If you've read the Noise Level FAQ at Neoseeker then you'll be equipped with the standard measurement expectations of the average user. Below is the chart (used with permission) that Neoseeker made to give you an idea of some common noise levels and how you can relate to them:

From the above chart you'd expect a quiet living room to be 20dBa, which means that a CPU fan rated at 21db such as the Thermaltake Silent Boost would be virtually indistinguishable from the ambient noise of a quiet living room. On the other end of the scale, if your cooler were rated at 50dB (which is on the upper end of the cooler noise spectrum) you'd expect it to be about the same volume as an average office.

From our tests in various environments, you'd best take the whole 20dB and 50dB range with a grain of salt.

Why Quiet is Still Annoying

The problem with the chart is not in its accuracy, but in a person's perception of noisiness. Yes, a very quiet office or library is rated at 30-35dB, but if you think that's really quiet computing you're in for a surprise: 35dB from your computer is surprisingly loud.

To figure out what was happening, we measured the decibels at ear level using the HHSL-1 and came up with some surprising results. Even though the HHSL-1 is accurate only from 35dB and up, we found it could measure noise levels far below that, though obviously with a lot less reliability.

Here's what we learned:

1. In a common house, even when there are absolutely no active noise sources, basal noise levels are at least 31.5dB.
2. Even something as simple as a fridge can kick your noise level up to 34.5dB.
3. When things are so quiet that our ears start playing tricks on us, and when the slightest noise sounds intrusive, we were measuring 28.8dB on average.
4. The absolute lowest measurement we could achieve was 26.5, and we could only get that in brief flickers.
5. Two noise sources with the same measured decibels don't necessarily sound as loud as one another
6. Measuring 21dB on consumer level sound meters seems as likely as a dinner and dance with Lucy Liu.

Ok so the first argument is that the HHSL-1 is simply too inaccurate to rely upon below 30dB... what the HHSL-1 measures as 28.8dB (our average in a human perceived noiseless environment) might actually have been closer to 20, 21, or 22 dB. However, given that the Silent Boost is rated at 21dB and yet can be heard by the human ear AND measured by the HHSL-1 as being around 30dB @ 1meter distance, I think we can at least safely say that many manufacturers who have spec'd out sub 25dB CPU coolers would likely measure much louder under our test conditions.

Cooler Measurement Standardization

The problem with measuring anything is the standardization of the measurement. Manufacturers of coolers and fans not only fail to disclose their noise measurement methodology, each manufacturer measures under different conditions and with different instruments.

We feel that a fair measurement of any cooling device is measured at approximately 1 meter from the computer. The computer should be set on the ground, beneath a desk, and the measuring device placed 1 meter away as if a person were sitting above the desk. This is a common set up in a home or office environment and simulates the location of your computer relative to the point where you will be hearing the noise. We encourage silent enthusiasts to measure their own results in the same manner, so as to be comparable with our own measurements and with those of others who can follow this most basic of standards.

Defining Quiet

We've established that a room is often louder than you'd expect from "accepted" decibel charts. We've also established that if a manufacturer's claims are to be accepted, most coolers under 31dB would be virtually noiseless in the average quiet living room, and anything under 29dB should be so quiet that your breathing would be louder to your ears.

Taking all of these things into consideration we can safely define a super quiet computer as being below 30 and even 29dB, as measured by the HHSL-1 at 1 meter from the case and around at ear level.

How do we define NOT quiet? If you find the hum of your fridge annoying, consider our discovery that a humming fridge is around 34.1dB, from 2 meters away. Here are some other common noises that might interest you (all MEASURED using the HHSL-1 and not just based on noise charts):

1. Our CD-ROM spins up at 42.9dB
2. Whispering really DOES fall within the range indicated on charts: we measured 38.9 - 41.0dB
3. Regular conversation is 48-52.5dB, measured.
4. A typical library is bursting with over 43.5dB of constant noise. I've always found studying at the local library to be unnervingly distracting, and now you can see why.
5. At our office, even with 0 conversation, and no computers turned on, measures at 36.5dB, mostly thanks to the HVAC (Heating, Ventilation, Air Condition) built into the ceiling T-bars.

The Impact of Extra Fans, and Case Placement

One thing we were interested in knowing about was the simplest of all noise reduction techniques: the placement of the case.

Everybody knows to place their case on ground to minimize noise - if only because the added distance from your ear would let noise dissipate and lose cohesion. We measured the noise difference between 3 positions around your working area:

1) Beside the desk: 40.2dB

2) Beneath the desk away from wall: 38.4dB

3) Beneath desk, beside wall: 40.7dB

4) On the desk: 42.4dB

As you have guessed, placing the system below the desk protects you the most, but what's REALLY interesting is that putting the system ON the desk versus below the desk is as big a difference as what we measured when a CD-ROM starts to spin up (versus it being idle) or when you unplug a single 25CFM case fan!

Noise: It's all Relative

I actually wanted to compare two systems that I characterized as being totally different in terms of the level of annoyance of their noise. One system measures 35.7dB and the other measures around 39dB... the 4dB difference seems measly, yet the fact that it's accepted that a 10dB increase in volume is perceived as being twice as loud by your ears, the 4dB is significant enough that we found the latter system to be significantly more annoying in a quiet (32-34dB) environment. In fact, during the day, when there's more ambient noise from outside traffic or activity, and general apartment or house noises, the quieter system becomes almost impossible to detect. Likewise, when we're watching movies on our noisy system, the system noise is not noticeable at all. Again, if there are distractions that are greater than the computer, its noise level becomes almost insignificant.

The thing about computer noise is that it's very relative. A simple 3-4dB boost in noise is very aggravating to the average person in a quiet environment. We also found that on average, 2-4dB above ambient is the most we could go before we considered the noise source to be a distraction.

Knowing this, we can suggest the following noise rules for the minimally acceptable computer:

1) In an office, your computer can be be around 38-40dB and still be "acceptable"

2) At home, when everything's quiet, anything above 35dB is getting too loud.

For true silence, you'd have to keep your system at, or below 31.5dB in a quiet home environment, and 35-36dB for a very peaceful office environment.

Keep in mind these guidelines are based on MEASURED noise levels using our OMEGA HHSL-1. Compared to specs claims by manufacturers our measurements will be around 10dB higher, give or take a few dB - so when shopping around for components that have NOT been measured using an actual sound level meter calibrated against real life equipment for reference points, try to look for noise specs BELOW the 25dB mark.

It doesn't end here

While noise and quiet computing measurements can't be completely covered by this simple introductory article, I hope I've given you some perspective at least. We've already seen the relative noise levels measured by an actual Sound Level Meter, and we've compared that to the "expected" noise levels based on commonly accepted dB levels. The bottom line is that here we're hoping to give you an idea of what to expect when we tell you we measured a certain noise level in an article using our sound meter, the HHSL-1 from OMEGA. You can't really compare against other meters and we don't encourage you to.

As a side note, our friends from Neoseeker wanted to let everyone know too that they are using the same sound level meter for all of their objective noise measurements, so you can know that what we've discussed in this article will apply exactly to the measurements they take in their reviews.