Guitar | Bass | Keyboard | Microphones | Mixers | Audio Interfaces | Monitors | Sequencers | Soft Synths | Live Sound | Drums | Club | Accessories | Blowouts
Setting up Word Clock
in simple and advanced digital audio recording systems
by TweakHeadz Lab
As your digital audio rig gets larger you will inevitably have to deal with word clock distribution. At first, synchronizing word clocks of different digital audio devices is easy. In most cases, you can generate and transmit the word clock signal along with the digital audio signals passing from master to slave through s/pdif coax, Toslink and (sometimes) ADAT optical cables and AES/EBU cables. But soon enough you will run into a situation where syncing through audio cables will stop working.
That is where you need to set up a way to distribute the word clock master to all the slaves that need it. This is very simple stuff, so don't let it freak you. But don't confuse word clock with MIDI clock synchronization, Midi time code, SMPTE, or anything that regulates tempo. We'll talk about those in the next article.
Sample Words and the clock that regulates them
First, without getting too technical lets get into why we need a word clock. Digital audio, as we know, is made up of 1s and 0s, or bits. These bits are grouped into sample words. The word size for common digital audio is 16 bits, 24 bits and sometimes 32 bits and even 64 bits. Indeed when we talk about 24 bit audio we mean data with sample words of 24 registers for bits. When these sample words start and end at the same exact time on several devices, we have word sync. That is what we want.
The word clock sets the rate at which these words are transmitted. This to takes us back to our old friend the sample rate. We might use a clock rate of 32kH, 44.1 kHz, 48 kHz, 96 kHz, and for those who have bought the hype 192 kHz.
When you use multiple digital audio devices they must use the same size word and travel at the same speed. We are keeping this simple. (More technical descrriptions) You might ask. What happens when you send a signal at a clock of 48kHz into a device working at 44.1. If you are lucky, it will just stop working. If not, you might find one of the devices explode into an intensely loud cascade of white noise. That is what can happen during an obviously big clock mismatch. (This is a good reason never to wear headphones when messing with word clocks! Keep your system volume WAY down too.)
But what about when the clocks are very close, but not perfectly, synchronized. Then we get what we can call "jitter". Jitter may be subtle or extreme. At it's extreme, there will be distortion that almost sounds like ring modulation in all the audio. Sometimes you will hear the rhythmic pulsing of soft white noise bursts happening about a second apart. (I usually hear that when I have 2 devices set to master.) As the rates get closer this is more like loud popping and crackling noise in the audio which still makes the audio unsuitable. As the rates get even closer, you might hear only a few microscopic barely audible ticks once every interval, sometimes 5, 20, 50 seconds apart. Many of us may actually have systems ridden with jitter but because the artifacts are so far apart we tend to ignore them like one would ignore the occasional record pop on a vinyl record.
SuperClocks. How much is hype?
The professional has to be particularly careful about distributing word clock because these ticks, even if far apart, are not acceptable. These tics and pops not only exist in the playback of audio, but are burned into the recorded data as well. While they can be surgically removed with audio editors, it is much better to get it right the first time! There is also a bit of audio voodoo here. Many professionals claim that even though there are no audible artifacts of jitter, a less that perfectly stable clock can cause degradations in clarity and on the width of the sound field. This leads them to buy special word clock generators known to have excellent stability. I call them SuperClocks. As you approach the high end digital audio converters, you will often see manufacturers tout why their clocks are better than their competitors. Is this hype? Much as with the claim that recording at 192kHz is "better", if there are differences in quality we may not be able to perceive them. A professional might simply buy a super word clock solution as insurance against jitter. Voodoo? Let your ears decide.
Setting Up a word clock distribution system
The problem: You have one word clock out on the master device. How do you get this one cable to go to all 3 slave machines? First, there can be only one master clock source. All other digital devices must be slaves. Second, word clock is a one way system. it does not have to loop around back to the Master's word clock in. Third, you can't assume that devices with a WC in and WC out will pass the signal at the in to the out. The WC out may be reserved only for signal internally transmitted.
Now lets apply a synchronized word clock to solve this problem.
There are two solutions here. 1) The more expensive solution is to get a word clock generator with multiple BNC outputs. Here the central device sends out the word clock to all four WC outputs. Each device reads the clock, locks into sync, and audio data flows smoothly from one device to the next.
Back of the Lucid Genx192 Notice how the inputs can send signal to various formats such as BNC, AES and s/pdif.
2) A second, less expensive method is to get several BNC "Tee" adapters. Every time you need to add a digital device with word clock, you get another cable and another Tee adapter and simply extend the system. You want to observe the gender of your cables to make sure you get the right ones. But I suggest the following rule. Because all WC ins and outs are female, all WC cables you buy should be male to male. That will mean that your Tee adapter should be male where you plug it into the unit and have two female arms. (see pic). That way you connect your male WC cables directly to each arm and on to the next device. Using the Tee basically splits the signal into two, so one goes to the device and the other passes on to the next.
To test your word clock connection just simply change the sample rate on the master. All the slaves should simultaneously change their sample rate. When you see that, you know you have wired the cables correctly. Now make a recording from each digital device listen carefully over headphones on playback for any evidence of ticks, pops, clicks, and more obvious forms of distortion. You should hear perfectly clean audio. If you do you have successfully applied a word clock distribution system. Congratulations.
Questions and Answers
Q) What is s/pdif, toslink, ADAT and AES/EBU
A) All of those a methods of transmitting digital audio data from device to device. Note that digital audio data is your audio signal after it has been converted to ones and zeros.
Q) Can you make your own Word Clock cables?
A) Yes. The cable is good old 75ohm cable just like your TV cable that has "F" connectors. Just go to an electronic store (or use my links to Amazon) and pick up some Female "F" to male BNC connectors that can be screwed on. Many video RCA cables are also 75ohm. With these just buy some Female RCA to Male BNC adapters. Coaxial s/pdif cable 75 ohm RCA cable. That can be used as well. BNC cables are used in professional broadcast systems and in home video surveillance systems. Places that serve these customers are more likely to have the adapters than your local music store.
Q) Is there any rule of thumb for deciding which device should be the master?
A) Yes. The device that does the most critical analog to digital conversion should be the master, when possible. The theory is that the "native" clock on the device doing the conversion will be better. However, a stable clock is better than an unstable clock, so in some situations the native clock rule does not hold.
Add you comments on this article at the following forum:
Definition of Word Clock in Wikipedia
Go to the Next Class
Go to the Previous Class
Tweak's Articles on Essential Studio Concepts