ID: | 5121 |
From: | rn...@igor2.repo.hu |
Date: | Mon, 13 Sep 2021 15:36:29 +0200 (CEST) |
Subject: | Re: [pcb-rnd] new: route-rnd 0.9.0 |
in-reply-to: | 5119 from Evan Foss <ev...@gmail.com> |
replies: | 5122 from rn...@igor2.repo.hu , 5123 from rn...@igor2.repo.hu |
Hello Evan, On Mon, 13 Sep 2021, Evan Foss wrote: >http://igor2.repo.hu/tmp/route-rnd/0.9.0/2dip_crbs.png > >What is going on under pin 1 in that image with the top layer of >copper? It looks like a tiny circle. Yup, you've spotted the bug that bothered me too. It's not a complicated routing bug but some technical bug about import/export. Well, crbs is not stable yet, so as advertised: do not trust it. Backround: The data model for the geometric rubber band sketch is rather complicated. It contains lines and stacks of convex and concave arcs arranged in sectors around points. Normally, when academic people approach these problems, they do some simplifications. They typically say there are only round pins/vias on the otherwise totally empty and clean board, then the rats nest. So lines, arcs and points ("vias") are enough. However, I absolutely wanted to have trbs and crbs to be able to live with existing copper objects and handle non-round copper pads properly. So what I do with the rectangular pad, which is technically a copper polygon in the transport format, is replacing it with 4 points in the 4 coners and add 4 pre-drawn, thin lines at the polygon edges - so the polygon does not need to be handled in the grbs data model. What you see is probably some miscommunication of one of the corners: it does not show on any internal routing dump, only after importing the route results in pcb-rnd. To make it funnier, it's visible only in the png export, it's not visible on screen! (This may be a pcb-rnd bug on some arc corner case unseen before.) Best regards, Igor2
Reply subtree:
5121 Re: [pcb-rnd] new: route-rnd 0.9.0 from rn...@igor2.repo.hu
5122 Re: [pcb-rnd] new: route-rnd 0.9.0 from rn...@igor2.repo.hu
5123 Re: [pcb-rnd] new: route-rnd 0.9.0 from rn...@igor2.repo.hu