On Wednesday 30 June 2010, Philipp Klaus Krause wrote: > I'm rather new to gnucap, so this could be a user error, but: > > I've seen the following problem three times: > gnucap just outputs lots of > @@@unreachable:u_limit.h:110:fet_limit_vgs messages (and a > R1.X1: RTEMP: effective length is negative or zero warning > before). Each time it occoured using level 3 models (and I > don't remember not seing the problem when using level 3 > models). I've tested on two different machines, both running > Debian.
Obviously, it isn't unreachable. It can get there if there is a numeric overflow. You didn't say what version of gnucap. There is a big difference between the stable branch (0.35) and the development branch (2009-12-07). In the development branch, with plugins, there are 4 level 3 mosfet models to choose from. It might be interesting to see what the difference is. Perhaps a convergence problem might be solved by using a different one. I didn't run it, but I see a few issues .... .MODEL BAV99 D + IS=7.4960E-9 + N=2.0077 + RS=.80464 *+ IKF=.1058 + CJO=515.09E-15 + M=.115 + VJ=.6389 + ISR=1.4182E-9 + NR=4.9950 + BV=90.375 + IBV=10 + TT=2.1640E-9 The * on the IKF line doesn't do what you think it does. It makes the line a comment, and also disconnects lines that follow, so the CJO line is an extension to the comment, not to the model. Gnucap does it like most programming languages do it. I didn't realize how spice was different. .print tran V(VccEN) V(pin) V(DC) V(logic) V(VppEN) i(VVpp) i(VVcc) V(inner) Look at the extra probes that gnucap has. A voltage source is not the best current probe. .MODEL RTEMP RES (TC1=8E-3 TC2=5.5E-6) R1 4 3 RTEMP 290E-3 When you use a the "RES" model for a resistor, the instance parameters are L and W, not resistance. I thought this was "spice compatible" but perhaps it isn't compatible with some particular variant of spice. _______________________________________________ Help-gnucap mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gnucap
