Tina-TI LC overshoot dependent on simulation time

Paul,

Apologies for the delay.

I was able to reproduce your observation by using a unit step driving a 1mH inductor in series with a 1uF capacitor.
The output was assumed to be the connection between the inductor and the capacitor. The simulator option was "trapezoidal" of order 2.
The output swing was dependent on the total simulation time.
With just a few cycles (1ms run time), the peak-to-peak output was 1.9V. When the sim was stretched out to 10ms, the swing dropped to about 0.6Vpp.

As an experiment I tried approximating the source as a voltage impulse.
This was done by changing the source from a voltage step to a narrow voltage pulse with a 1us duration. The rest of the circuit was unchanged.
The output swing became about 59mV peak-to-peak and seemed to be more-or-less independent of the total sim time.
Increasing the duration of the pulse caused the output swing to increase (normal) but the final voltage swing appeared to be independent of the total run time.

My TINA configuration uses the default selection for the max step size used by a transient simulation. To see this parameter select the Analysis/Set  Analysis Parameters tab, and scroll down to TR Maximum Time Step. The default setting is "10G" which means the simulator chooses and varies the time step automatically during the simulation. This allows the solver algorithm to balance accuracy and sim time while conforming to the limits imposed by the other analysis parameters.

 My guess is that with the idealized, lossless circuit, the solver was picking time steps that resulted in an aliasing effect and missed the waveorm peaks.
In other words, with a short sim time the solver picked small time steps that caught the waveform peaks, resulting in a larger swing.
When the sim time was lengthened, the time steps grew and the the solver missed the waveform peaks.
To test that theory, I tried changing the max time steps from the default  "10G" to a fixed value of 1us.
The result : the output swing became independent of the total sim time (~2Vpp).

So with the ideealized lossless circuit, the smaller time steps avoided the aliasing effect brought on by the automatic step size.

The next step was to introduce a very small amount of loss in the form of a 1uohm series resistor.
Even with the automatically chosen step size, the output swing became independent of sim time.

Hope this helps. Please let me know if you have any more questions.

Regards,

John