DRC and info panel do not agree with unconnected


This happens a couple of times now: I finished the routing, if I run DRC in the unconnected box there is nothing while in the info box it says there is 2 unconnected nets

I can try to find these unconnected nets by zooming and turn-on/off layers but the DRC should point it for me, isn’t?

This is a known bug in open gl mode.
DRC has more information than the info box and the ratsnet drawing engine.
(you should see a ratsnet somewhere where you already have made a connection.)

Normally if you restart pcb_new the info box will also say there are 0 unconnected pads and the wrongly drawn ratsnets will vanish. (It seems some stuff is not correctly updated in opengl.)

If the info box says that there is something unconnected even after a restart, you have found a new bug. (And should report it)

Also (unrelated to your question) i found a weird zone in your pcb.
The border is on a weird angle. If it would be straight the top part could connect to the rest of the zone.


I tried restarting it and the problem persist, just to be a bit paranoid I also reboot the system, but it insist in 2 unconnected nets, I will check the bug report method.
The weird zone is due a not so well draw exclusion zone, I’m trying ground planes with different ground zones to minimize contamination of the reference ground of the IC with the ground of the zone with big currents (this is a DC-DC converter).

This looks like a problem. What version of kicad are you running? Could you share your project here? (So that others could investigate if this bug persists in other kicad versions than yours.)

Kicad bug reporting:
How to: https://kicad.org/help/report-a-bug/
Bug tracker: https://bugs.launchpad.net/kicad/+bugs?orderby=-date_last_updated&start=0

If you separate gnd zones you should not have any signal crossing the gab. Otherwise you could create a gab antenna.
(I’m not sure if this applies in your case. It seems something is crossing the gab on b.cu if this is a signal and not a gnd pour it could create a problem.)

version is:
Application: kicad
Version: 4.0.5-e0-6337~49~ubuntu16.04.1 release build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1009,GCC 5.4.0,wx containers,compatible with 2.8)
Platform: Linux 4.4.0-62-generic x86_64, 64 bit, Little endian, wxGTK
Boost version: 1.58.0

files are on https://github.com/jano2358/psu

about the zones and GNDs, as a ‘general rule’ we should not separate ground planes to have a less impedance path return, but I’m trying to follow the directives of this paper: https://www.maximintegrated.com/en/app-notes/index.mvp/id/2997
or my interpretation of it :slight_smile:

Turning on/off all layers and searching a bit seems like the right info was on ratsnet, there was two tracks with ‘unconnected’ ends (track are there and they are touching each other but the end of one is not the exact point of the beginning of the next).
So DRC ‘sees’ a everything is connected and ratsnet ‘sees’ a no-so-precise connection, who is right?

Make gerbers and inspect them with viewer. If you see connection there then it is ok.

thanks Andy_P and Indir_Okanovic, but you are missing the point here, the post is not about how to solve this particular case or how to avoid the problem, but if I need to inform to the developer team if there is a bug, like its the case since ratsnet says there is connections missing (true) and DRC don’t think this is important, one is wrong and the other should say the same.
In this case I think ratnest should be more flexible with the connection points of the tracks.
I already submitted the bug.