Much ado is made of the concept of noise, among both the engineering community and the listenership, but what exactly is meant by the word noise? We can understand the noise of a washing machine or an air conditioner when we are trying to listen to a delicate passage of music, and we will usually insist that all sources of noise be extinguished before any serious listening occurs. But, when it comes to the noise levels of electronic equipment such as amplifiers and preamplifiers, we rarely hear an audible noise level from them before any music is playing. Now, if an air conditioner happens to be so quiet that we are unaware that it is running, why are we okay with that but critical about electronic noise we cannot hear by itself?
We are all aware that audible background noise sources are not only annoying, they are defeating to the cause of fine listening because they mask the important microdetails that help our brains reconstruct the sonic environment and subtle texture of the musical performance. This may seem to suggest that it is enough to fire up the system and see if we can hear a noise floor from the listening position. Since that noise floor is almost never audible, is our concern over noise specifications unfounded? Actually, quite the opposite is the case.
To understand this apparent paradox better, let us look briefly at the problem from a psychoacoustical perspective. When sound is present, the cochleae of the inner ear divide the sound into spectral slices, or frequency bands (critical bands). The brain has independent control over the threshold of audibility (lowest discernible sound level) of each of these spectral slices. This is very much a survival asset, because it allows for sensitive hearing of a very faint sound at one frequency, even in the presence of a louder sound at another frequency. Noise, however, presents a more challenging problem. The noise generated by electronic equipment is spread, often uniformly, across a very broad frequency range that overlays the entire musical content. This means that nothing in the music gets by unscathed; everything is subject to noise that directly overlays the signal, regardless of frequency. Each sound has to compete with noise that is in the same frequency slice as itself. It is a bit like trying to converse in the din of a loud cafeteria. But that is not all. When a background noise level is present, even if it is inaudibly small, the brain ratchets down the sensitivity of the hearing perceptic in all the effected frequency bands. Experts in the field refer to this as a rise in the threshold of audibility. Thus, extremely small noise levels are able to raise the threshold of audibility for sounds that would otherwise be audible, to the point where they are no longer audible. What parts of music are so faint and quiet that they are easily masked by this process? It is the microdetails that we pay so much money to retrieve. With these details intact, we are absorbed by the music and lost in the sense that we are there at the live performance. When these details are masked by a rise in our threshold of audibility, it is as though they were never included in the recording, resulting in an inexplicable boredom and a wandering of the mind.
Noise levels in electronic playback equipment are usually measured by costly test equipment, such as noise level meters and spectrum analyzers. There is, however, a simple test that will help, though not completely, determine if noise is a problem in your system or its components. This test is made possible by a convenient characteristic of electronic noise: The majority of noise energy in preamplifiers and power amplifiers is constant, regardless of the volume level of the playback. Stated another way, the preamplifier and power amplifier noise sources are after the volume control, not before it. This means that when we turn the volume up, the signal level rises but the noise level from the foregoing components remains constant. As you increase the volume from very soft levels, the microdetails are elevated above your own threshold of audibility, i.e., they become audible. You will know that this happened because the sound will become more detailed, lush, and textured. Once this occurs, continue turning up the volume. The sound should become louder without becoming significantly more detailed. If the sound continues to become more detailed with volume, it is a good indication that electronic noise is present in the preamplifier and/or amplifiers, and that it takes a good deal more loudness to overcome this electronic noise. Of course, this experiment must be conducted in an absolutely quiet listening room. Stated alternatively, systems that need to be played loudly to present the microdetails are probably infected with electronic noise. Some systems fail to recover microdetails at all, regardless of volume. It is a credit to any system that it can be played quietly and still retain the details that make listening enjoyable. We are frequently interested in how loudly a system can be played (note the loudness of audio show listening rooms), but seldom inquire as to how softly they can be played before loosing detail.
There is a caveat to the experiment mentioned above. We hear changes in detail when the level of those details changes while the noise floor remains constant. Consider the basic architecture of a playback system, including a front-end source (perhaps a DAC), a preamplifier, and power amplifier. Everything in the chain that precedes the preamplifier (e.g., DAC or other source) is affected by the volume control, both signal and noise. This means that the noise floor of the DAC, for instance, follows the volume of the music signal as determined by the volume control. The intent of the experiment is to change the signal to noise ratio. If the predominant noise source is the power amplifier, the volume control can be used to change the signal to noise ratio because the amplifier noise floor will not change with volume, while the signal level will change. The signal to noise ratio will improve with higher settings of the volume control, and improved microdetail retrieval is indicative of a noisy preamplifier or amplifier. A noisy DAC, on the other hand, will not respond to this experiment because the DAC is before the volume control – both signal and noise will change together with changes in volume. Incidentally, the volume control is at the input of the preamplifier, meaning that the rest of the preamplifier will respond to this experiment.
Most serious listeners have noticed that playback sounds better during the late night/early morning hours. While some have postulated that cleaner AC power is available due to the lower usage of electric machines, etc., and that this cleaner AC explains the better sound we hear in the wee hours, the foregoing suggests a more mundane explanation as well. Since microdetails can be masked by subaudible electronic noise floors, the same should hold true of subaudible acoustic noise floors such as traffic and tire noise. You may think your listening room is perfectly quiet because you don’t hear street and traffic noise, or some buzz saw operating too far down the street to be heard. Yet, this “inaudible” noise can be raising your threshold of audibility, causing the tiny but precious details of your music to fall below this threshold. This is especially true of uniform noise sources such as a fan or HVAC that occupy a wide swath of spectrum. Granted, you could combat this by turning up the volume, but that has its own problems. If subaudible noise can raise the listening threshold of audibility, how much more could an unpleasantly loud volume level do the same? Furthermore, a background noise will do nothing to raise the sound pressure threshold of pain or fatigue. It simply reduces the sound level range into which the entire musical passage must be situated. You can literally find yourself in a position where you are fatigued or pained by a listening level that is high enough to retrieve the small details, spoiling the whole endeavor.
And so, we introduce the concept of optimal sound pressure level (OSPL) for music playback. If playback is below the OSPL, microdetails are lost to equipment noise floors, acoustic noise floors, and even our own absolute threshold of audibility. If playback is above the OSPL, the loud passages will suppress our threshold of audibility, fatigue the listener, or literally shut down the ears, making the simultaneous hearing of microdetails impossible (think rock concert). When playback is nicely situated within the OSPL, the listener will be maximally aware of the greatest dynamic range attainable. When the noise levels of the electronics and the surrounding air space are absolutely minimized, the beautiful simultaneity of fine detail and striking dynamics will make a thrilling and compelling case that you are present in the original recording space. And so, while the old days of audio were focused on power and loudness capability, the modern audiophile should graduate to a deeper level of concern about how softly a system can be played. And, if enough regard is given to this concern, the audiophile will not only enjoy the whole experience more, he or she might have many more years of useful hearing with which to do so.
The information contained herein is presented in good faith and with all hopes of helping the reader to find the best equipment for an audio system. Neither the author, nor Bella Sound LLC, assumes any responsibility or liability due to loss on the part of the reader, owing to or believed to be owing to the content of this article, or resulting from or believed to be resulting from an application of the information contained herein, regardless the level of expertise engaged thereto.
