What's wrong with the VCO phase noise response

[qianxuan]

I have run a VCO simulation by spectre. The VCO is a differential inverter with a pair cross-couple PMOS. Output frequency is about 1GHz when input vctl equals 705mV.
The problem is shown in attachment.
curve1:vctl=706mV，out=998MHz
curve2:vctl=710mV，out=985MHz
curve3:vctl=704mV，out=1.004GHz
There is only a little difference between vctl but the change of phase noise is huge, more than 60 dBc/Hz in worst case.
My question is why the spurs happen at 1, 2 and 4 times signal frequency and and why they change so much

 

[chuckey]

Very mysterious! , are you sure that your oscillator is working properly, because Vctl/Fvco is not monotonic. Also could there be some unwanted resonance somewhere?
Frank

 

[qianxuan]

Very mysterious! , are you sure that your oscillator is working properly, because Vctl/Fvco is not monotonic. Also could there be some unwanted resonance somewhere?
Frank

they're nomotonic but I mess them up. I'll realign them.

 

[skal81]

The spurs are you VCO oscillation frequency harmonics. You are integrating the noise up to 10GHz whereas your VCO frequency is only 1GHz.
You should set the upper limit of the integration to half the VCO freq, otherwise
- peaks appears in the spectrum
- the noise at the harmonic is not taken into account -> setting to half the freq takes the noise into account by folding.

 

[sevak112]

Try to bring Kvco less than 200 MHz for better operation of PLL ...For that design voltage reference circuit very accurate for diff pair.and its also depend where you give your control voltage in diff pair.