app
Adom Fields
Public Made by Adomby adom
Adom Fields: where the current and the heat go on your board.
main
John Lauer
0.10.3: KiCad 8 numbered nets (issue 21); same-net copper pieces kept apart (issue 22)
5ed6b7b
adom-fields 0.10.3
- KiCad 8 boards (issue 21).
- The bug: a track or via whose net is a number (
(net 41), as KiCad 8 and earlier write it, with the name only in the board's net table) was read with no net. It dropped out of every solve, and nothing said so. - The fix: numbered nets resolve through the net table, as KiCad does, and a board whose tracks or vias name a number the table does not list is refused with the count.
- The check: the same board in KiCad 8 and KiCad 10 format now solves identically, with the same vias, copper and peaks on every net.
- The bug: a track or via whose net is a number (
- Same-net copper closer than a cell (issue 22).
- The bug: two pieces of one net's copper that the board keeps apart (two fills 0.14 mm apart, joined only through other layers) were joined by the 0.2 mm grid. That invented a neck in the gap and could set the net's peak.
- The fix: each net's copper on each layer is now split into its pieces from the board's geometry (the fills, tracks, pads and via rings that touch or overlap). A solve cell belongs to the piece with most copper in it, and cells of different pieces do not conduct.
- Reporting: every spot where two pieces came closer than a cell is listed in
nets[].piecesKeptApartand logged;--pitchhalf the default resolves the copper there exactly. - The check: on a test board with two fills 0.14 mm apart joined only through the bottom layer, 0.10.2 put the peak in the gap. 0.10.3 sends the current through the vias and the bottom track, where the peak really is.
# adom-fields 0.10.3
- **KiCad 8 boards (issue 21).**
- **The bug:** a track or via whose net is a number (`(net 41)`, as KiCad 8 and earlier write it, with the name only in the board's net table) was read with no net. It dropped out of every solve, and nothing said so.
- **The fix:** numbered nets resolve through the net table, as KiCad does, and a board whose tracks or vias name a number the table does not list is refused with the count.
- **The check:** the same board in KiCad 8 and KiCad 10 format now solves identically, with the same vias, copper and peaks on every net.
- **Same-net copper closer than a cell (issue 22).**
- **The bug:** two pieces of one net's copper that the board keeps apart (two fills 0.14 mm apart, joined only through other layers) were joined by the 0.2 mm grid. That invented a neck in the gap and could set the net's peak.
- **The fix:** each net's copper on each layer is now split into its pieces from the board's geometry (the fills, tracks, pads and via rings that touch or overlap). A solve cell belongs to the piece with most copper in it, and cells of different pieces do not conduct.
- **Reporting:** every spot where two pieces came closer than a cell is listed in `nets[].piecesKeptApart` and logged; `--pitch` half the default resolves the copper there exactly.
- **The check:** on a test board with two fills 0.14 mm apart joined only through the bottom layer, 0.10.2 put the peak in the gap. 0.10.3 sends the current through the vias and the bottom track, where the peak really is.