Word Clock Basics

by Steve LaCerra
in Theory and Practice
Fig. 1: Word Clock in and output jacks in BNC format on the rear panel of a Yamaha CL5 digital console.
Fig. 1: Word Clock in and output jacks in BNC format on the rear panel of a Yamaha CL5 digital console.

Does Anybody Really Know What Time it Is?

Last month, we touched upon the idea of routing digital audio from an outboard channel into a digital mixing console. Any time that digital audio is transferred between devices, there are two distinct components to the signal: the audio, and something called word clock. Often (but not always) these signals are combined on a single cable, so we don’t realize that word clock is being sent along with the audio. Understanding word clock is critical when interfacing digital gear, and that requires a brief review of digital audio. So here we go.

Audio entering our digital mixing consoles is passed through an analog-to-digital converter (A-to-D or ADC), where it is electronically chopped into slices called samples. A sample rate of 44.1 kHz means that there are 44,100 slices of audio or “samples” per second. Each of these samples is represented by a number or a digital “word.” Every digital audio device has an internal clock to regulate the speed at which these samples are taken, and to make sure that there are in fact precisely the correct number of samples per second as specified by the sample rate (e.g. 44.1, 48, 88.2, 96 kHz etc.). If not, we run into problems (errors) that can create noise, loss of signal and possibly an undesired change in pitch.

When using only one digital device, miniscule errors in the clock (known as jitter) can go unnoticed because the system is “closed” and there is nothing to compare it to. For example, if the clock in your digital console was actually running at 44,101 kHz (which would be considered a massive error), you probably wouldn’t notice — that is, until you tried to patch the digital output of that console into the digital input of another device such as a system processor. In order for the two devices to properly exchange data, we need two things to happen. First, they must be set to the same sample rate, and secondly, their word clocks must be synchronized. The first part is easy due to the fact that most digital devices have a menu where you set the sample rate. The second part is not so easy.

The Champagne Analogy

Here’s an analogy: Imagine that you and your significant other are running on the beach together (cue romantic music please). So that you can look each other in the eyes lovingly, you are running at the same speed (it would be quite rude if one of you was trying to outrun the other). This would be like you are running at the same sample rate. But as you look down at the warm ocean water flowing over your toes you notice that your feet do not hit the sand at the same time as your partner’s. In fact, your partner’s left foot is hitting the sand fractions of a second after your right foot hits the sand. You are not in sync (like, you are each following a different clock). Because your body is bobbing up while your partner’s body is bobbing down, you have difficulty passing a glass of champagne (digital word) between you. The champagne spills, resulting in an error and a loss of information.

You need to synchronize your running to a higher standard. You need the same word clock. This would ensure not only that you run at the same speed (sample rate) but also that you take the exact same steps, at the exact same moment in time so that when you exchange the champagne (data) each of you is at precisely the correct position to give or receive the glass without error. Either you must be locked precisely to the motion of your partner or vice versa. One of you must serve as the master timing device while the other follows (the slave). That’s what a word clock signal does.

Word clock travels in all sorts of ways. Those S/PDIF and AES/EBU digital connectors on the rear panels of your gear carry two channels of digital audio plus word clock. An ADAT optical (TOSlink LightPipe) connection carries eight channels of digital audio (at sample rates of 44.1k or 48k Hz) plus word clock. And your Dante network connector? Multiple channels of digital audio plus word clock! The BNC connectors you see labeled “word clock” carry — you guessed it — word clock, but nothing else (see Fig. 1, rear panel of a Yamaha CL5 console).

A Real-World Example

Let’s suppose that you purchased a digital system processor and you plan to connect it to your digital mixing console. If the console has a digital output and the processor has a digital input, it makes sense for you to digitally route audio from the console to the processor — avoiding a few unnecessary D/A and A/D conversions in the process.

Next, we see that the console has an AES/EBU digital audio output, and that the processor has an AES/EBU digital audio input. These are XLR connectors, so you whip out a 25-foot microphone cable and patch them together. That’s a mistake. You need an AES/EBU cable, which looks a helluva lot like a mic cable, but, alas, is not. The impedance of an AES/EBU cable is 110 Ohms as opposed to the impedance of a mic cable which is somewhere in the vicinity between 45 and 70 Ohms. You might get away with using a short run of mic cable for the purpose (perhaps 5 feet) but once the run gets longer, you’ll need the proper cable impedance or the cable can degrade the signal, introducing — guess what? — clock errors.

Fig. 2: On the Yamaha CL5 (and most other digital consoles), the Word Clock settings are made on a menu page, such as this one that selects a 48 kHz internal clock signal.

We go back to the bench, make a few proper AES/EBU cables and patch the desk’s output to the processor’s input. The next step is to dig into the menu of each device and set the sample rate the same. If you then listen to audio through the system, it might sound okay for a while. Or it might sound okay for a few seconds until you hear what sound like static noises (clicks and ticks). Those are clock errors. In severe cases, you will hear no audio at all and a lot of static.

The last step to this happy marriage is sync’ing their word clocks. We want one of them to be the master, and the other to be the slave. In the clock menu of a typical digital device there should be several options such as internal (the device uses its internal clock as the master), AES/EBU (the device uses the clock derived from an AES/EBU input) or perhaps Word In (the device will look for a clock signal at the rear-panel BNC Word Input jack). See Fig. 2, a screen shot from the Yamaha CL5, showing its Word Clock settings. In our example, the console’s word clock is set to Internal 48 kHz. It will send word clock to the system processor along with the L/R audio bus. If we set the system processor’s clock to external, it will receive the clock via the AES/EBU connection, and follow the steps of the mixing console’s clock so that they can run on the beach into the sunset, sipping champagne but never spilling it.

I’ve hurt your brain enough this month. Next month we’ll finish our discussion of word clock.

Steve “Woody” La Cerra is the tour manager and FOH engineer for Blue Öyster Cult.