Electronics Forum / CAD / September 2005

Spelling?

Actually the main difference is user interface... HSpice is just plain
butt-ugly gawd-awful to use.

                                       ...Jim Thompson

           
I love to cook with wine.      Sometimes I even put it in the food.

You might want to check out XSpice too. It's public domain, extensible and the
source code is online.

http://users.ece.gatech.edu/~mrichard/Xspice/

Microcode Engineering used it in their CircuitMaker CAD and worked well.

Protel/Altium bought them out, scared of the competition I guess. Too bad, they
had a good product they ported from Protel's DOS PCB cAD to windows back when
Protel couldn't.

PSpice is a version of SPICE sold by Cadence, HSpice is sold by Synopsys.
Both seem to get sold to other companies one or twice a decade though. :-)

PSpice is aimed more at discrete component circuit design, I'd say, although
there certainly are folks doing IC design with it (such as Jim Thompson).
HSpice is aimed almost exclusively at IC design... there are many IC fabs out
there that use HSpice as the 'golden standard' -- if you simulate your circuit
with HSpice using the fab's transistor models and the actual IC behaves
significantly differently in production than the simulation indicated, the fab
assumes it's their process that's at fault and re-runs the wafers rather than
figuring you screwed up your simulation somehow and would you like to write
another check for some good chunk of a million bucks (or more!)?

PSpice is insanely expensive (4-5 digits, depending on the options) whereas
HSpice is absolutely obscene (5-6 digits).

For the price, both are actually not that horribly impressive, either in the
SPICE engine itself nor in their user interfaces.  It's primarily momentum
that keeps them both alive, as far as I can tell.  HSpice has had some really
dorky bugs in it over the years -- for instance, there was a version where, in
the log file, it'd print out whether your MOSFET was in cutoff, saturation, or
its linear region... but they didn't bother taking the absolute value of
Vds/Vgs, and therefore it always claimed P-channel 'FETs were in cutoff!

---Joel