Mail Archives: geda-user/2015/10/22/13:31:18
On Thu, 22 Oct 2015, Larry Doolittle wrote:
> DJ -
>
> On Thu, Oct 22, 2015 at 12:43:15PM -0400, DJ Delorie wrote:
>> One of our old problems is "how to tell where a short really is".
>> Perhaps that problem and "where is a star ground" are really the same
>> problem? We'd just need some way of saying "we expect these nets to
>> be shorted".
>
> No, that's the wrong answer, since it's exactly the same as having
> one net. We're looking for a way to say "we expect these nets to be
> shorted _in_exactly_one_place". And an acceptable solution involves
> specifying where that place is.
>
I agree. In pcb-rnd I solve the "where the short is" problem using the
minimal cut algorithm. It basically says "let's consider combination of
cuts that would separate the shorted nets so that the problem is resolved;
now let's find the solution(s) with the minimal number of cuts". A cut is
a deletion of an object (e.g. a trace or a via).
While working on it, I enumerated cases in example inputs and found that
there are often more than one solution to the problem. Even using
different weights on edges, it's easy to construct symmetrical shorts
that will yield two or more equally good solutions. For the star topology
this means two equally good looking star points. Even if it is prepared to
accept only one short, how the algorithm would decide which one is the
good one?
I think Larry is right about PCB needs more information on the intention
of the designer. I'd go one step further: it would be cool if we could
pass generic DRC rules/considerations from the schematics to PCB, e.g.
"the distance between this pin and that pin should be less than x" or if
we somehow explain the star topology "the distance between the star point
and this pin should be less than x".
Regards,
Igor2
- Raw text -