RFICDUDE
|
The length of time you simulate is really dictated by what you are trying to measure. If you are interested in power spectral density then the time length of the simulation determines the frequency resolution of the resulting power spectrum (i.e. delta_f = 1/tstop). This determines the effective resolution bandwidth of each frequency bin. My guess, and experience, says that you reasonably need a frequency resolution that is about 50 to 100 times smaller the bandwidth of your signal. So if you have a 1 MHz bandwidth signal then you probably need about 2kHz frequency resolution; although, you may be able to accept substantially less depending on your application.
A behavioral model (AM-AM, AM-PM for amplitude modulated systems or PM-AM, PM-PM for phase modulated systems) characterization of a circuit should greatly speed up the analysis. You don't even need envelope simulation if you have the simulated characteristic. You can use MATLAB or something similar for faster analysis after you have the input/output characteristic.
|