The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> How does Spectre calculate phase noise?
https://designers-guide.org/forum/YaBB.pl?num=1411746772

Message started by carl_chao on Sep 26^th, 2014, 8:52am

Title: How does Spectre calculate phase noise?
Post by carl_chao on Sep 26^th, 2014, 8:52am

Hi,

Does Spectre measure the final phase noise like this:

1) Measure the periods (T(i))of the clock waveform, and calculate the average period (T)from it;
2) Cauculate time domain phase noise like this: 2*pi*(T(i)-T)/T;
3) Perform FFT on the time domain phase noise to get the phase noise spectrum.

Thanks

Carl

Title: Nobody knows or is it too simple a question?
Post by carl_chao on Oct 24^th, 2014, 11:22am

??????

Title: Re: How does Spectre calculate phase noise?
Post by Ken Kundert on Oct 24^th, 2014, 12:22pm

No, that is not the way SpectreRF computes phase noise. In fact, when you use the phase noise direct plot feature in ADE you are not plotting the phase noise at all, rather you are plotting the normalized noise (the time-averaged noise divided by the signal power). It happens to be an excellent approximation to phase noise when applied to oscillators, but not when applied to non-autonomous circuits

If you want to know how SpecreRF performs noise analysis, take a look at Introduction to RF simulation and its application.

-Ken

Title: Re: How does Spectre calculate phase noise?
Post by carl_chao on Oct 30^th, 2014, 9:08am

Thank you very much for the reply! So Spectre calculates both the amplitude and jitter noises in its phase noise analysis. After the oscillator stablizes, I expect the noise should come mostly from phase noise. Especially for sampling circuits, is it more suitable to calculate the phase noise in the way as I said in my first post? No that it's called "phase noise", why doesn't Spectre just consider the phase portion of the noise by extracting the time information at the threshold crossing points (like half supply voltage). It should be hard for Spectre to do that.

Best,

CZ

Title: Re: How does Spectre calculate phase noise?
Post by Ken Kundert on Oct 31^st, 2014, 1:42pm

What you suggest is a considerably more expensive and less accurate simulation, and as you point out an oscillators noise is generally dominated by phase noise, so the easiest and fastest thing for the simulator to do is to compute the time-averaged noise and have you, the user, compute the normalized noise and interpret it as phase noise.

Your response was difficult to parse. Please read what you write before you post and make sure you are actually saying what you mean to say.

-Ken