I have been laying out a board with +12V, +5V and +3V3 nets for some weeks with no problem. But when I recently needed to update the pcb from the schematic, the +3V3 tracks have been assigned as +5V and the +3V3 footprinrts are now linked as rat’s nest.
I can’t find any issue with the schematics or the netlist. Updating the schematic from the pcb and back again has no effect.
First, please alway tell you exact KiCad version. The detailed version information is found from Help → About → Copy Version Info and can be pasted here.
Second, it’s difficult to say anything without further details. Apparently what you described isn’t normal and therefore we can’t just tell that hey, KiCad works this way, do this and that.
Have you used the ERC check, and what is the result?
Finally, many times to easiest way to find the problem is that someone can open the actual files. If possible (if it’s not confidential) you could attach the zipped project, or a stripped down version of it, here.
Quite often things like this are because of some subtle error in the schematic. For example a wire connection drawn right through a schematic symbol is sometimes hard to spot. When you see small squares that indicate open wire ends anywhere in the schematic, or junction dots that indicate wire connections in places you do not expect them you should become suspicious of that area immediately and examine it closely. In the example below I shorted pins 3 and 6 of the connector. You can barely see the wire, but the extra junction dots are quite visible.
When you use the Highlight Net function it often also becomes more clear and zooming in also helps to get more clarity.
Another method to find such shorts is to make a copy of your project, and then delete schematic symbols one by one and look at the wiring.
Another common error is to connect two nets together with labels. If two different labels are connected with a wire, then KiCad merges the nets together. ERC does issue a warning about this though.
Thank you all for your suggestions, and apologies for the delay. I found two nets on the pcb touching a through hole connector pin, one as per the schematic, the other accidentally passing on the back side.
I’m sure the DRC would have found it but relaying the tracks sorted it.