What's your best way to get SPICE to converge?

[solutions]

I try sprinkling multi megohm resistors in to try to get it to behave. Sometime slow rise/fall of signal generatiors.

What have you done, or has become a habit, to get SPICE to converge?

Is there a better SPICE implementation (convolutional simulator like ADS for example) that you've experienced?

[optikon]

** Changing solution method (Gear versus trapezoidal) - Gear lends itself better to switching type circuits

** Adjusting .options, namely ITLn and loosening up tolerance (VNtol, Charge tol, Rel tol)

** Replacing subcirucits with simpler versions (sometimes opamps for example)

** Adding realistic parasitics. Series R, C to Ground some L sometimes. Can help with AC/transient events.


I have only experimented with ADS a little bit but I like what I see so far. Most Spice simulators use the same basic core engine.

[solutions]

Correct.  But ADS does not - it uses a "convolutional simulator". Much faster.