.option MAXORD=1 uses Backward Euler method.
Integration (IDT) problem. Why? - QSPICE - Qorvo Tech Forum
Thank you for the clarification on what “maxord” exactly means. The question is, what is the most optimal setting for this particular circuit? To me, maxord=1 seems to be the most suitable.
I have faced a similar issue in some circuits, representing an idealized connection of switching topologies. I try to use QSPICE instead of SIMPLIS. The solution that oscillates is unacceptable when I need to process (externally) the simulated data (e.g., extract voltage/current point to estimate switching losses). Even using gear still showed some overshoots.
In principle all three methods work correctly and give results that are accurate to within the tolerance specified. However, TRAP sometimes shows an unpleasant (to the eye) “ringing” on the signal. The ringing can be removed by using a smaller TSTEPMAX, but then the simulation runs slower. The GEAR method will never show ringing and damps the artifacts of the TRAP method. Much more important is that the GEAR method is extremely fast for ‘stiff’ networks (where very small and very large time constants are mixed together).
In my experience almost all convergence problems go away when using method=GEAR, reltol=0.1m and 80-bit math. The downside is that it is sometimes a bit slower than necessary.