I want to create a split ground plane. This would be like opening a rectangular hole in a ground plane. And put another rectangular filled region inside that rectangular hole.
In other words, it would look like concentric rectangles.
This method is mostly useful if you want to connect the inner zone to the same net as the outer zone.
If you want to connect the inner zone to another net, then just drawing the zones on top of each other and working with zone priorities as Naib mentioned may be a better solution. The inner zone should then have a higher priority, so it’s drawn first, and the zone around it will automatically keep a clearance from the higher priority zone.
In the example below I drew two overlapping zones. In the properties of the highlighted zone (on B.Cu, just as the other) I set the Zone Priority Level to 5. For the other zone, it’s left at the default (which is zero). “5” has a higher priority, so this zone is filled first, and the other zone maintains a clearance from it, just as it does with tracks from other nets, and Edge.Cuts.
Neat. I might guess that they need to have different net names. Presumably I could connect them through a 0 ohm resistor and they will still remain separated.
In my case, the regions are going to connect to each other through a narrow area around one through-hole pin. So, I might guess the software will see that as one net and loose the separation between them?
Yes, they must have different net names. If the names are the same, then the zones will just merge together. You could use a zero ohm resistor in the schematic, but that would also require you to put it on the PCB. KiCad does have a special mechanism to have continuous copper with different net names, and that is called a net tie. KiCad has both schematic symbols and footprints for net ties.
I would like to mark two solutions here, both the zone priority method and the keep-out method.
Aside, the idea of the one who asked the question being the one to decide what is correct seems counterintuitive. By definition, that is the one person who typically is not the one who knows the most about the subject and yet we ask that person to assert what is correct. It seems like a deeply flawed construct.
The other thing that seems odd is voting on truth, by a collection of experts and the opposite, one vote each.
“Perfect” is a difficult term too, it is incomplete, it needs at least one criterion. And QM pretty much guarantees any such criterion needs a range for noise.
How about, It’s an ad-hoc word. Evolution tends to produce the most optimal solution to what is ultimately one fixed criterion.
And yet the whole thing seems to have worked out for a very long time, until intelligence came on the scene.
You’ll have to use words in a separate post and then indicate that as the selected solution. A single tick box doesn’t handle all cases. See Wheeler’s aphorism.
No it doesn’t. E.g. our bodies have so many design flaws that it’s a wonder we live as long as we do. But with each generation the genes throw the dice again, to counter entropy.
True, I misspoke (or misswrote). It tends to produce a set of solutions weighted by reproductive success and probabilistic opportunity from conditions at each step, with underlying mechanisms in events such as snps and flips. The winding path of the urethra and late development of the prefrontal cortex being oft cited examples.
Actually, my purpose in indicating a correct answer is to give credit in whatever sort of system has for granting privileges for points.
So, the solution you mentioned is clever, in so far as redirecting readers to the two answers, but still bumps into at least one fundamental flaw in the design of the system overall.
And that is a disproof by example, of the aforementioned aphorism, that indirection solves all problems.
In this instance, it is built into the topology of the problem in a way that cannot be resolved by any amount of of indirection.