What is jitter?
It's the undesired deviation of a periodic signal from the ideal timing...
So, even though the above definition is correct, it doesn't really let you know why jitter is bad for a digital audio signal, how jitter actually degrades the signal or what the aural effects of jitter are. Let's take a closer look at jitter in order to answer these questions.
First, it's helpful to understand a bit about word clock. Here's a graphical representation of a word clock signal:
A word clock signal acts like a conductor, providing a timing signal to all the parts of a digital audio system so that each process may be triggered at a precise moment. If you think of your digital audio system as an intricate mechanical device, the word clock is analogous to the teeth of the gears that make various parts of the device move together. Think of Charlie Chaplin in "Modern Times":
Here's another way to think about how a digital audio system works. Start with an analog waveform:
Now, in order to digitize this waveform (i.e describe it using numbers), let's overlay a piece of graph paper:
Using the graph, we can describe the waveform with letters and numbers:
At A the wave is 8;
at B the wave is 11;
at C the wave is 14,
at D the wave is 15, and so on.
We could store these numbers and use them to recreate the waveform later. Please don't bust our chops about the obvious inaccuracy between the red waveform and the blue points, it's an analogy, people.
So, to forge on with the analogy, imagine that the letters are samples and the numbers are bit levels. Now we're recording digital audio. It might look like this:
At sample 1, digital level is 0111;
at sample 2, digital level is 1011; and so on.
Notice our word clock signal at the top of the graph. It's the signal that triggers each sample in a digital audio system.
Once we've recorded our waveform as numbers, we can "re-plot" the points and re-draw the waveform to get back our original signal. We've just recorded and played back a digital audio signal.
The process of converting our waveform to numbers is called analog to digital conversion - here's what it might look like:
The process of converting the numbers back into a waveform is called digital to analog conversion - here's a movie of that process.
Now, throughout this whole process we've assumed that the vertical lines on our graph paper are evenly spaced. Imagine that we got some faulty graph paper where the vertical lines weren't evenly spaced and we attempted to re-draw our waveform using the numbers we previously recorded. Clearly the resultant waveform would not be like the original:
Look at the top of our graph, that's what a jittery word clock looks like. Notice that the distance between the transitions is uneven - this is jitter.
Though we've greatly exaggerated the amount of jitter in this graphic, it does show how a jittery word clock causes samples to be triggered at uneven intervals - this unevenness introduces distorsion into the waveform we're trying to record and reproduce.
For another take on jitter, let's think again about Charlie Chaplin and "Modern Times". A film functions in a similar manner to digital audio, in that a film camera doesn't record every instant of a scene, but rather "samples" a scene by taking a series of still pictures at a fast enough rate to fool the eye into thinking it sees fluid movement. As you can imagine, the regularity of the exposure (and subsequent projection) of film frames is crucial to maintaining the illusion of fluid movement. Back in the days of Chaplin, early film cameras weren't so even - frame jitter made the movement seem jerky and unnatural.
If jitter gets into our D-to-A stage, it degrades the playback. The original digital recording is intact, and we just need to remove the jitter or get a better D-to-A converter to resolve the issue. Going back to our film analogy, if our projector is uneven, our film is probably fine, we just need a better projector.
On the other hand, if jitter gets into the A-to-D stage, those errors are "baked into" the digital data - there's no recovering the original waveform. If you've shot your film with a hand cranked film camera, there's no way you're getting that jerky motion out of the film. This is why getting the A-to-D stage of a digital recording system right is so crucial.
So, what does digital audio affected by jitter sound like? To answer this question, imagine a recording of an orchestra made with a perfectly placed stereo pair of mics. If you listen to the analog output of the mic pres, ideally you'll have a precise and wide image of the orchestra. If you closed your eyes, you could point to the triangle player, there, behind the second violins.
Smaller amounts of jitter start to cloud the precision of the stereo image - you can't pinpoint sources as clearly. As greater amounts of jitter degrade the system, the stereo image starts to shrink - your orchestra isn't so wide. With increasing jitter, changes in timbre occur that accentuate ugly harmonics.
To avoid jitter, use a finely engineered clock source as found in all of Apogee's interfaces. If you're using multiple digital devices, use a Master clock like Apogee's Big Ben and make word clock connections using our Wide Eye cable.
Finally, be aware that frequency drift, often mentioned when discussing digital audio clock, isn't jitter... see here! https://knowledgebase.apogeedigital.com/?p=2093