Raising this old chestnut again, as I am doing a bunch of ports at the moment. I’m not sure how common my experience is, but I do see companies engaging long time OrCAD designers, then realising the result is not very accessible to a wider audience and converting them to KiCad. Whether that justifies the effort necessary to make a converter work, I don’t know…
My guiding goal is to make a maintainable KiCad project that produces effectively-exact gerbers to the original. So I was comfortable redrawing the schematic by hand, since that’s pretty quick, and provides a quality starting point to an enduring KiCad schematic.
I’m now refining my PCB porting method and thought it would be worth sharing where I’ve ended up.
- All footprints have been redrawn to match the originals. Swapping to KiCad footprints would have resulted in too many differences. I was going to have to redraw the less common footprints anyway, so doing all the jellybeans as well was just an extension of that work, rather than something fundamentally new.
- Then for the layout, I make key layers visible one-by-one in OrCAD. Eg. TOP ETCH, then TOP VIA, then TOP SILKSCREEN, etc.
- Export to DXF v12, auto-generating layer names, possibly including the outline if it helps with alignment, but not bothering with polyline or fill options since I’ve not found a combination which actually results in directly useable polygons (probably even worse then those that would come from importing the gerbers).
- Import graphics in KiCad putting each DXF on a separate User layer, placing them all at the same spot, with 0.1mm line width and millimeter units.
- Trace over them using native KiCad tools (potentially using the wonderful “Create from Selection” feature recently described here for complex traces) to generate a high quality digital design that is physical equivalent to the original.
Any stories of similar experiences or suggestions on a superior workflow gladly received.