Jitter: Part 3 – Classifying Jitter

#3 in a series of articles about wander and jitter

This posting is a simple one… It’s a setup for the next bunch of postings in this series.

In the last posting, we saw that jitter can be separated into two categories, looking at whether the root of the problem is in the time or the amplitude domain.

A different way to categorise jitter is to start by looking at whether the variation in the timing error is random or deterministic.

If the timing error is random, then there is no way of predicting what the error on the next clock “tick” will be. In this case, the error is caused by some kind of random noise (I know – that’s redundant) somewhere in the system.

However, if it’s deterministic, then the timing error will be correlated with some measurable, interfering signal that is not just random.

An Analogy for Obfuscation

One way to think of this is to imagine the sound coming from a poorly-made piece of audio gear.

  • You’ll hear the signal
  • you’ll hear some distortion artefacts that are somehow related to the signal
  • you’ll hear some “hiss”
  • and you’ll also hear some “hummmm”.

The signal is what you want to hear.

The other three are things-you-don’t-want-to-hear: stuff that would traditionally be included as TDH+N or “total harmonic distortion plus noise”.

The distortion artefacts are things that are unwanted, but somehow related to the signal – so they are not periodic, but they’re deterministic.

The “hiss” is independent noise – a random signal that is added to (but also unrelated to) your signal.

The hum is not random – it’s periodic (meaning that it repeats itself) – and therefore it’s also deterministic.

 

The sum of all jitters

 

Fig 1: A chart showing the classification of jitter types

 

If you have a digital audio system, it will have jitter (remember – this might not be anything to worry about… sometimes it just doesn’t matter…). The total jitter that it has is the sum result of all of the different types of jitter that contribute to the total. So, in the chart in Figure 1, you can sort of think of each of the black lines as representing “plus signs”. (Only “sort of”, since it depends on exactly what you’re measuring, and for how long…)

My plan is that I’ll address each of these blocks individually in the coming postings in the series.