 3 V, because not all the fans could work fine even at 5 V voltage setting. We recorded maximum current for each voltage setting and fan rotation speed. Moreover, we also determined the startup voltage for each tested fan (several start/stop cycles). Don’t be surprised if the operational voltage is lower than startup voltage: it means that once the fan blades reach stable rotation speed the voltage may be lowered and the fan will just keep on working fine.

Besides, for PWM-controlled fans we also determined the dependence of the rotation speed on the voltage for two reasons: to maintain unified testing methodology and because the tests performed at low voltage are considered heavier than tests of PWM-control.

Our second testing device was a thermal anemometer – a device used for measuring airflow speed and temperature. We used Pro’s Kit MT-4005 thermal anemometer with 0.01 m/s or 1 ft/min precision:

Although our anemometer only registers the speed of the airflow passing through its fan (which is very light and creates almost no resistance), we can use the size of this fan to calculate  the airflow in more common units used by most fan makers - cubic feet per minute (CFM).

To ensure higher measurement precision, namely, to make the entire airflow generated by the fan go through the anemometer blades we had to be creative. We took a 5-liter plastic bottle and removed its bottom and neck. We cut off the bottle neck so that it matched the anemometer frame exactly, then attached the anemometer to it and sealed the contact area with insulating tape. The bottom of the bottle was removed to accommodate the tested fans. For that purpose we made a special polyurethane foam ring with a square cutout measuring 115x155 mm. The diameter of the ring was 3 mm bigger than that of the bottle. During the tests we inserted a fan into the square opening in the polyurethane foam and then inserted the ring with the fan into the bottom of the bottle:


As a result, this device was very airtight from the sides and had an admission opening and a discharge outlet. For example, if we inserted the fan into the bottle without the polyurethane foam ring, the airflow speed was only half as high. The distance between the anemometer fan blades and the tested fan was 190 mm. Things were a little more complicated with 140 mm fans because they were too big and didn’t fit into the bottom of the bottle. Therefore, we simply pressed them against the bottom part of the bottle and sealed everything with sticky tape. The only exception was Scythe Kaze Mary fan that fit into the bottle together with the foam ring, just like all 120 mm fans.

The whole thing was placed at the edge of the computer desk to ensure free airflow to the fan almost over its entire diameter:


Before we took any of the readings down each fan worked for about 5 minutes to warm up the bearings and make sure the characteristics have stabilized. We took the maximum airflow readings for each fan voltage for the performance diagrams.

The noise level of each fan was measured between 1:00AM and 3:00AM in a closed room about 20 m2 big using CENTER-321 electronic noise meter. The noise meter was always placed along the same axis as the fan rotor at an exact same spot. To ensure precision we set special marks for the noise meter as well as fan placement. The distance between the noise meter receiver and the fan equaled 350 mm:

As you can see, we set the fan on special stand made of polyurethane foam (with placement markings as well). We had to give up shock absorbing retention and reduce the distance between the fan and the noise meter, because CENTER-321 registers 30 dBA as the minimum reading. And since some fan models generate less noise at minimum and even medium rotation speeds, we had to resort to this specific measuring technique. We will provide not only the actual measurement results, but also our subjective opinion on the noise level of each particular fan model.

We are going to provide an individual graph with test results for each fan. The vertical axis on the left stands for the fan airflow and noise level, the one on the right shows fan power consumption in watts. The horizontal axis indicates fan rotation speed. All results are summed up in a table below the graph.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 ]


Comments currently: 13
Discussion started: 06/23/09 10:35:38 AM
Latest comment: 11/29/10 08:58:10 PM

View comments

Add your Comment