I'm simulating my upconversion mixer.  

After parasitic extraction, I ran PSS (Flexible Blance Method)analysis and Transient analysis. Surprisingly, Those two simulation are totally different simulation results. PSS simulation showed me reasonable (good?) result while the transient simulation showed very poor performance.  I tweaked the tolerance options but the transient simulation result didn't change much. Before parasitic extraction, both analysis showed me identical result.  

Does anyone know the reason of this weird behavior of the simulator?  

Thanks

Thanks for your opinion.

The wave form from the two simulation differ in regard to amplitude rather than the shape of the wave. I'm not sure if it is some information which decide something, though.

I appreciate for your interest.

For transient analysis, I confirmed it is in steady state. FYI, my application is upconversion mixer(LO=44GHz). I looked at the output of the mixer in transient wave form.
For PSS (Flexible Balance) analysis, I setup tstab=20ns. After this simulation I also looked at time domain wave to compare.

I also compared Flexible Balance and Shooting method. Those two method showed identical results. As far as I know, Shooting method also rely on transient analysis. If it is right, shooting method must be the same as transient analysis. It also confused me.
Is it possible for circuit components in the extracted view work correctly only for a specific analysis type?

Before parasitic extraction, both transient and PSS analysis showed identical or very close results. With parasitic extracted view, the simulation result became significantly different.  The amplitude in the wave form differ in the order of 100. The transient analysis showed very poor performance.

I will post the simulation captures soon.

Thanks