原创 Setting-up HFSS/SIwave for SPICE Transient Simu

- Use Discrete Sweep rather than Fast or Interpolating.
The algorithms used within HFSS and SIwave to do interpolation do not preserve causality.

- Use a frequency sweep that gets as close to DC as possible.
You may be planning to excite your structure with a source of
limited frequency extent, but if you’re using SPICE’s transient
solver, you’ll need a good DC point, since DC must be solved to
give transient a starting point.
Also, since sine waves and other ostensibly band-limited sources
must start from zero for times less than zero in SPICE’s
transient, there may be more low-frequency content than you expect.

- Remember the signal-integrity rule of thumb, that the
knee frequency of a pulse is somewhere around 0.35/risetime to
0.5/risetime of the signal.(There can be significant energy above that point, depending on the
shape of the pulse.) So a 100ps risetime signal requires a top
frequency of simulation of at least 5GHz (and often much higher),
no matter what the period of the signal is.
-Avoid dielectric materials with a constant non-zero loss tangent; they are non-causal. Use either lossless materials, or “frequency-dependent” ones.

- If you’re planning on using convolution, it’s best to use a uniform frequency sweep.

- Eliminate any unused ports; extra ports put a burden
on the state-space fitter, particularly if you choose to enforce
passivity.