[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: gEDA-user: gnetlist errors



Stuart Brorson wrote:

Stuart,

I've reviewed the tutorial and am following that flow - partially.
What I'm really doing is porting a bunch of layout and schematics
from another CAD tool into gschem. I'm not actually drawing
schematics at this time (but I soon will).



Hmmm. . . Can you tell us which tool you are leaving, and how you are
doing the import? Is is sarlacc? I'm curious to know how easy this
is. Thanks.




After putting everything in a clean directory and correcting a typo
pointed at by a previous post (Thanks!) I now have a usable
netlist.



Great!


Stuart




Stuart,

I'm not familiar with sarlacc...

We formerly used tools from Ivex. Windraft was their schematic editor and Winboard their layout editor. After receiving decreasing levels of support (next to nothing actually), we began looking for other solutions. Ivex eventually closed up shop altogether. The tools still work since the licenses were not time limited. We first ported the layouts to PCB and now are moving the whole design flow from schematics on down to gEDA/PCB.

My first step was to build a perl script that translated netlist output from Ivex into something PCB would read. Of the dozen or so formats Ivex supported I used "P-Cad NLT" since it looked like the easiest translation.

The schematic translation was quite a bit more work. I used their "decompiled" schematic and library format, which is ASCII and looks like EDIF, although I'm not sure it is 100% compliant. In any case, there are two parts to the porting process. One perl script translates the library, outputting a gschem compatible symbol file for each Ivex part definition it finds. The second script translates the actual schematic page. To reduce the number of possible errors in the end result I wanted to insure that very little manual cleanup would be required. So far, that goal has been met.

Altogether we needed to port about 100 schematic symbols, and a dozen moderately dense schematic pages. PC footprints were either built or leveraged from existing elements already in the PCB libraries. We employ some not-so-common parts so I don't make any use of m4-generated footprints.

This involved more time than I would have liked, primarily because I'm a very novice perl programmer. I'm sure that someone who knew what they were doing would have written much more elegant/efficient perl code and done it in much less time. Using yacc to do the schematic translation might be another option.

Does this answer your question?

Joe T