Skip to main content

Unlocking the Mystery Of AES3 Digital Audio, Part 2

The AES3 signal is described by such terminology as linear pulse code modulation, linear two's complement binary, serial digital bitstream, biphase mark format, and time division multiplexing. Last time we discussed the first two terms, this time we'll tackle some more.

In a serial digital bitstream, bits are transmitted sequentially, one bit at a time. This simplifies connections as only one signal line is needed rather than multiple wires and multipin connectors as for parallel connections, where a number of bits are transmitted simultaneously. AES3 specifies shielded twisted pair 110-ohm cable, and AES-3id calls for a coaxial cable.

When bits are transmitted in sequence, there has to be a way for a receiver to recognize the value of each bit, know where each digital word (audio sample) starts and ends, and where to find and how to interpret any other types of bits in the stream. Also since AES3 can carry two channels of digital audio, there has to be a way of transmitting them, with minimal delay between the two, and distinguishing them at the receiver.


Fig. 1: Graphical representation of the biphase code for digital sample word 11010010, with the most significant bit to the left, and the least significant bit to the right (adapted from AES3-2003). Let's start with how the bits themselves are represented. They could be simply sent as a logic high (like +5 V) to represent a digital one, and a logic low (like zero volts) for a digital zero. But this method could result in a long string of ones or zeros which would make it difficult for a receiver/decoder to distinguish one bit from another, unless a separate clock reference signal is transmitted at the same time. That would end up making the AES signal more complex than it needs to be.

So AES3 specifies a different way of representing individual bits, called biphase mark coding.

Instead of representing the logic value of each bit by an absolute voltage, a bit is represented by a transition from one voltage amplitude to another of the same value, but opposite in polarity (for example, a transition from (for example, a transition from –5 V to +5 V, with the next from +5 V to –5 V).

But how do you determine whether a transition represents a logic zero or one? For a logic zero, the timing of each transition occurs at the bit-rate of the given bitstream. The time between transitions for a logic zero is the inverse of the bit-rate and is called its time slot. A logic one is differentiated by an extra transition in the middle of the time slot.

Let's look at a graphical representation of the biphase mark code of an example sample word 11010010 in Fig. 1. Note that the overall clock rate is twice the bit-rate, to allow for the intermediate transition for logic one.


Biphase mark coding has several advantages. The transitions themselves carry the clock information, obviating the need for a separate clock signal to be carried along with the digital audio signal. The AES3 receiver looks for these transitions, not the voltage values, and where these transitions occur on the timeline, to extract the bit values and timing information.

Fig. 2: Graphical representation of time division multiplexing of channels 1 and 2 in an AES3 digital audio signal (adapted from AES3-2003). It doesn't matter whether a transition goes from positive to negative or vice versa, nor what the starting polarity is. This is why the AES3 signal is polarity insensitive, unlike analog audio.

If using an XLR connector for analog audio, and you swap the wires on pins 2 and 3, you've inverted the polarity on the audio signal. However, for the AES3 signal, if you swap pins 2 and 3, you will invert the polarity of the electrical signal waveform that forms the bits, but not of the audio signal represented by the bits. Again, it's the transitions that matter.

In practice, it's a good idea to be consistent with AES3 wiring as any other signal wiring. Keep to a convention and stick with it. A good suggestion is to use the same one for analog audio. With red and black wires of a twisted pair, for example, connect the red wire to pin 2 and the black wire to pin 3 of the XLR connector, and the shield to pin 1.

Another advantage to biphase coding is that since the signal goes in the positive and negative directions for an equal amount of time, if all is working well, the average or DC value of the electrical signal is zero. AES3, as an AC signal can be coupled with a transformer, as it often is in receivers and in balanced to unbalanced converters.

Since there's no (or minimal) DC component, transmitting power is low, and the cable gauge can be small as well, like AWG 26 as for analog audio.


Now for the time division multiplexing (TDM) aspect of the AES3 signal. TDM provides a method for carrying two digital audio channels in one AES3 bitstream. (TDM in general allows many channels to be carried in single bitstream, but the AES3 standard only specifies two channels—stereo, two independent channels, or mono which can be carried on channel one or repeated on channel two as well.)

The bits that make up an AES3 signal are arranged in a specific order called a frame, and each frame occurs in a given amount of time. To handle the two channels, each AES frame is divided into two subframes (that's the time division part), with one subframe for each digital audio channel. Each frame contains one sample word for channel one, followed by one sample word for channel two. As each frame proceeds in sequence, word samples alternately progress between channels one and two. That's the multiplex part.

Fig. 2 shows a graphical representation of the frame structure. The X, Y, and Z parts are preambles which identify the start of each subframe, and which will be discussed further in another article, as will blocks. Channel 1 indicates one sample word for the first channel, and Channel 2 indicates one sample word for the second channel. VUCP are the validity, user, channel, and parity bits (one bit each) associated with each channel. These bits immediately follow each sample word. More on these later as well.

In this and last month's installment, we briefly explained the key terms that describe the AES3 digital audio signal. Next time we'll look more closely at the subframes.

Mary C. Gruszka is a systems design engineer, project manager, consultant and writer based in the New York metro area. She can be reached via TV Technology.