For that pic I turned off display of the zone fill on both F.Cu and B.Cu to make the dangling ratsnest line from PTH 4 somewhat visible. But the area around pad 4 (error) and pad 5 (no error) is filled GND zone on both layers as indicated in this wide view:
Seems like a reasonable guess, but – this is my first board ever – I don’t know why or where it might be divided. That’s why I mentioned the nearby pad 5. Also, there’s the GND PTH way at the bottom of the 2nd pic, which also be seems to be in the same contiguous ocean of GND zone on, as indicated by the color, both Cu layers.
If two regions are not connected than KiCad will not draw a connection line between each pad from one region and each pad from second region. You will get lot of them while only one connection is lacking.
Just don’t use any autorouter (I have never tried any) and using your brain design PCB.
The easiest way to get GND connection with no problems is to have one layer destined only for GND and make all connections on the other layer. This is also the best you can do at PCB level for EMC compatibility as you give to each signal the best return path at adjacent layer.
Don’t start with freerouting. It mostly makes a mess of your PCB. The first and most important rule of PCB design for the last 30+ years is to have a good GND plane, and freerouting usually completely destroys that, and cleaning that mess up is about the same amount of work (or even more work) then routing the PCB manually in the first place.
Freerouting can be (slightly?) usable tool, but I advise you to learn about the importance of GND planes and manual routing first. It is quite often that beginners think that an “autorouter” is some kind of magic tool that designs the whole PCB layout for you, but it is not. It’s just a dumb tool that can draw tracks faster then a human can, but without the intelligence of a human being. There are many professional PCB designers who do not use auto routers at all.
It is divided by signal tracks that GND zone can’t jump over them.
In a case of 2 layer PCB destining one layer to GND makes you have to design it (except GND) practically at one layer. It can seem at first moment very difficult and it is difficult but not so ‘very’.
You should draw all tracks at top (or at bottom - your decision) and when you have to cross them you should use as short as possible track at second layer to not cut GND zone too much.
At your PCB you can probably solve the problem by adding some vias conneting GND zone at top with GND zone at bottom what will give a current a way from one this line end to the second. But it is not a good solution. The much better is to have as much unbroken GND zone at one layer as possible.
I design 2 layers board with continuous GND at one layer. When I can’t avoid crossing tracks I prefer to use 0R for it than to break GND zone. An example of such design:
All vias at this PCB are GND vias. There are 2 or 3 0Rs there.
@Piotr@paulvdh More generally… I understand that you are advising best practices, and I appreciate the time and effort you’ve taken. But the extent of my PCB ambitions is to get this board, and another very similar one for my other hand – they’re keyboards – that I personally can use, with perhaps one or two further iterations with minor changes in key switch positions.
Learning to draw some tracks manually is easier then learning to use freerouter. And for a properly designed (digital) PCB, you simply need to design it at least half properly.
And it really is not much effort to do so. Just assign one layer to the GND zone, and put all the tracks on the other layer. You can do this in less time then it takes to read that thread from 2022 with 46 posts in it.
