1
$\begingroup$

Regarding white (uncorrelated) and coloured (correlated) timing jitter. I know that for free-running oscillators (such as VCOs), various kinds of noise causes the jitter, for example (1/f) coloured noise and white noise...As a result, the jitter is expected to be coloured in oscillators because the phase noise power spectral density will have regions of (1/f^3) and (1/f^2) due to (1/f) and white noise respectively.

However, I am wondering if the timing jitter at the output of PLL is white or coloured?

From one perspective, I can say that PLL acts as a lowpass filter, so it filters out the jitter, as a result: the white jitter becomes coloured. Therefore, we always expect to see coloured timing jitter at the output of PLL.

From another perspective, I can say that PLL is intended to reduce the jitter, so we end up with white timing jitter due to white noise (which is always there). Therefore we expect to see white timing jitter at the output of PLL.

Could you please help me to know if the jitter at the output of PLL is white or coloured?

thanks

$\endgroup$
1
$\begingroup$

The noise is colored. If we are to talk of spectral densities (color) then we would best refer to phase noise and not jitter (jitter is the result of integrating the phase noise, and typically after a high pass function given by the typical cycle-cycle jitter measurement).

The PLL's phase noise would be a low pass filter of the phase noise from the reference oscillator, and a high pass filter from the phase noise of the unlocked VCO, resulting in a typical plot of composite phase noise of the locked VCO as shown below. Overall this clearly not white in phase or in frequency, however there are regions of frequency offsets over which the frequency noise (FM) is sufficiently constant (where ever there is a 20 dB/decade roll-off in phase noise is constant FM).

Typical Resultant Phase Noise for a Locked Oscillator:

composite phase noise

The combined phase noise above is the result of the summed high-pass and low-pass contributions of the VCO and reference phase noise, where the filter transition between the two is given by the loop bandwidth.

Side note with regards to the quick sketch above: The phase noise of the reference as it would effect the resultant PLL will typically have many different slopes (I show just one slope above) increasing as we approach smaller and smaller frequency offsets. Such transitions can be predicted using Leeson's equations.

$\endgroup$
1
  • $\begingroup$ Thank you very much for your answer... $\endgroup$ – Amro Goneim May 31 at 18:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.