Andrew Beckett
Senior Fellow
![* *](https://designers-guide.org/forum/Templates/Forum/default/starsilver.gif) ![* *](https://designers-guide.org/forum/Templates/Forum/default/starsilver.gif) ![* *](https://designers-guide.org/forum/Templates/Forum/default/starsilver.gif) ![* *](https://designers-guide.org/forum/Templates/Forum/default/starsilver.gif) ![* *](https://designers-guide.org/forum/Templates/Forum/default/starsilver.gif)
Offline
![](https://designers-guide.org/forum/avatars/elmerfudd.gif)
Life, don't talk to me about Life...
Posts: 1742
Bracknell, UK
|
A straight conversion of the analog primitives into VerilogAMS won't speed it up (if anything it could be slightly slower).
Depending on what is the root cause of the slowness, you could get it faster by writing a more direct VerilogAMS model of the motor though, but all the time it has continuous nodes (continuous in time) it will still need an analog solver (e.g. spectre) to solve and won't be as fast as the pure digital simulation.
So one approach is to use the real-number modelling technique - see <IUSinstDir>/doc/wreal/wreal.pdf (the Verilog-AMS Real Valued Modeling Guide) which shows how to use event-driven reals instead, which can be used entirely in the event-driven (i.e. digital) solver.
Regards,
Andrew.
|