On Aug 13, 2007, at 11:45 PM, Felix Fujishiro wrote: > On Mon, 13 Aug 2007 07:43:42PM -0700, Ben Jackson wrote: > > >On Mon, Aug 13, 2007 at 07:33:40PM -0700, Felix Fujishiro wrote: > >> > >> I tried to do this by adding the 'net=GND:4' attribute, and that > works, but > >> only if pin #4 is visible. > > > >I'm not sure what you mean. My Altera FLEX board used a chip-shaped > >symbol with all the power and gnd hidden using net= attributes. > There > >are no pins for those anywhere. > > I guess I'm not being clear.... I'm using gschem for custom digital > IC design, perhaps there's a difference in the way the tool treats > MOS components and PCB components? If I remove pin #4 (the > substrate/well connection) from the MOS symbol, gnetlist spits out > a three-terminal device instead of a four-terminal device (which I > want). >
Which back end are you using? Perhaps spice-sdb? If so, I think I understand what's going on. Spice-sdb assigns pins by their pinseq= attributes, while the net= attribute works on pin number. Without a pin, you have no place to hang a pinseq= attribute, so spice-sdb won't understand that it needs to make a connection. If you're not adjusting transistor parameters much you could make three terminal .SUBCKT models, each containing a single transistor, with the substrate connected to a global node. Use model-name= (and maybe file=, depending on how you organize things) to associate the symbol with the subcircuit. The problem here for me would be that I do mixed signal VLSI and I'm often tweaking individual transistor parameters (L, W, M). In the subcircuit approach, you need a different subcircuit for every geometry. So for my work I just use the four terminal symbols and connect the substrate node explicitly. I have very small VSS and VDD symbols for this purpose. And I generally just box select and copy components in preference to the dialog, so I can copy the transistor+substrate combo together. John Doty Noqsi Aerospace, Ltd. http://www.noqsi.com/ [EMAIL PROTECTED] _______________________________________________ geda-user mailing list [email protected] http://www.seul.org/cgi-bin/mailman/listinfo/geda-user
