Jitter or phase noise in LTSPice

[fireworks]

Is it possible to simulate the jitter of an oscillator with LTSpice ? If not what software can do that (preferably free) ?

If it is possible, an explanation of how to do it would be much appreciated. Also, please explain how to get the numerical values from the simulation results (voltages / currents).

[Elvee]

Something like that?

You can change the jitter profile by inserting any mathematical expression into B1's value.

[fireworks]

Originally Posted by Elvee

Something like that?

You can change the jitter profile by inserting any mathematical expression into B1's value.

What I want to do is create the schematics of an oscillator in LTSpice, simulate it and compute its jitter / phase noise.

[Elvee]

It is possible to reproduce any measurement you make in the real world in simulation: the FFT tool is the equivalent of a spectrum analyzer, and by measuring and integrating the power vs carrier offset, you could derive the phase noise.

You have to configure carefully the simulation and fft parameters to avoid measuring sampling artefacts.

Since this is not an audio-related topic, you should post your question on the yahoo LTspice group, this will maximize your chances of getting pertinent answers.

[jcx]

high Q oscillators are painful to sim due to the huge number of complete cycles of oscillation you need to sim, each at good resolution for the phenomena you want to see

the other problem is noise modeling in a .TRAN analysis isn't "native" - you have to add the component noise as additional models with the statisitics you expect

better oscillator circuits have low enough electronic noise that thermal drift, microphonics from external vibration become important - you would need to add even more "custom" sources to the standard Spice models to see these effects in a sim

[fireworks]

Originally Posted by Elvee

by measuring and integrating the power vs carrier offset, you could derive the phase noise.

Tanks, do you have a link to a more detailed description of this method ?

[fireworks]

Originally Posted by jcx

high Q oscillators are painful to sim due to the huge number of complete cycles of oscillation you need to sim, each at good resolution for the phenomena you want to see

the other problem is noise modeling in a .TRAN analysis isn't "native" - you have to add the component noise as additional models with the statisitics you expect

better oscillator circuits have low enough electronic noise that thermal drift, microphonics from external vibration become important - you would need to add even more "custom" sources to the standard Spice models to see these effects in a sim

I want to start by modeling a simple RC oscillator where I would only need a reasonable amount of cycles to get observable effects.
Do you have links to papers, books etc. that explain the process and maybe give examples ?

[Elvee]

Originally Posted by fireworks

Tanks, do you have a link to a more detailed description of this method ?

Just enter the terms I used in google:
"spectrum analyzer" and by jitter measurement and integrating the power carrier offset you could derive the phase noise - Google Search

But the devil is in the details, and implementing any method properly and efficiently (timewise) in LTspice requires care and attention, as jcx reminded you, that's why I recommended you ask the question on the specialized LTspice forum.

[ingenieus]

Make me think one I heard quite a few times as a student: "If you want an oscillator, design an amplifier! And vice versa, of course."

I have occasionally come up with a good oscillator, when I followed the above truism.

Sorry, this doesn't answer the question. Just wanted to share it. The Yahoo group should come up with better advice than this.

[fireworks]

Originally Posted by Elvee

Just enter the terms I used in google:
"spectrum analyzer" and by jitter measurement and integrating the power carrier offset you could derive the phase noise - Google Search

But the devil is in the details, and implementing any method properly and efficiently (timewise) in LTspice requires care and attention, as jcx reminded you, that's why I recommended you ask the question on the specialized LTspice forum.

Thanks, I'll take a look at the links and try the LTSpice forum.
Of course I am also interested in the method to do that in a typical SPICE software not just LTSpice.