Netlist import Connects Paste Pads?

Schematic
1

I must be missing something simple here…

I have a SOT23 design that works nicely, and converts like this

(pad 3 smd rect (at -1.1 0.95) (size 1.06 0.65) (layers F.Cu F.Mask)
  (net 7 GND))
(pad 3 smd oval (at -1.1 0.95) (size 1.06 0.65) (layers F.Paste))

but I find if I import a .NET file, as a cross-check step, then PcbNew then changes the above very slightly, to give this instead ?

(pad 3 smd rect (at -1.1 0.95) (size 1.06 0.65) (layers F.Cu F.Mask)
  (net 7 GND))
(pad 3 smd oval (at -1.1 0.95) (size 1.06 0.65) (layers F.Paste)
  (net 7 GND))

and that now reports not connected F.Paste error ? (because it is NET tagged)

Somehow, PcbNew thinks a paste layer is electrical & adds the NET tag to it ?!
Is there some way to build the PAD info to avoid this behaviour ?

Addit: If I separate out the *.Mask defines, they also connect to NET, and give a DRC error.
I can manually edit the kicad_pcb file, and then DRC passes.

2

Your problem will probably sit inside the code that handles the (pad …) itself and always expects those to be connected to nets.
When I create (in BZR6971) a mask only pad it looks like this:

(pad “” smd circle (at 0 0) (size 2.75 2.75) (layers F.Mask)
(clearance 1.625))

Maybe try without the number and just “” in there?

2 Likes
3

Magic :slight_smile:
That’s the nudge I was looking for.
I can now create this, via translate

(pad 6 thru_hole circle (at -3.81 3.81) (size -0.3 -0.3) (drill oval 1.4 1.2) (layers *.Cu *.Mask)
  (net 1 +5V))
(pad 6 smd rect (at -3.81 3.81) (size 2.5 1.5) (layers F.Cu)
  (net 1 +5V))
(pad "" smd rect (at -3.81 3.81) (size 2.1 1.4) (layers F.Mask))
(pad "" smd rect (at -3.81 3.81) (size 2 1.3) (layers F.Paste))
(pad 6 smd roundrect (at -3.81 3.81) (size 3.5 1.7) (layers B.Cu)(roundrect_rratio 0.25)
  (net 1 +5V))
(pad "" smd roundrect (at -3.81 3.81) (size 3.3 1.3) (layers B.Mask)(roundrect_rratio 0.25))
(pad "" smd roundrect (at -3.81 3.81) (size 3.1 1.2) (layers B.Paste)(roundrect_rratio 0.25))

and that seems to pass DRC, and connect fills ok, and allows complete independent XYS control of all PAD shapes.
(no concentric caveats)