Digital audio details II

Last time, signal encoding and data structure were covered as well as some information about AES3 — the balanced version of the standard. This tutorial will start with the AES3id unbalanced version and its parameters.

AES3id specifies the physical cabling as being unbalanced 75ohm coax cable using a BNC connector. The better high-frequency capability of coax allows it to carry the AES3id signal at least 10X further than what is possible on AES3 balanced, shielded, twisted-pair cable. Again, return loss should be considered whenever high-frequency cabling is installed to prevent signal distortions that can cause a loss of receiver lock and the subsequent signal dropout. The voltage used with AES3id is only 1V, and reflections can cause more interference with it than with the AES3 2V-7V signal range.

Because the frequencies used are lower in AES digital audio, the impedance of the BNC will not affect the return loss, because even with the highest bandwidth AES/EBU sampling rate, the quarter wavelength is about 7ft. The length of the quarter wave for the frequency being used is very important for return loss, because any disturbance in the impedance of the cable at quarter wavelength distances results in much more return loss and greater signal distortion. This also demonstrates why it is that improper termination and cabling can have adverse effects on the received AES signal, especially on short cable runs. In AES/EBU, this means you can use 50ohm BNCs, but if they will be used in a plant that also has SDI video, it would be best if all the coax cables are made with the same components using true 75ohm BNC connectors.

When interconnecting AES3 and AES3id, high-quality balums should always be used; these usually come in small boxes or rack-mount panels with a BNC and an XLR. Splitting audio in an analog plant could easily be accomplished by just attaching the wires together, but in digital audio, this would create an impedance mismatch and a signal loss. Again, a high-quality, impedance-matched splitter for both AES3 and AES3id must be used. Both balums and splitters are available from several manufactures.

Synchronous AES

Here is an example of what can happen when digital audio is not synchronous. A broadcaster required an AES audio embedder to feed SD-SDI and AES/EBU from a professional digital satellite receiver to another piece of equipment. When he installed the audio embedder, the only audio received was a tone going up and down in level. After testing the AES3 output and determining that it’s good, he acquired another embedder, which performed no better. After contacting the manufacturer, the question of whether the AES3 is synchronized with the SDI video was raised. After doing some research, no device could be found that displays or measures AES/EBU synchronization with video, analog or SDI.

The only test is to force synchronization with video. This was accomplished using one of the AJA model ADA4 audio A/D–D/A converters. This unit can be configured as a synchronizer with a reference of analog black supplied. With this, the audio embedder worked, proving that the AES3 coming out of the digital satellite receiver was not locked to its SDI video. When an audio embedder that accepts asynchronous AES3 was installed, the system worked as expected and continues to do so.

Another issue with unlocked AES digital audio is the clicking and popping that can be heard when switching between sources. When AES digital audio sources are not locked, a switch can occur that splices two different sections of the audio blocks together causing the D/As to produce pops and clicks, which can be reduced when the AES digital audio signals are locked together.

Synchronizing AES digital audio with other sources can be accomplished in one of two ways: using a word clock or a digital audio reference signal (DARS). A word clock is a 5V square wave of the sampling frequency that will keep all of the samples synchronized with each other. This type of clock is used in recording studios where digital audio equipment is fed to a digital audio mixing board and all the sources must be locked together. The word clock works well to lock the samples but does nothing to lock the blocks or frames of AES/EBU digital audio; this is where DARS comes into play.

DARS, or AES11, is an AES3id (coax/BNC interface) signal with no audio (AES black) that is locked to a video reference. This one signal then carries all the information needed for a digital audio source to fully lock to and be synchronized with a video source as well as other AES digital audio signals. Some video sync generators provide DARS/AES11 as an output, while others just provide an AES tone output that should not be used for synchronizing. When AES digital audio carries an audio signal, the variation in pulses can cause some jitter in the regenerated clock, and because this is the reference, it should be as stable is possible. There are video-referenced AES/word clock generators available that accept analog video black, SDI or even trilevel HD sync to lock to and provide alarms if the video reference is lost.

Little attention is paid to AES/EBU lock to video, but there has been a great deal of discussion about lip-sync errors. And although most of these errors occur because of compression encoding, a good place to start correcting timing errors is before encoding. Digital audio and digital video must first be synchronized before the picture and sound can be synched over time.

Next time

The next “Transition to Digital” will cover how AES/EBU is locked to a video source and how it can be measured without specialized equipment.

Do you have any experience with measuring AES synchronization? Share your experiences in Broadcast Engineering’s Forum or send your feedback to!