Trim threshold: absolute A/mm2 variant next to the relative one
Build PCM package / build (push) Successful in 6s
tests / ubuntu-latest · py3.11 (push) Successful in 54s
tests / fedora:latest (push) Successful in 1m11s
tests / ubuntu:24.04 (push) Successful in 1m22s
tests / NixOS (FHS wrapper from docs/NIXOS.md) (push) Has been skipped
tests / ubuntu-latest · py3.13 (push) Successful in 2m42s
tests / archlinux:latest (push) Successful in 55s
tests / debian:12 (push) Successful in 3m5s

The dialog threshold field gets a unit selector: % of mean |J|
(default, as before) or absolute A/mm2. The absolute variant applies
at the solved operating point - |J| scales with the test current, so
it is meant to be used with the real operating current entered as the
test current (config comment, README and dialog say so).

Switching units swaps in the other unit config default
(TRIM_THRESHOLD_A_MM2 = 1.0 for absolute) but never clobbers a number
the user typed. Validation stays per unit: 0..100 for %, > 0 for
A/mm2. pipeline.run takes trim_pct/trim_abs (at most one), compute()
mirrors that, and the JSON records threshold_mode + threshold_value
next to the resolved threshold_a_per_mm2.

Suite on dev 3.13 and the Python 3.9 mac-stack venv: 151 passed each;
dialog wiring exercised offscreen (defaults, unit swap, validation).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
janik
2026-07-24 14:21:13 +07:00
co-authored by Claude Fable 5
parent 6d8634802d
commit 8aff1c0be9
7 changed files with 146 additions and 43 deletions
+7 -5
View File
@@ -13,11 +13,13 @@ from .solver import Result
def run(problem: Problem, outdir: Path | None, show: bool = True,
i_test: float | None = None, freq_hz: float = 0.0,
contact_model: str | None = None, overlay=None,
trim_pct: float | None = None, trim_push=None) -> Result:
trim_pct: float | None = None, trim_abs: float | None = None,
trim_push=None) -> Result:
"""overlay: optional callback(stack, result) run after the solve
(EXPERIMENTAL in-KiCad overlays); its failures are non-fatal.
trim_pct: mark copper below this % of the mean |J| (None = off):
per-layer areas are printed, polygons saved to
trim_pct / trim_abs: mark copper below this threshold (% of the
mean |J| / absolute A/mm2; at most one, both None = off): per-layer
areas are printed, polygons saved to
<outdir>/low_current_copper.json and handed to trim_push, an
optional callback(trim_result) that pushes them into the board
(failures non-fatal)."""
@@ -57,8 +59,8 @@ def run(problem: Problem, outdir: Path | None, show: bool = True,
except Exception as e:
print(f"overlay push failed: {e}")
if trim_pct is not None:
tr = trim.compute(result, stack, trim_pct)
if trim_pct is not None or trim_abs is not None:
tr = trim.compute(result, stack, pct=trim_pct, abs_a_mm2=trim_abs)
print(trim.summary_line(tr))
if outdir is not None:
trim.write_json(outdir, tr)