How to simulate phase noise of the injection locked oscillator

[TheOddPandy]

Hello everyone,
I'm trying to simulate injection locked oscillator(ILO) with cadence.I use the subharmonic as the injection source,I design a 20G oscillator to inject the 60G oscillator.

But when I use pss and pnoise to simulate,I found a problem.The phase noise is almost equal to the phase noise of the 20G oscillator,but theoretically use a 20G injection source with low phase noise to a 60G oscillator should cause a 20log3(about 9dB) phase noise deterioration at least.

I'm attaching my pss,pnoise setup and spectre.out(Vout3 and Vout4 is the different output of the 60G oscillator).You can see "the periodic noise analysis 'pnoise':freq=20.4564GHz" in the spectre.out.I think this maybe the reason why my phase noise is almost equal to the phase noise of the 20G oscillator.

So I think I may make mistakes in pss and pnoise setup.I didn't know how to simulate two oscillators with different frenquencies together.What's mistake in my setup,or my circuit maybe wrong,fail to locked?Should I try to use ads to simulate?

Please help me.
Thanks a lot!

 

[jiripolivka]

Hello everyone,
I'm trying to simulate injection locked oscillator(ILO) with cadence.I use the subharmonic as the injection source,I design a 20G oscillator to inject the 60G oscillator.

But when I use pss and pnoise to simulate,I found a problem.The phase noise is almost equal to the phase noise of the 20G oscillator,but theoretically use a 20G injection source with low phase noise to a 60G oscillator should cause a 20log3(about 9dB) phase noise deterioration at least.

I'm attaching my pss,pnoise setup and spectre.out(Vout3 and Vout4 is the different output of the 60G oscillator).You can see "the periodic noise analysis 'pnoise':freq=20.4564GHz" in the spectre.out.I think this maybe the reason why my phase noise is almost equal to the phase noise of the 20G oscillator.

So I think I may make mistakes in pss and pnoise setup.I didn't know how to simulate two oscillators with different frenquencies together.What's mistake in my setup,or my circuit maybe wrong,fail to locked?Should I try to use ads to simulate?

Please help me.
Thanks a lot!

View attachment 96804View attachment 96805

I do not know about simulation. What I know is that any controlled oscillator phase noise is defined by the reference-oscillator phase noise.
If the controlled oscillator is injection-locked, then it by itself has no own phase noise. The same holds for phase-locked oscillators.

 

[BigBoss]

If Cadence finds the oscillation frequency as 20.xx GHz, Pnoise will also compute Phase Noise at this frequency by nature of spectre.
Have you ever tought to change "Relative Harmonic" to 3 ??

 

[TheOddPandy]

Thanks for your reply.
I have tried to change "Relative Harmonic" to 3, now the frequency is equal to 60.xxGHz in the pnoise simulation, but the phase noise is still incorrect.
The output phase noise is just -24dBC/Hz,but the phase noise of the injection source is about -103dBc/Hz.