New downblower: Noctua NH-C14S in the test

The typical tower design for CPU coolers is actually rather unpopular with mainboard manufacturers. The exhaust air on motherboard components described in many design guides from AMD or Intel is missing. Therefore, among the well-read users there are always preferences for models with top / downblowers. Noctua pays tribute to this group with its new NH-C14S. Our test clarifies what the premium manufacturer of CPU coolers has created here.

Intro

 

With a top or downblower, the fan sits on top of the cooler and presses its air through the fins towards the mainboard. This is actually the preferred approach for mainboard developers, as the exhaust air can also cool components of the mainboard's power supply at the same time.

In the meantime, cooler development has clearly moved in the direction of tower cooling, and the top models in this series practically no longer provide for such mainboard cooling. With top mainboards in the price range from around 150 euros and higher, the manufacturers counteract the development with better components. Typical, heat-sensitive electrolytic capacitors (electrolytic capacitors) are usually still used on cheaper circuit boards, which can benefit from a downblower.

With its new NH-C14S, Noctua also wants to serve this user warehouse, which attaches great importance to this fact. Noctua's focus is of course still primarily on good CPU cooling and, with the NH-C14S, also on improved compatibility with motherboards. Our test clarifies what to expect from the new Noctua cooler.

Bookmark (CPU) cooler: Know-how:

• Cooler structure (radiators, heat pipes etc.)
• Cooler designs and special features
• Fans and bearings
• Thermal compound (backgrounds and order)
• especialities

Test environment

Hardware

Test candidates:

• Noctua NH-C14S (Amazon offers / Caseking offers)

Reference cooler being tested

• Arctic Freezer i30 (HT4U test / Find it on Amazon*)
• Alpenföhn Himalayas (HT4U test / Find it on Amazon*)
• Alpenföhn Matterhorn (HT4U test / Find it on Amazon*)
• beQuiet Dark Rock Advanced (HT4U test / Find it on Amazon*)
• Enermax ETS-T40-TA (HT4U test / Find it on Amazon*)
• Noctua NH-D15 (HT4U test / Find it on Amazon*)
• Noctua NH-U14S (HT4U test / Find it on Amazon*)
• Noctua NH-U12S (HT4U test / Find it on Amazon*)
• Noctua NH-U9S (HT4U test / Amazon offers / Caseking offers)
• Noctua NH-D9L (HT4U test / Amazon offers / Caseking offers)
• NZXT Kraken X31 (HT4U test / Find it on Amazon*)
• Prolimatech Super Mega (HT4U-Test / Find on Amazon)
• Raijintek Ereboss (HT4U test / Find it on Amazon*)
• Scythe Mugen Max (HT4U test / Find it on Amazon*)
• Scythe Ashura (HT4U test / Find it on Amazon*)
• Scythe Grand Kama Cross 2 (HT4U test / Find it on Amazon*)
• Scythe Mugen 3 (HT4U test / Find it on Amazon*)
• Thermaltake Frio Extreme (HT4U test / Find it on Amazon*)
• Thermaltake Water 3.0 Ultimate (HT4U test / Find it on Amazon*)
• Thermalright Macho 90 (HT4U test / Find it on Amazon*)
• Thermalright True Spirit 140 (HT4U test / Find it on Amazon*)
• Phanteks PH-TC14PE (HT4U test / Find it on Amazon*)
• Xigmatek Dark Knight Frostbourne Edition (HT4U test / Find it on Amazon*)

We recently had our new test stand in ours CPU cooler reboot article presents. To put the new test stand to the test, we used a total of 14 coolers of different designs. There are smaller as well as larger models, younger and older models. But the old representatives are still available on the market and can prove how the cooler segment has developed in recent years.

Of course: you will always miss one or the other favorite. Unfortunately, we cannot eliminate this fact in our consideration. As soon as options and rooms are available, we complete the comparison list. But the selected portfolio should actually be able to offer everyone a basis for assessment.

Hardware components:

• Processor: Intel Core i5 4670K* @ 4,3 GHz @ 1,3 volts (turbo mode: on)
• Motherboard: MSI Z87-GD65 Gaming* (Intel Z87 chipset) - BIOS 1.5
• Memory: 8 gigabytes (2 x 4 GB) Kingston HyperX Gray DDR3-1600* CL9 at 1,60V
• Graphic card: MSI GeForce GTX 660 Ti PE*

The choice of the above components has been made with care. Besides the fact that the Gaming series MSI board When it comes to an enthusiast mainboard that leaves little to be desired on the feature list, there were two other main reasons for our decision. On the one hand, after various test runs, it showed that it did exactly what we asked it to do - be it voltages and clock rates or sufficient space in the area of ​​the socket and the distance to the PCI Express slot so that expansive coolers can also be used. We will come to the keyword socket keepout area later in the test.

We left the fan control curve - preset in the BIOS - in the delivery state. This applies to the speed behavior and the volume shown later.

The choice of the less spectacular Kingston DDR3 memory was also made for two reasons. On the one hand, the test phases showed that it works absolutely stable with the motherboard, on the other hand, the memory comes with flat heat spreaders, so that we run less risk of a bulging CPU cooler colliding with bulging heat spreaders. Oh yes: Visually, Kingston's Gray series also fitted a lot better.

For the actual heart of the test, we chose the aforementioned Core i5 processor, which is still in an affordable price range and is therefore likely to be more widely used than the more expensive i7 models or even those for the Socket 2011 platform.

Contrary to some instructions on the net regarding overclocking the new Haswell processors, we took a different, simpler path. We overclock the CPU to a maximum at the turbo clock and increase the multiplier there. So that sufficient temperature development is guaranteed and the clock rates are kept stable, we increase the pure core voltage to 1,3 volts and at the same time set the power limits (long duration and slow duration) to 155 and 165 watts, respectively. In this way, we ensure that the turbo cycle does not drop when there is sufficient cooling, and at the same time that it is reduced when the cooler is no longer powerful enough. Rather, with this approach, we ensure that we produce a power consumption of more than 125 watts and thus of course reach corresponding temperature regions in which the test coolers can prove themselves.

Other hardware:

• Power adapter: be quiet! Straight power 680 watts
• Hard disk: Seagate Barracuda 7200.10 SATA 500 GB
• Drive: Samsung USB DVD Burner Model SE-S084
• Housing: Cooler Master LAB (Benchtable)

One last word applies to our open test stand. Since there is practically no PC case that could be representative of the home user in any way, we rely on an open test stand. Depending on the housing used at home, this can be an advantage or a disadvantage. With well thought-out case ventilation, some coolers should show themselves better in terms of noise behavior, in average concepts probably on the level of the open test stand and in poorly ventilated cases with clear disadvantages. But that, in turn, is all dependent on many factors, and since this is so, we see a reasonable and reproducible way of doing these tests with our test stand.

However, we can currently attest that with the hardware used, the temperatures are usually driven into those areas where the fans have to reach their limits. Only carefully ventilated cases could optimize the cooling behavior here.

Software and test procedures

• • Windows 7 32-Bit - SP1 including all updates up to February 2014
  • NVIDIA GeForce / ION driver version 314.07 WHQL
  • DirectX as of June 2010

tools

• AIDA Extreme
• Core2MaxPerf
• CoreTemp
• LinX
• TES V: Skyrim

In addition to major changes in the hardware, we have also made fundamental adjustments to the software and the determination of the CPU core temperatures.

Load conditions:

We have now decided to work with a total of four instead of three scenarios, which - apart from load-free operation - represent worst-case scenarios for the CPU.

The applied load times are 15 minutes in idle and in LinX, and the last 60 seconds are averaged over all values. The load pickups with Skyrim and Core2MaxPerf are held for 30 minutes and the last five minutes are averaged. To do this, we log the temperatures completely over the running time via AIDA Extreme and at the end only give a temperature of the four processor cores. Under Idle, C2MP and Skyrim we give the averaged value of the recorded average temperatures, under LinX the average value of the maximum temperature values.

Why four scenarios?

The Idle value is obvious. If the PC has nothing to do or if it is only lightly loaded with office applications or surfing, it is less important to know how warm the processor will be, but rather how loud the cooler's fan is. We cover this area with this recording, which is taken 15 minutes after the system start.

This represents another category of daily PC use 3D games A large part of the younger generation likes to use the PC as a basis for gaming, since there potentially better picture quality can be offered compared to consoles. But at this point you have to differentiate between the applications. There are games that are very graphics card-heavy and demand very little from the processor. However, there are also games that paint the opposite picture and those that require both components in the course of the gameplay. To do this, we set out again to get hold of a suitable worst-case scenario. By this one should understand that this is only the worst case of the cases, which is permanently not to be found in the game. For this purpose, we have chosen a position in Skyrim where we can generate a lot of load on all four CPU cores and which runs through over a period of 30 minutes.

This is followed by the use of Core2MaxPerf as the next level of difficulty. Our measurements have shown that this synthetic load tool behaves in roughly the same way as when you ask the PC to Videos using handbrake to convert. From our point of view, this represents another common area of ​​everyday PC life and is more or less perfectly emulated by this procedure. There are actually no harder applications in everyday life today.

Core2MaxPerf and Coretemp in actionFinally, there are still special programs from the scientific fieldwhich can use the main processor of a system powerfully and excessively; even higher than the previously mentioned applications. One such case is LinX, which performs such calculations as an example. In our comparison, LinX represents the worst case for a CPU - and actually the best explanation for why you might need a very powerful CPU cooler.

A bit of office processing, surfing the Internet and reading e-mails also fall out of a meaningful consideration, as one moves in these areas practically on load-free operation for processors, which with their sophisticated energy saving measures usually work at the lowest clock frequency .

Benchmark scene SkyrimHowever, there is certainly another area of ​​application that many users indulge in every day: Computer Games. Usually it is precisely this area - the gaming scene - that manufacturers of accessories from the cooler sector want to address in particular. We have therefore also selected a game title for today's test, which is supposed to cover this practical test as an example. We chose a representative through the previous selection who loads the processor sufficiently without being completely CPU-heavy. A large number of current game titles should not demand the (four-core) processor as much as we do in our test.

Note: The results of the newly selected test scene in Skyrim are of course not comparable with the results of the earlier tests.

Fan control:

All manufacturers offer fan controls for the fan connections on today's motherboards. However, they usually have nothing in common with each other, and so the fan control on our MSI mainboard is certainly different from that of ASUS, ASRock or Gigabyte, for example. There are no specifications from the CPU manufacturer, and therefore each manufacturer cooks its own soup.

Modern fans are primarily only equipped with 4-pin PWM fans. In a few cases, however, you come across 3-pin models that do not support pulse width modulation. While earlier motherboards offered the option of being able to regulate 3-pin models, we had to find out with the MSI Z87-GD65 Gaming that apparently only 4-pin fans are regulated here.

The control curve is kept very flat up to 40 ° C and then rises steeply in the range between 40 and 70 ° C. From a CPU temperature of 70 ° C, the control system tries to counteract these temperatures by means of maximum speed. In view of the variety of coolers on the market, we think the settings made make sense. In individual cases manual interventions - adapted to your own case cooling system and the CPU cooler - could have a positive effect on the noise behavior. Of course, our test cannot cover these situations.

temperature fluctuations

Our tests are carried out at a room temperature of 20 to 21 ° C. Of course, there are temperature fluctuations in the immediate area of ​​the test station due to the hardware's own heat generation. Fluctuations in room temperature, even in midsummer, are not to be feared in our measurements because we are working with an air conditioning system.

Hardware: special features

Meters

We like to use high-quality measuring devices in our tests. Volume measuring stations, thermographic cameras, infrared thermometers, clamp ammeters or simply voltage measuring devices (voltmeters) are used.

Depending on the area and purpose, we sometimes rely on well-known manufacturers such as Fluke or Tenma, in other cases also on Conrad's own Voltcraft brand. When it comes to noise emissions, we use special devices from ulteaudiotechnikwhich enable us to carry out sone measurements in addition to dB (A) measurements. Further details on the measurement technology we use can be found here .

• DAAS USB
• Tenma 72-2065A (temperature measuring device)
• Voltcraft DT2L / K (tachometer)
• Voltcraft MS-9160 measuring station
• Tenma 72-6185 (clamp ammeter)
• Thermal imaging camera PCE-TC 3

Recording of the background noise

In addition to the cooling behavior, the background noise from the combination of cooler and fan is an essential criterion for assessing such a product. Regular readers should know that we have specialized in this segment. As a rule, our measurements can hardly be compared with the results of most other publications, because they mostly use inexpensive hand-held measuring devices in the 100 to 200 euro price range, the microphone noise of which starts at around 30 dB (A) . In this respect, it is hardly surprising why quiet values ​​are given as 30 dB (A) in such comparisons, although a standardized 30 dB (A) value is anything but quiet.

In addition, in order to meet or at least approximate the norm, such measurements lack a low-reflection room and all too often information on the distance is missing. A dB (A) value without specifying the distance to the test subject must be understood as a standard value, and this corresponds to a distance of one meter.

Our current device DAASUSB from the house ulteaudiotechnik is in a significantly higher price segment and offers us the opportunity to take significantly more precise measurements.

The calibrated device allows us to take measurements in the dB (A) and sone range, and we give the measurement results, as usual, standardized, which corresponds to a distance of one meter. In addition, the test subjects are measured in a kind of anechoic room.

The spectral analyzes also allow an impression of the spectral distribution of the level of a fan in combination with the radiator of the individual test candidates and show how the combinations behave in the different frequency ranges.

Spectral analysis of a test candidateIn order to be able to quickly see the essential values ​​in the spectral analysis, we have marked the detailed field information here with red arrows. The measured value in dB (A) and sone is shown at the top right and the information on the load condition at the bottom - in the middle. Next to it on the right, we list the test subject's fan speed.

In addition, of course, the actual graph provides clear clues to die-hard connoisseurs. Stochastic signals (random signals as defined in the acoustic range) in the range up to around 700 Hz are less misperceived by most people than in the more sensitive range (around 1.000 to 4.000 Hz). In addition, tonal signals in the range from around 200 Hz upwards can significantly impair subjective perception and be perceived as unpleasant (a so-called whistling tone as an example).

However, experience in this sector in recent years has taught us that larger CPU fans are much more gentle on our hearing in those areas. The age of the booming 40, 60 and 80 mm fans is over. Nowadays, hardly any cooler manufacturer starts developing below a diameter of 92 mm.

The test candidate at a glance

Key data and scope of delivery

We usually have to praise the scope of delivery from the manufacturer Noctua in this chapter. The installation material for the different mainboard sockets is neatly distributed in separate boxes. The instructions are more than detailed, but are only required in rare cases with the most sophisticated installation structure.

Due to the fact of the wide radiator, which sits above the cooling and installation surface, a particularly long screwdriver is required to mount the NH-C14S - the tool is of course also included in the scope of delivery.

Noctua describes itself as a premium manufacturer of CPU coolers, which is evident not only in the processing of its products and in the scope of delivery, but also in the pricing. The NH-C14S is also not a bargain, but falls into the high-priced segment with its purchase price of around 75 euros. In return, the user can be sure that the cooler can be used on the next mainboard with a different socket, because Noctua also regularly offers installation kits for newly released mainboard sockets. So you don't have to invest in a new CPU cooler.

Impressions

Seven nickel-plated heat pipes of the 6 mm standard meander through the fins of the radiator, the dimensions of which have been slightly redesigned compared to the previous model. With the NH-C14, Noctua already has a powerful downblower on the market. The NH-C14S comes with detailed improvements, especially in terms of compatibility with mainboards and memory modules. If the fan is attached to the top of the cooler, memory modules with an overall height (related to their passive cooler) of up to 65 mm can be used.

Due to the changes to the radiator, the fan of the NH-C14S can also be mounted below the radiator, which means that the construction is significantly smaller in height, can be used in smaller housings and is also suitable for use in an HTPC housing under certain circumstances .

The problem-free bracket attachment of the fan allows an installation in suction or blowing form, which should always be compared with the air flow of your own case ventilation. We found the best cooling performance in our test setup when the fan was on top of the radiator - sucking or blowing on the bottom gave almost identical results, but slightly worse than those on the top. With a case that has a fan in the lid, these values ​​can change completely.

In addition, the NH-C14S shows itself in the usual Noctua quality. The heat pipes are soldered and not pressed in the area of ​​the fins and the base plate. The pipes as well as the base plate are nickel-plated, which counteracts signs of aging.

There are two openings in the radiator, which are provided for mounting the cooler. The tool included in the scope of delivery must be inserted here to tighten the screws.

Editing

As usual, the well thought-out and simple installation options of Noctua coolers should be praised. The backplate including its fastening screws is delivered in one piece and does not have to be laboriously assembled.

The plate is pushed through from the back of the mainboard - you just have to pay attention to the correct orientation. Then the plastic spacer sleeves are attached, the metal locking rails are inserted and screwed on with thumbscrews - almost done.

Afterwards, all that remains is to apply the thermal paste to the CPU and to put the cooler on and align it with the screw points, so that it can then be tightened using the tool provided.

Test results

Speed ​​behavior

Particular attention is paid to the speed behavior and thus the PWM control (pulse width modulation). For several years now, the CPU fan has been controlled by the mainboard, depending on the temperature in a "curve" programmed by the manufacturer. Depending on the design, the finally measured temperature or volume differs significantly. There is no specification from either Intel or AMD, so that every motherboard manufacturer can implement their own ideas, which is done diligently. Sometimes changes can only occur from one BIOS version to another, but at least from one model to another.

Many manufacturers have also integrated the option of adapting the PWM control to their own parameters. In this constellation, we typically only use two of the parameters: Fan control deactivated or activated. Some modern motherboards are currently no longer able to control 3-pin fans - this effort has been abandoned in favor of the common 4-pin fans.

The following table gives us an overview of the behavior of the test candidates consulted today (in alphabetical order):

To determine the readout options of the motherboard as well as a Voltcraft DT2L / K digital speedometer used. The latter is used to compare the information provided by the monitoring software and as a control.

When looking at this table, we quickly see that after idle mode, hardly any of the representatives can handle different speeds in the load states that are still present. Our different modes in connection with the test platform and the preset CPU ensure that the maximum fan speed is required there in the majority of cases.

So whether under CPU-demanding games or video encoding or simulated full load: Most coolers need the full fan speed at this point after only 20 to 30 minutes of load.

Noctua NH-C14S

Noctua names the fan speeds with a minimum of 300 RPM and a maximum of 1.500 RPM and gives a tolerance of +/- 10 percent. In idle, we never saw a speed in the range of 300 RPM in connection with our motherboard - the lowest values ​​were around 390 RPM when the fan was attached to the top of the radiator. There were actually never speeds higher than 1.450 RPM - taken as a partial, maximum value.

Attached to the underside - sucking or blowing did not matter - we recorded idle values ​​of at least 440 RPM and maximum load values ​​of 1.510 RPM. However, this is due to the method of attachment and probably the distance between the fan and the cooler block.

Temperature behavior

Idle - dormant desktop

As already indicated elsewhere, here again to clarify: It is basically completely the same whether the cooler cools the four CPU cores at an average of 23 or 30 ° C in idle mode. Far more interesting and important in this operating state is the background noise with which it works. After all, what use are the few degrees Celsius difference if the fan is constantly annoying at high volume?

The Noctua cooler shows the absolutely correct means here by accepting a higher temperature, but then acting practically absolutely silent - in this operating state the NH-C14S cannot be perceived from a closed housing.

Games - worst case

 

The fan control of the mainboard changes this behavior already in the first load state. Noctua turns its fan up almost to maximum speed, shows a reasonable cooling behavior, but works clearly audible.

Video encoding simulated<

In the next load test, the NH-C14S behaves identically to gaming load. We see speeds in the range from 1.430 to 1.450 RPM and a perceptible fan speed behavior, which can be perceived from a closed housing.

Full load - mathematical calculations

We saw the highest fan speeds in the maximum load behavior, which sometimes went up to 1.510 RPM. The cooling performance is on a good level; slightly worse than the Noctua tower cooler NH-U14S, which can be neglected due to the small differences. The background noise changes only insignificantly.

Background noise: sound pressure (dB (A))

Idle - dormant desktop

We have to make it clear that reported values ​​below 9 or 10 dB (A) are clearly too much for our measuring system and that significantly higher measuring tolerances can occur here. And yet the values ​​are a certain indication.

We would also like to note that all values ​​below 15 dB (A) - we call the standardized value at a distance of one meter - can be classified as absolutely quiet. In the majority of cases, a healthy, young, human ear hardly perceives such a value - such a value can no longer be perceived from a closed housing (based on the standard distance).

In the range between 15 and 20 dB (A) you are usually still in the audible segment, but these values ​​also represent a quiet operating behavior. One can speak of clearly perceptible background noise, which subjectively sometimes as loud and sometimes as less is rated loud. And so the around 21 dB (A) of the Phanteks fans are still pleasant, but the almost 28 dB (A) of the Thermaltake model can be classified as a clear source of noise in a quiet system.

Noctua NH-C14S

After the above, it is of course clear that the NH-C14S passes as a whisperer with its results at around 10 dB (A). This background noise can definitely no longer be heard from a closed case.

Games - worst case

And now we come to the point of the evaluation at which the top cooler in the temperature range and equipped with two fans now also take the top positions in terms of volume - to be understood in a negative sense, of course.

In addition, we have representatives in the range between 25 and 30 dB (A). At this point we say “acceptable”, although in most cases the operating noise would be too loud for us to have our own inconspicuous computer, because it can be clearly identified from the system.

Noctua NH-C14S

Noctua's youngest offspring ends up in the region between 25 and 30 dB (A) in terms of its noise behavior, so no noise, but also not really quiet, but perceptible. Subjectively, this is annoying for one ear, but pleasant for the other ear.

Video encoding and full load

And so our comments made earlier remain.

Full load - mathematical calculations

Background noise: Loudness (sone)

We present the sone considerations to complete our results. In doing so, sone would like to present a subjectively clearer value, which we cannot grasp either through spectral analyzes or other facts.

It may be interesting that the test candidates were reloaded for the first sone measurement years later, but their assessment of the loudness was completely different from years before. "Sone" crap! Jokes aside: sone is a quantification based on field trials with humans. An evaluation option was derived from this, which should come close to human perception. The evaluation schemes resulting from such an approach are always subject to a certain change as a function of time. Translated: The aging of people naturally also contributes to the fact that noises are perceived in a different form, perhaps as quieter, perhaps as louder, and so the previous inventory of these people has its corners.

Nevertheless, sone remains a value that is gladly seen today, and that's why we show it too.

Idle - dormant desktop

Games - worst case

Video encoding and full load

Full load - mathematical calculations

Summary

 

Noctua has redesigned its NH-C14S downblower. Slight revisions in various areas enable better compatibility with memory modules, and installation height can also be saved by installing the fan below the cooler. The cooler can therefore also be used in smaller housings.

In addition, we see the usual praiseworthy image of the manufacturer. The hardware is absolutely top-quality, the assembly is well thought out and it can hardly be implemented more simply. There is also the advantage that Noctua reacts to new bases at short notice and offers assembly kits for them - a long-term product from the premium manufacturer.

When it comes to cooling performance, you are at the usual high level and you have to admit to being easily beaten by the tower designs in-house. In return, there is the right cooling for the components of the power supply of the motherboards. Where they are still necessary, a fact that should not be neglected.

The background noise in idle mode on our test board can be described as absolutely silent; Our test candidate only moves in the middle under load, although this must of course always be put in relation to the cooling performance. The price is high, as is customary at Noctua - but this is owed to the company's own quality standards for refined copper components or soldered heat pipes.

[pg], June 10, 2015