Designer’s Guide Community

Visjnoe

Senior Member

Offline

Posts: 233

Ring oscillator phase noise integration limits
Jun 06^th, 2008, 12:01pm

Dear all,

assume a ring oscillator with a square wave like output and frequency fo. If one wants to calculate the RMS jitter of this oscillator,
one should integrate the phase noise, take the square root and divide by 2*pi*fo.

But what is the upper integration limit of the phase noise integral? In several discussions on this forum, I have found 'fo2/' as an answer.
But this was always for driven circuits like dividers, clock buffers...

Does this also hold for an autonomous oscillator?
If not, what is the integration interval to get an accurate jitter estimate?

Regards

Peter

Back to top

IP Logged

David Lee

Junior Member

Offline

Posts: 29

Re: Ring oscillator phase noise integration limits
Reply #1 - Jun 6^th, 2008, 3:49pm

Visjnoe wrote on Jun 6^th, 2008, 12:01pm:

Dear all,

assume a ring oscillator with a square wave like output and frequency fo. If one wants to calculate the RMS jitter of this oscillator,
one should integrate the phase noise, take the square root and divide by 2*pi*fo.

Which rms jitter is of interest to you?
a) absolute jitter? This is infinite for a free running oscillator.
b) period jitter?
Your method does not yield b), I'm afraid. There are two common definitions for phase noise. If you take the normalized ratio of the power spectrum of the noisy waveform over that of the noiseless waveform, total area under the curve is 1. If you take the power spectral density of the phase deviation (in radian^2/Hz), the total area is infinite. Your circuit simulator does not report rms period jitter?

Back to top

« Last Edit: Jun 6^th, 2008, 10:13pm by David Lee »

- David

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: Ring oscillator phase noise integration limits
Reply #2 - Jun 6^th, 2008, 9:00pm

I would think of the equivalent noise bandwidth for your oscillator. Basically, if you open the loop and perform and ac analysis right at the quiescent point at which each inverting stage is switching you will get the gain is maximum. However such gain will decrease with frequency down to a point at which it will not be enough to guarantee oscillation once you close the loop. That would be a good spot for setting the upper limit of the integral for calculating the phase noise.

Tosei

Back to top

Keep it simple

IP Logged

Visjnoe Senior Member Offline Posts: 233	Re: Ring oscillator phase noise integration limits Reply #3 - Jun 7^th, 2008, 6:13am Dear, I mean RMS (period) jitter. My simulator (HSPICE RF) does report RMS jitter but the user has to provide integration boundaries. Regards Peter
Back to top	IP Logged