To properly measure how quiet are our computers, we tested one of our machines using a professional grade digital sound pressure level (“SPL”) meter. Testing was done with all heating units in the house shut off and all refrigerators stopped. The listening room, with windows closed on a moderately warm January day in Montana with little wind, is 3 miles from the nearest blacktop highway. We are talking here about a VERY quiet sound room, but without special sound absorbing material as might be found in a “test chamber” used by some audio companies.
We chose this kind of practical environment, because no audiophile actually enjoys his/her music in a test chamber. Not only would the music sound stunted in such an environment, but the equipment could not remotely reproduce the experience of hearing a live concert, jazz quartet, or rock band.
The first things to understand about extraneous noise within a good music system or home theater setup are that a) the amplifiers, DACs or pre/pros, and other equipment – without any media computer in the room – generate heat and therefore cause air to move and, thus, make a sound (even though this sound might not be heard by the human ear); b) even the most quiet of listening rooms generates a base-level of noise that is astonishingly high.
In our Montana listening room, as described above, the base level of noise was measured as 33.1 decibels!! Since the lower limit of human hearing is below 20 dba – a very quiet listening room is much louder than “pure silence” as heard by the human ear. These noises are generated by many things, including refrigerators in other rooms, wind outside the house, and of course the breathing of the person conducting the measurements. And the noise constantly fluctuates, by more than a decibel or so, on a more or less continuous basis.
To this base level of noise, a good quality 2-channel audio system on standby adds only about 0.1dba additional noise – see the photo shown next. Taking the photo at precisely the lowest level of fluctuating noise gives us a conservative lower bound to the system’s base level of noise, before booting the computer – about 33.2 dba.
The booting of the computer causes the optical drive to make a noise of about 4dba above the ambient noise level, for less than 0.2 seconds. The noise is so short in duration that there is no time to squeeze the flash on the camera to capture the reading on the SPL meter. [note: This optical drive sound has been eliminated in our latest builds.] Then, the computer on standby, along with the rest of the audio system on standby, generates a noise level of 33.3 dba, when the SPL meter is placed at 1 meter away from the side fan opening of the computer. So, the computer adds about 0.1 dba to the sound system itself on standby.
The noise level out at the listening position (10 feet away) is, again, 33.2 dba – the noise level of the audio system without the computer. The computer is silent by measurement out at the listening position. Only at 1 meter can the noise of the computer be measured, but it can’t be heard even at that close distance.
When the JRiver software is playing a piece of music, but the audio system is placed on Mute, the measured noise at 1-meter is still only 33.3 dba. Even when playing a blu-ray file, the computer, which generates additional heat for such a job, does not cause the SPL meter to rise above this 0.1 additional dba noise level – until the blu-ray file runs long enough (e.g., an hour or so) to cause the computer fan to add another 100-200 rpms or so to the fan speed to control heat build-up. Depending on the length of the movie being shown, and the amount of air the user has provided around the opening(s) in the computer case, this somewhat higher fan speed might add another 0.3 dba to the reading on the SPL meter – the meter might show 33.6 dba.
So, armed with this information, can you live with 0.1 to 0.3 dba of fan noise, measured at 1 meter from the computer? Before you answer, there is one more piece of information you need. All of the measurements taken above were obtained with the user holding his breath. This is because, normal breathing at 1 meter from the SPL meter (and camera) adds about 0.5 dba to the noise measurement. That is, the listener at a distance of 1 meter from the sound meter (even with no audio system in the room) will see about 33.8 dba on the sound meter! The computer playing a blu-ray disc with the fans running at higher speeds at 1 meter shows a reading on the sound meter of 33.6 dba!
The uncharitable way of summarizing all of this is that an audiophile’s insistence on a completely passive computer – one without any fan – must flow from that audiophile’s hearing being sharper than that of a professional sound meter, or the audiophile’s habit of holding his breath while listening to his audio system. The truth, however, is that today’s incredibly silent fans, modified to fit within a very small case, will not be heard by the audiophile at a distance of a meter or even less. Some audiophiles, however, will continue to associate even ultra-low-noise fans with “psycho-acoustics” (Chris Connaker’s famous term for non-measurable noise, or, in the opposite direction, non-measurable acoustic improvement).
But, we can see how the insistence on a fan-less machine arose. The Noctua fans we modify for our machines generate less than half the dba of the next most quiet fan! That is, an 80mm Noctua fan, set up with the lowest range of speeds (less than 1000rpm for the CPU fan) will generate about 7dba in a padded test chamber (according to the company). The next most quiet 80mm fan generates something like 20dba in a similar test chamber. This difference is -- in the world of exponential dba measurement -- quite massive. Our prototyping also discovered that small cases generate less noise than large cases, because the smaller case has less volume to create an echo. The echo doesn’t increase measured sound but it increases the duration of the sound. But small cases present other physical problems – mainly the difficulty in fitting a large size, ultra-slow-speed fan into the small box -- that we have spent much time to overcome.
The 2-channel enthusiast may think that he doesn't need a powerful computer with fans, because he isn't interested in video. But he is missing out on playing ripped blu-ray concerts without such a computer. Indeed, most audiophiles have never in their lives heard a 48 khz/32-bit concert on their beloved stereo system. And only a near-silent Windows computer can deliver such an experience.
Also, we are now in the 21st century, and many high-end sound systems – systems costing many tens of thousands of dollars – are built within a Home Theater set-up. For these owners, the acknowledged leader in Home Theater computers is Kaleidescape™. Their systems start at around $11,000 or so. Our computers produce better sound with lower noise than Kaleidescape systems and are far more flexible in terms of the sources of media. And we can prove it – by looking at the measurements of resampling rates that are shown BOTH in the JRiver software as the computer file is being bit-streamed AND on the panels of the high-end DACS and pre/pros which reside in the best Home Theater setups.
Best of all, our prices are so much more reasonable than Kaleidescape prices that the Home Theater builder can be much more inventive in the way in which the Home Theater system is designed and the way in which multiple-room connections are made. See our discussions of this topic elsewhere on our website.
The listening room is a dedicated, 2-channel listening room with Wilson Audio MAXX2 speakers, Krell monoblocks (650Mc’s), Krell HCT pre-amp, Berkeley Alpha DAC, and Marantz SA11S2 SACD/CD player. Speaker cables are Transparent Reference XL ; interconnects are XLR by MIT. Power conditioner is by Tice. Sound Pressure Level meter is a NADY digital DSM-1 mounted on a tripod at a distance of 1-meter from the computer. Sound meter settings are LO (30-100DBA) and Fast (to capture variable sound peaks). A flash digital camera is used to record the sound at its lowest level for the base measurement and at a modal level for the computer.