start-up aspect of colpitts oscillator simulation

[Ian.Y.Jin]

colpitts oscillator simulation in ads

Dear All,

I'm simulating with a conventional Colpitts oscillator in Cadence. As the SpectreRF User Guide suggests, to initialize the oscillation, it should require some initial conditions on L or C, or some brief impulsive stimulus into oscillatory node. However, to my surprise, the colpitts oscillator can start up
without any initial conditions or brief impulse. The simulated center frequency, phase noise and tank voltage are in perfect match with that of colpitts oscillator with some brief impulsive stimulus.

Can anyone give some clue?

Best Regards,
Ian

 

[dick_freebird]

the colpitts oscillator

With a loose enough DC solution, the numerical error
residue may be enough to kick the oscillator when you
get into transient solution. This can be 100uV or so by
default.

However in other oscillators (like xtal) small numerical
errors can inhibit startup. You have to try out ranges of
tolerance settings and see which are both robust (for
startup) and realistic.