rfmems wrote on Jul 25th, 2012, 5:15am:I have been trying to figure out an accurate way to simulate a high Q DCXO, to get a idea of its amplitude, frequency and phase noise.
But I found the amplitude depends heavily on how the oscillation is started, give different stimulus or initial condition, the amplitude simulated are hughly different.
This doesn't make sense to me. The *initial* amplitude may vary but should stabilize. The final ampl. should be heavily dependent on your ESR and Cap values.
The other thing is its extremely painful (long) to simulate high Q values, so de-Q-ing, with modified series L-C values will greatly speed up transient simulations.