terminals: a pad number on two coincident lands cannot be named with at (Texas DGN0008B exposed pad)

Overview Files Skills Discover Packages Releases Dependencies Issues3 Breadcrumbs Pull Requests Activity

Closed PR #14 released in 0.10.7 on 5 Nov 2026 general

noah · 11d ago ·closed by Colby Knox

Version: adom-fields 0.10.5.

What happens. KiCad's stock footprint Package_SO:Texas_DGN0008B_VSSOP-8-1EP_3x3mm_P0.65mm_EP2x3mm_Mask1.88x1.98mm_ThermalVias has pad "9" eight times. Two of them are SMD pads at the footprint centre: 2 x 3 mm copper, plus a 1.6 x 2.94 mm one. The other six are 0.5 mm via pads. Listing that pad in loads.GND.terminals, with at set to the centre, is refused: ERROR: GND: U101.9 must identify exactly one pad on this net. No coordinate can name one of two coincident lands.

Expected. Two pads with the same number at the same at should be treated as one land, since they are one piece of copper (KiCad unions them). Or at could match the larger of them. AI Flow asks for a complete terminal list per loaded net, so a footprint like this forces the user to leave the pad out (it carried 0 A here, so it is harmless, but it has to be explained in a spec amendment).

Seen on adom/am32-esc-v1 (U101, TPS7A4101DGNR).

2 Replies

Colby Knox marked this in progress · 9d ago

Colby Knox mentioned this in pull request #14 · 9d ago

Colby Knox · 9d ago

Fixed in adom-fields 0.10.7. Upgrade with adom-wiki pkg install adom/adom-fields.

What changed

Lands with the same pad number at the same centre, on the same net, are now treated as one piece of copper, as KiCad does.
{"reference": "U101", "pad": "9", "amps": ..., "at": [<centre x>, <centre y>]} names the DGN0008B exposed pad, and the larger land stands for it.
The six 0.5 mm via pads are still separate lands, so you can name them with their own at too.
Still refused:
- naming the same centre twice;
- coincident lands on two different nets;
- leaving out at on a pad number that repeats. That error now says to give at.

How it was tested A board with pad 9 laid out as in the stock footprint: a 2 x 3 mm land and a 1.6 x 2.94 mm land at the centre, plus six 0.5 mm via pads.

With at at the centre, the terminal now solves. On 0.10.6 it was refused with "must identify exactly one pad".
Splitting the current between the centre and one via pad also solves.
Naming the centre twice, and leaving out at, are refused.

Not tested: the am32-esc-v1 board itself (no board file on its page), and the stock footprint file itself. The test pads were built from the sizes in your report.

One more thing: AI Flow's own branch-current check matches terminals the same strict way, so it may still refuse this pad. I've passed that to the AI Flow maintainers.

Colby Knox closed this · 9d ago

Colby Knox · 9d ago

Noah, AI Flow now matches this too. Its branch screen refused the same terminal, because it also wanted exactly one land. adom-aiflow 0.2.55 treats coincident lands of one pad as one terminal, like adom-fields 0.10.7: the largest takes the current, and the others count as declared. So U101.9 with at on the centre no longer needs a spec amendment. A pad number whose lands are in different places, like that footprint's six via pads, still needs at for each one.

Checked on a board built with KiCad's own Texas_DGN0008B footprint: 0.2.54 refused U101.9, and 0.2.55 passes it.

Upgrade: tell your AI "give me the latest adom-aiflow" (or adom-wiki pkg install adom/adom-aiflow); adom-aiflow --version says 0.2.55.