Have you reviewed the documentation for this analysis tool by chance? And if so, are there any aspects of it that you find insufficient? I ask because many of these aspects are covered already there.
But anyway, I’ll attempt to answer all of your questions here…
The steady-state analysis requires that you have a system with periodic operation. This could be, for example, an AC frequency in the system, a PWM frequency, or a combination of one or more than one of each, where the least common multiple of all periodic sources is used to determine an overall “system period”. The analysis completes when via an iteration-based method, two consecutive cycles are determined to be within a user-defined “termination tolerance” of each other such that the algorithm decides the system is at least close enough to steady state.
The simulation start time parameter is simply a time offset for when the simulated steady-state waveforms are shown. It should not affect convergence or the final waveform behavior.
The number of final cycles is the number of “steady-state cycles” that will be displayed in all scopes when the analysis is complete. You just need one final cycle, but having more than one can be helpful for visualization of the repetition of the various dynamics in each given cycle.
So the analysis should proceed and succeed based on proper selection of the system period (or using the auto option), the termination tolerance, and potentially some additional tweaks to the number of initial cycles and relative perturbation for the Jacobian, but these latter two can be considered more advanced settings. The start time should not be considered as part of your process to correctly achieve steady-state values.
Also, it’s certainly possible that within a periodic cycle at steady state that you will observe some repeated transient in a given waveform, and you should not just expect a flat line for every signal, but if you are seeing true transients that are not repeated from one steady-state cycle to the next then it sounds like you are not achieving a correct steady-state operating point and the analysis hasn’t succeeded.
Of course, if you are able to post a model that would be most helpful in allowing someone to help give you guidance on your specific question.
At what number of final cycles are you getting an error? And what is the error message?
And sure, it is possible to run a steady-state simulation from a script- please review the section of the PLECS Documentation: Simulation Scripts > Scripted Simulation and Analysis Options.