A problem of my LCVCO phase noise

[incol]

I have designed a 5GHz LCVCO. when i simulate the phase noise of it, i find the phase noise is bad. Seen from the noise summary, the mainly contribute to the noise is the transistor from bandgap. The bandgap only generate a reference voltage, and convert to bias current for the VCO. When i use ideal voltage source to replace bandgap, the phase noise can be improved. so i am confused is the problem of bandgap or VCO which worse the phase noise. Can anyone give me some suggestion? thanks a lot!

 

[incol]

Does someone have met this problem before?

 

[RCinFLA]

To have good phase noise there are two main criteria. First is minimize conduction time of active devices, hard saturation. Second is biasing. There should be no resistors used for direct connections for biasing oscillator device. The bias current must be clean of noise and injected via chokes so resistor noise does not degrade oscillator SBN performance.

 

[incol]

To have good phase noise there are two main criteria. First is minimize conduction time of active devices, hard saturation. Second is biasing. There should be no resistors used for direct connections for biasing oscillator device. The bias current must be clean of noise and injected via chokes so resistor noise does not degrade oscillator SBN performance.

thank you. So the noise mainly come from bias current, and i should bypass the bias current's noise, is it?

 

[RCinFLA]

Bias injection is a common mistake. It is not the only cause.