High-Quality A/D/A Conversion Vs. Analog Wire
In theory, the more bits per sample, the greeter the system’s dynamlc range, meaning the greatest range between silence and loud sounds.
But greater bit depth requires more bits and will consume the available data pipe of a digital system with fewer channels.
For example, with a 100 Mbit pipe, if 25% of the bandwidth were required for protocol overhead, 75% (75,000,000 bits of data per second) could be used for audio. With 24-bit sample sizes at a 96 kHz sample rate, we could fit a maximum of 32 channels (24 bits x 96,000 x 32 channels = 73,728,000 bits) on the connection.
If we dropped the bit depth to 20 bits per sample, we could fit 39 channels, and with 32-bit samples, only 24 channels could fit on the same 100 Mbit pipe.
Sample rate directly affects frequency bandwidth. The faster the sample rate, the higher the maximum possible. Frequency reproduced by the digital audio system and the more accurate, open and transparent the result can sound.
As with bit depth, higher sample rates use up the available data pipe with fewer channels. Setting the system sample rate of 96 kHz will provide half the number of channels of that same system running at 48 kHz: doubling the sample rate requires halving the channel count to keep the total number of samples (and therefore bits of data) constant.
While various new digital audio technologies have developed over the years, copper wire is still coveted by many for its superior sound quality. Both the dynamic range and the bandwidth of analog wire have made it difficult to match in the digital audio world.
However, limitations in other parts of the analog signal chain put the analog vs. digital comparison on a more level playing field. This debate is highlighted when considering digital audio networks because the premise is to replace copper wire with a digital system.
Dynamic Range (Signal-to-Noise Ratio): A good audio cable, especially for short distances, performs very well. An analog signal going down a cable can have noise level below -122 dBu, and it can have an average signal level of +4 dBu with peaks going to +18 dBu or higher.
That adds up to a dynamic range of 140 dB – a superb performance. (Of course analog systems have other circuitry that limits their dynamic range, but since we are discussing the replacement of wire infrastructure with an active digital network system, it seem, an appropriate perspective.)
Contrast that to our best A/D and D/A with dynamic ranges of 115 – 120 dB, and the superior dynamic performance of a good, short, analog cable is obvious. However, since virtually every modern recording or live performance goes through a digital process somewhere in the signal chain, we can readily conclude that 115 dB sounds fantastic when used properly.
Bandwidth: A good audio cable can also have an extremely high frequency bandwidth. In short lengths, high-quality wire rarely is the limiting factor for maximum high frequency; it’s usually the rest of the analog circuitry. In longer lengths, every wire’s capacitive and resistive effects attenuate high frequencies.
In digital systems the maximum high frequency is determined by the sample rate. In theory that frequency is called the Nyquist frequency and is half the sample rate [e.g., for a 48 kHz system, the Nyquist frequency is 24 kHz).
In practical design, the required low-pass filter eats up some of this theoretical bandwidth; hence, 48 kHz systems often have a 20 kHz upper limit specified. Many believe that 96 kHz systems sound better, mere open, transparent, and with better imaging. One thing’s for sure, with a Nyquist frequency of 48 kHz, it is pretty easy to design a system that has high frequency performance that is better than human hearing, i.e., no high frequency roll-off in the human hearing range.