The section is a quiet computing guide to selecting components best suited for your needs. The “recommended/reference” articles are updated periodically as new information comes to light and new products surpass old ones. Please read this Introduction, which covers a lot of ground about the how and why behind our recommendations. Enjoy.
SPCR’s Recommended Quiet / Silent Components is a guide to choosing
PC components, peripherals and accessories suitable for the level of quiet performance
that you seek. The guide is by no means definitive; it is to a growing and living
database open to change. New reviews, strong evidence contradicting original
assessment, and contributor inputs all influence the database. Product availability,
new models and price also have their effects. In short, the recommendations
are updated continuously.
The guide is most useful as an overview and introduction to quiet component
options. You may wish to do more in-depth research as you narrow your search.
Recommended sources are articles and reviews here at SPCR, manufacturers’ web
sites, and other computer hardware web sites, some of which pay at least a little
attention to the issue of noise.
Our recommendations are based on:
Not every one of the recommended components are known by SPCR firsthand. The
ratings for such component are marked with a question mark (?), which indicates
it is subject to confirmation. We want to hear about your direct experience
with these products in reduced noise operation.
Some of you will want to jump right into the recommended lists. The links to those pages are just below. You’ll want to return here to review the rest of our discussion about product assessments.
What does it mean if a product is not on the recommended lists?
The Myth of the Objective Review
Some people like to believe that there is a clearly ordered universe of PC
components with the best on top, and all the others lined up neatly in a sequential
column of diminishing goodness. This is not quite the case, because the emphasis
placed on various characteristics changes a product’s position on any ranked
1) Benchmarks are biased. While sophisticated tests and benchmarks
allow simple comparisons between components, the weight given to various aspects
of performance has a strong impact on the result. The choice of testing benchmarks,
therefore, is itself a built-in bias in every review. Most hardware sites focus
almost exclusively on performance and price. We rank and recommend products
on the basis of noise first, performance second, and price third. Our logic
is simple: noise is our primary concern, and it is a characteristic most often
ignored or glossed over by other review sites.
2) Product application changes ranking. Another aspect of reviews is the question of how a product will be used. This is a very important point when seeking quiet in a world mostly mad about computer performance at any price, about speed, speed, and more speed.
Most high performance CPU heatsinks, for example, are designed for overclockers who want the highest cooling power. They usually have very powerful, loud fans. We would rank such heatsinks low if used as intended. However, when run with a quiet fan at 6V, the monster might be highly efficient at cooling a hot CPU that has not been supercharged and turbocharged to 50% above its normal speed.
3) Low sampling rate is a basic limitation. Even rigorous firsthand reviews can only provide information about one or two particular samples. There are quality and tolerance variations in every product line as well as in the sub-components. Even with very tight tolerances, these variations can lead to significant differences, especially when it comes to noise emissions.
A Few Words about Noise
Assessing noise at very low levels is a difficult challenge. Even sophisticated
scientific testing with expensive instrumentation is not a complete solution
to the challenge of assessing noise. Two components that measure similarly do
not necessarily sound equally noisy.
Neither sound pressure level (dBA – SPL) nor sound power measurements tell
us anything about the quality of the noise, which we can discern
without the aid of any instruments. Is it steady or something more cyclical?
Is it smooth broadband noise or does it have chattering or clicking? Is it lower
pitched or high? What is the frequency balance? How much does the noise vary?
How does it interact with the noises of other components in the PC?
All of these things can and do differ in two noise-making components that measure
at the same SPL. The challenge is even greater when dealing with very low noise
levels (under 20 dBA@1m). The rule of thumb about noise measurements is that
the instrumentation should be capable of reading accurately 10 dBA below the
noise source level, and the testing environment must have an ambient noise level
10 dBA lower than the noise. These requirements dictate highly sophisticated
audio analysis gear in an anechoic chamber. After years of discerning noise
characteristics at very low levels by careful listening, we actually built our
own 10-11 dBA anechoic chamber
in 2008 and equipped it with instrumentation capable of accurate measurement
to below that ambient level.
Noise analysis by listening is naturally subject to factors such as the ambient noise in the testing environment, the noise made by other components, and the sensitivity of the reviewer. Remember, too, a PC that is too noisy for you may be quiet enough for someone who is less sensitive to noise, working in an already noisy environment, or playing games with sound effects at full blast.
How does SPCR listen for PC component noise? In many ways:
One final aspect of noise worthy of note: Human hearing is incredibly selective and adaptive. In a very noisy environment, we are often able to tune out the high background noise and focus selectively on the sounds we want to hear. Someone less adapted to that noisy environment will have trouble hearing anything other than the noise for a while, but then in a matter of minutes, be able to carry on conversation with others. What happens in noisy pubs every night all over the world illustrates this adaptive and selective ability.
Conversely, as noise recedes to very quiet levels, people often become sensitive
to subtle nuances of sound. More than just hearing the sound of a pin dropping
on the floor, we might be able to ascertain whether the head of the pin is plastic
or metal. The closer one’s PC gets to absolute silence, the more acutely one
begins to hear fine differences caused by mounting, different components, or
a minute change in fan speed. It can certainly lead to obsessiveness.
For more detailed discussion about subjective aspects of silent computing,
please see the article, What
is a Silent Computer?
Fans and spinning drives vibrate. The question is how much and whether it results in noise. One simple definition of noise is that it is unwanted airborne vibration. Vibration is not normally provided as a specification by fan makers. Some hard drive manufacturers offer this information. However, even when specified, it is very difficult to correlate vibration with overall perceived noise, because the way vibration adds to noise is interactive.
A fan or drive is normally mounted in a case with screws. The close mechanical coupling allows vibrations from the component to transfer efficiently into the case. If the case is solidly built and the level of component vibration is low, there may be no perceptible increase in noise. On the other hand, a component with some lower frequency vibration may be very quiet suspended in free air, but when bolted in a case, cause a lot of panel vibrations and noise. The panels in a PC case, in effect, become transducers driven by the motors of fans and drives, and act much like loudspeakers.
We take note of vibration levels, especially when they appear higher than usual.
In most cases, decoupled mounting solves the problem. But this solution is not
applicable for all setups.