LC Oscillator Simulation – Why Ideal LC Oscillator is Highly Damped in Spectre but Not in PSpice

oscillatorpspicesimulation

I'm just curious about why a ideal LC oscillator (See Figure) has a damped response in Spectre Simulator but not in Ngpsice. In both cases I'm using 5V of initial condition. I'm also using gmin=0 and gmin_in Spectre in order to avoid some energy losses. In Ngspice, I'm using default convergence options.

schematic

simulate this circuit – Schematic created using CircuitLab

The Spectre results gives me something like:
enter image description here

Does anyone has a hint about it?

Thanks

Best Answer

I'm going to guess that it's the integration method used. Trapazoidal integration is good for oscillators, but Gear integration introduces numerical damping that tend to make oscillators damp out rather than oscillate.

Most simulators provide an option to select the integration method. Try the same method in both simulators to see if you get similar results.

See here:

"The one area where you should definitely avoid Gear is in LC oscillators. A pure LC tank circuit with an initial voltage on the cap will oscillate forever with trapezoidal integration, but will rapidly decay with Gear integration."