Mail Archives: geda-user/2015/10/16/11:49:47
On 10/16/2015 10:12 AM, Stefan Salewski wrote:
> For your idea of allowing prerouted manual traces: One solution would be
> to cover that traces with pads and pins, so autorouter will know that
> the area is occupied. That is some work coding it. For very special
> cases that may make sense. But generally I think that the effort is not
> really justified, and it will restrict the autorouter very much. Why do
> you think you need manually routed traces at all?
One big reason is in design reuse, or reference designs. Also radio sections
integrated by RF inexperienced system designers into a board.
Some existing circuitry will be desirable to route up to,
leaving the rest alone, but connecting to some
of its nets. Usually the pads ot traces to connect to are not a huge number for a
reusable module, so defining those as ones to connect to by some manual work
could be worthwhile. Defining a schematic that does not include the "reused or existing"
circuitry is fairly easy. If the existing layout could be put on a non-routed set of layers
then new pads or traces on the routing layers could be added with them as a reference, (on top of),
to be ready for routing. The outline to route in would be an irregular shape defined by an
outline area. The pads or traces to route to would need netname attribs.
Attribs on traces seems necessary for this.
- Raw text -