Digital audio is used throughout the world to provide clean, accurate, noise-free sound, and many varieties and standards are currently in use for applications ranging from home theater systems and professional sound studios to satellite and terrestrial transmissions. For broadcast engineers to be able to install, troubleshoot and maintain the digital audio used within their facilities, we need to understand its structure and parameters.
Almost all current professional digital audio in use today conforms to the standards set down by the Audio Engineering Society (AES) and the European Broadcast Union (EBU) in a set of specifications normally referred to as AES/EBU digital audio. The official name of the AES specification is AES3 for the balanced 110ohm version and AES3id for unbalanced 75ohm. The actual signal is made up of square waves at a frequency of 6.144MHz when the sampling frequency is 48kHz. (See Figure 1.) But the AES specifications allow for sampling rates of up to 192kHz, which equals a data rate of 24.576MHz, and while this may seem to be too high to be transmitted over anything other than coax, it can be accomplished.
AES3 data is transported by a series of pulses, and a clock is required to accurately read these pulses; because no separate clock signal is sent, the clock must be recovered from the edges of the pulses themselves. To ensure that there is a data edge to recover the clock from, the data is encoded with biphase-mark encoding; otherwise, a string of ones or zeros would cause the loss of edges and a loss of the clock. Biphase-mark encoding uses a symbol rate (clock rate) that is twice the data rate, so there are two transmission bits for every data bit. Every data bit begins with a clock edge, and when the data bit is one, there is a clock edge in the middle of the data bit. But when the data bit is zero, no transition happens during the data bit. This ensures a clock edge for every data bit and an extra edge for every data value of one. (See Figure 2.) One benefit of using biphase-mark encoding is that it makes the interface polarity independent (you can hook it up either way), and it also removes the DC component from the signal, making recovery much easier.
The data carried by AES3 are arranged in blocks that contain 192 frames, and each frame contains two subframes, one for channel 1 and one for channel 2. Inside each subframe are 32 slots, each of which corresponds roughly to a single bit. These bits carry the audio data as well as other information. Normally, 20 bits are used for the actual audio data, but in the case of 24-bit audio, the auxiliary bits can be used for the extra audio bits. There is a channel-status bit used to indicate whether this is a professional or consumer format and a parity bit for each subframe.
To test digital audio, the first thing to do is measure the amplitude of the signal. AES3, the balanced 110ohm format, can have an amplitude of 2V and 7V, whereas the AES3id unbalanced 75ohm format must have an amplitude of 1V. Next comes jitter, which can cause errors in clock and data recovery and is specified by AES as +/- 20nS. As the square waves travel down the wire, their amplitude is attenuated and the rise and fall times become longer. As this happens, it becomes harder for the AES receiver to find the edges of the signal. To measure jitter, either a numeric or an eye diagram display is used that shows an overlay of the pulses. A clean, sharp display indicates a lack of jitter, while a fuzzy or blurred display means excessive jitter.
In SDI video, return loss is a large factor in designing and building high-quality digital video plants — the same is true in digital audio.
AES3 specifies an impedance of 110ohms with a tolerance of 20 percent or a range of 88ohms-132ohms using shielded, twisted-pair cable with a maximum distance of 100m. Although many different types of cables have been used to interconnect AES3 (some of which vary greatly from 110ohm impedance), it is best to use a high-quality 110ohm cable. It has been found that the majority of problems with mismatched impedances come from short cable runs where the resistance of the cable is too low to attenuate the reflected signal. And because many facility cable runs are relatively short, this is where trouble can begin. On the whole, twisted-pair cabling is very tough and is generally not affected by the deforming effects of cable ties as are coaxial cables; this allows for more flexibility in the installation process and in field use. Generally the worst-case return loss when using proper twisted-pair cable is –19dB, which is still transferring 99 percent of the signal to the receiver. One major advantage the twisted pair has over coax is its ability to reject external noise and interference because of its high common mode rejection ratio (CMRR). Once again, when used in the field where the environment is not controlled, AES3 twisted-pair cabling is highly recommended because any noise or interference will reduce the receiver’s ability to accurately detect the incoming signal.
The discussion on AES/EBU digital audio will continue with the properties of AES3id as well as other aspects and uses of digital audio.