In an ideal world, an oscillator’s output waveform would only be at one specific and exact frequency and also be stable with regards to time; however all oscillator output signals also include ‘noise’ whether that is in the form of fluctuations in frequency, time or amplitude. Therefore, we need some metrics to measure and make sense of these imperfections. That is where jitter and phase noise come in. So what exactly is jitter and phase noise and why should it be taken into consideration for applications?
Variations in the frequency domain are referred to as "phase noise" and variations in the time domain as "jitter". For this, consider a signal with two states, ‘on’ or ‘off’; this signal would ideally have a constant time period between the pulses, and all the pulses would be of equal lengths, and it would be easy to predict when the next pulse arrives.
However, if the signal is corrupted by ‘noise’, be it from within the device or via an external influence, this can cause the pulse to arrive early or late (Figure 1). This is essentially jitter and can be problematic by causing system performance degradation.
There are different types of jitter: period jitter, cycle to cycle jitter, phase jitter, and peak to peak jitter. At IQD, we mainly specify the phase jitter which is calculated from a phase noise plot.
It is important to note that all types of jitters describe the same phenomenon, but from a different point of view. This means that the values of different jitter types cannot simply be compared with each other.
