Pushing the Limits, Part One

Pushing the Limits, Part One

Written by J.I. Agnew

Back in Issue 122 and Issue 123, I had gone into the concept of loudness in the articles “How Loud Is a Record” and “Eight Decades of Wrong Assumptions: the Loudness Wars.” Today, I am going to go into a somewhat related topic, which is that of signal levels. This topic is complicated enough to require an entire book dedicated to it, leather-bound and letterpress printed, to really do it justice. As such, I will narrow it down to a discussion of signal levels on popular music distribution formats and will entirely skip the more involved concepts such as LUFS (Loudness Units Full Scale) and so on.

When recording audio signals onto any medium, metering and describing signal levels is a very important part of the process. If done properly at the production end and assuming that properly-engineered playback equipment is used, and the equipment used at the listening end is compatible with each other (some bold assumptions to be making in this sector), the average consumer may never need to worry about or have any understanding of signal levels. The serious listener, however, or those who may be using professional audio equipment at home, may wish to understand how signal levels apply to their enjoyment of recorded music.

Sounds in real life are a localized modulation of the atmospheric pressure caused by acoustic disturbances. The lower limit of signal levels in acoustic phenomena has been defined as the average threshold of audibility of the human auditory system, described as 0 dBSL, which corresponds to a pressure of 20 µPa (micropascals). This is our reference for sound pressure level (SPL). Leaves in the wind in an otherwise quiet surrounding read approximately 20 dB SPL, an airplane taking off would register around 120 dB SPL, and go on.

But is there an upper limit? Not in any real natural terms. Sound can get loud enough to cause pain, discomfort, serious injury, permanent damage, or even death. Sonic weapons have been in development for a long while and the numbers in the dB SPL scale will far exceed the levels at which anyone reasonable would listen to music. As such, the dB SPL scale has a lower limit but no upper limit. Technically, even the lower limit is not strict, as the numbers can be negative, but it probably won’t be audible anymore. The dynamic range of sound possible in terms of physics and acoustics exceeds the capabilities of the human auditory system and far exceeds the capabilities of sound recording and reproduction.


Basking in the glow of VU meters: a familiar sight to probably every Copper reader. Here's a TASCAM 122 MK II and 122 MK III, courtesy of Agnew Analog Reference Instruments.


Once sound has been converted to electricity after being picked up by a microphone, it becomes an AC signal. A common reference in the electrical realm is the dBu, with 0 dBu corresponding to 0.775 VRMS (volts RMS). This is neither an upper nor a lower limit. In the electrical world, signals can go much higher up than 0 dBu (or +4 dBu, which is a common studio reference), and also much lower. The audio signals can become so low as to become entirely buried within the inherent noise floor of the electronic circuit they are passing through. At that point, the audio signals are no longer recoverable, so this becomes a lower limit. However, each electronic circuit has a different noise floor, so this lower limit is left undefined. It will depend on the equipment being used. Ideally, the equipment used at the recording and mastering stage will have a much lower noise floor than the equipment used by the consumer at home, so the ultimate noise floor would be limited by your choice of playback equipment and not by the recording itself. Ideally, that is, in very much the same way that people holding any kind of public office should ideally not be corrupt.

The upper limit is also not defined. Essentially, one could keep on increasing the electrical signal level up until the electronic circuit starts to clip the signal. But even this is sometimes done on purpose to enhance the sound in the process of music production, with the most popular example being guitar amplifiers.

Simplified illustration of signal clipping. When an amplifier (or device) can't pass the signal without distortion, clipping occurs, so named because it looks like the peaks of the waveforms are cut off. Courtesy of Wikimedia Commons/Gutten på Hemsen.

One could increase signal levels even further, to the point that the electronic circuit is entirely destroyed. That would be a pretty hard limit. However, each electronic circuit will exhibit a different clipping point and a different damage threshold. As such, an upper point cannot be easily defined. We can go much higher and much lower than the electrical dBu reference, which is just a reference for convenience of measurement.

The electrical signal, with only a nominal level and undefined upper and lower limits, will subsequently need to be recorded onto a convenient storage medium.

This is where things get interesting. Storage is no longer part of the electrical realm. Signal levels can no longer be described using an electrical reference.

First came the mechanical grooved disk (aka the shellac, then the vinyl record of today), where signal amplitude is proportional to stylus velocity. The reference there is essentially miles per hour, but arranged in metric measurements and the more convenient format of cm/s (centimeters per second). The correct SI units would have been m/s, but the numbers would look daunting. The most common reference was established by the NAB (National Association of Broadcasters) and was a peak lateral velocity of 7 cm/s at 1 kHz, which equals an RMS lateral velocity of 5 cm/s, again at 1 kHz.

As with dBu, this is neither a maximum nor a minimum. The minimum would be the point where the signal disappears under the noise floor of the medium, which, in the absence of serious defects, is the random signal generated by the playback stylus tracing the individual molecules of the material the record is made of. Different materials and different stylus geometries have different noise floors. To further complicate matters, the noise floor is not linear with frequency. The lower limit, therefore, remains undefined.

The upper limit is also not linear with frequency, which means that at different frequencies, the maximum level capability is different. As with the lower limit, there is no hard upper limit on disk records. A limit can be reached in one of several ways.

If the duration of the recording on one side of a record is long, the grooves need to be packed closer together (recording pitch), and especially at the lower frequencies, one groove could even have the possibility of cutting into the neighboring groove. Even if the duration of the material permits a wide spacing of the grooves, very high levels at low frequencies could cause the playback stylus to jump out of the groove (which is technically a playback limitation, as the cutting stylus and cutter head will have successfully recorded that information onto the disk). At high frequencies, velocities could get so extreme as to render it impossible for the playback stylus to trace the groove, either due to geometric constraints or to inertia, or both. Extreme signals at low frequencies could mechanically damage the cutter head and at very high levels at high frequencies, the drive coils can melt due to excessive temperature being reached.


The FloKaSon Reference Series TestDisk-001. Courtesy of FloKaSon.


Between 1 to 3 kHz, where our hearing is most sensitive, (please refer to the Fletcher-Munson equal loudness contours for further reading on this phenomenon) the usable range between the noise floor and the loudest signal that can be recorded and successfully played back can reach 120 dB. At low and high frequencies this figure is reduced.

In the next episode, we will look at the minimum and maximum of signal levels encountered in magnetic tape recording.


Header image: Waves Audio VU meter plug-in. Courtesy of Waves Audio.

Back to Copper home page

1 of 2