On Tuesday 03 April 2007 09:20, Stuart Brorson wrote: > >> Here's what crashes for me: > >> > >> $ ~/local/bin/gnucap spice.netlist.wpd > >> gnucap> plot ac v(Vout) > >> gnucap> ac dec 1Hz 1MegHz > >> > >> #Freq > >> Segmentation fault > > > > 1. You need to do "op" before "ac". > > I was thinking about this during my morning commute. It's > fine that Gnucap wants an op performed before doing anything > else. However, it shouldn't segfault if the user forgets or > doesn't know this.
It doesn't segfault for me. He is using a development snapshot, the equivalent of a CVS checkout. > Rather, Gnucap should do a check before > running an analysis to verify that the operating point has > already been computed and is known. If it's unknown, then > Gnucap should print out a warning like "No operating point -- > you probably need to run op". That would be an improvement, but the power off test is a valid test, and you don't need to do it for a linear circuit. It probably should print a warning whenever "ac" is done after anything other that "op", because only after an "op" will it give the same answers as Spice. If you do "ac" after "tran" it will use the last step of the "tran" as the operating point. This is extremely valuable if you are doing real analog work. I really need to write up that Class-B amplifier example to show it. Probably, it should always print a note: "using operating point xxxxxxx". > > 2. The AC line is incorrect. The "Hz" is ignored, so you > > asked for 1 step per decade, with a start frequency of 1 > > meg, no stop frequency. It assumes the stop frequency is > > also 1 meg, so you get a single point. > > Perhaps this is what caused the segfault? Again, a little > args checking prior to running an analysis would help clean > this stuff up. There is plenty of args checking. It worked correctly for me. _______________________________________________ geda-user mailing list [email protected] http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
