MCAD-driven placement: apply IDF fixed positions to layout by jonfodi · Pull Request #876 · diodeinc/pcb

jonfodi · 2026-06-10T17:07:45Z

Adds a pcb-mechanical crate that consumes IDF (.emn) exports from mechanical CAD and pins MCAD-owned components at their exact mechanical pose. Claims are matched to components by reference designator, translated through a footprint-datum catalog
(mechanical/footprint-datums.toml) that maps each mechanical datum into footprint-local coordinates, and applied to the schematic as fixed, locked placements that the layout sync enforces on every run while HierPlace packs ECAD-owned parts around them.

The .emn is discovered by convention (mechanical/.emn next to the board's .zen or at the workspace root) or declared explicitly via [board.mechanical.idf] in pcb.toml. Includes an end-to-end example (examples/usbc_mcad) with a USB-C receptacle and mounting holes fixed by the enclosure.

Note

Medium Risk
Touches layout sync placement logic and introduces new resolution failures (missing datums, ambiguous IDF paths, hash mismatches); incorrect pose math could misplace MCAD-owned parts on every sync.

Overview
Adds MCAD-driven placement so pcb layout can honor mechanical CAD exports before KiCad sync.

A new pcb-mechanical crate discovers IDF board files (.emn) via mechanical/<board>.emn or [board.mechanical.idf] in pcb.toml, parses .PLACEMENT for MCAD-owned refdes, and resolves footprint origins through mechanical/footprint-datums.toml (optional footprint_hash validation). Resolved poses are written onto schematic instances as BoardPose / Instance.placement.

pcbc layout calls apply_mcad_positions after the schematic build. The layout lens treats compiler fixed_placement as authoritative: it overrides saved KiCad placement, creates new footprints at that pose (locked), re-applies pose on updates, and excludes those parts from HierPlace auto-packing (fixed footprints count as existing content). JSON netlist parsing forwards per-instance placement.

Includes examples/usbc_mcad demonstrating USB-C and mounting holes fixed by IDF while ECAD parts stay auto-placed.

^{Reviewed by Cursor Bugbot for commit 17001ce. Bugbot is set up for automated code reviews on this repo. Configure here.}

devin-ai-integration

Devin Review found 3 potential issues.

View 5 additional findings in Devin Review.

devin-ai-integration · 2026-06-10T17:14:54Z

+fn resolve_footprint_pose(claim: &IdfPlacementClaim, local: LocalDatumPose) -> BoardPose {
+    let footprint_rotation = normalize_degrees(claim.pose.rotation_deg - local.rotation_deg);
+    let theta = footprint_rotation.to_radians();
+    let cos_t = theta.cos();
+    let sin_t = theta.sin();
+    let dx = local.x_mm * cos_t - local.y_mm * sin_t;
+    let dy = local.x_mm * sin_t + local.y_mm * cos_t;
+
+    BoardPose {
+        x_nm: mm_to_nm(claim.pose.x_mm - dx),
+        y_nm: mm_to_nm(claim.pose.y_mm - dy),
+        rotation_deg: footprint_rotation,
+        side: claim.pose.side,
+    }
+}


🚩 IDF coordinate system assumed to match KiCad frame

The resolve_footprint_pose function at crates/pcb-mechanical/src/placement.rs:128-142 passes IDF X/Y coordinates directly through to KiCad nanometer positions without any Y-axis inversion. Standard IDF 3.0 uses a right-hand coordinate system (Y-up), while KiCad uses Y-down internally. The README explicitly states "IDF coordinates map 1:1 onto the KiCad sheet frame", making this an intentional design choice — the MCAD export is expected to already be in KiCad's coordinate frame. This works for the described workflow but would silently produce wrong results if a user feeds in a standard Y-up IDF export from SolidWorks/Fusion 360 without pre-transforming coordinates. Consider documenting this assumption prominently (e.g., in the datum TOML or idf_for_board docstring).

Was this helpful? React with 👍 or 👎 to provide feedback.

devin-ai-integration

Devin Review found 1 new potential issue.

devin-ai-integration · 2026-06-12T17:02:31Z

+            .get(&instance_ref)
+            .and_then(|instance| instance.attributes.get(ATTR_FOOTPRINT))
+            .and_then(AttributeValue::string)
+            .map(str::to_owned);
+        let Some(footprint) = footprint else {
+            errors.push(format!(


🚩 ATTR_FOOTPRINT constant assumes evaluator stores footprint under "footprint" key in attributes

The new constant ATTR_FOOTPRINT = "footprint" at crates/pcb-sch/src/lib.rs:51 is used by resolve_mcad_positions (crates/pcb-mechanical/src/placement.rs:49-54) to look up the footprint from instance.attributes. However, the Python JsonNetlistParser reads the footprint from instance.get("footprint_fpid") — a top-level JSON field, not an attribute. The test at crates/pcb-mechanical/src/placement.rs:188 manually adds "footprint" to attributes, so it passes regardless of what the real evaluator does. If the Zener evaluator stores the footprint under a different attribute key (e.g. "footprint_fpid"), datum-based hash validation would always fail with "has no footprint" for every MCAD claim. Verifying the evaluator's attribute key convention would confirm correctness.

Was this helpful? React with 👍 or 👎 to provide feedback.

Adds a pcb-mechanical crate that consumes IDF (.emn) exports from mechanical CAD and pins MCAD-owned components at their exact mechanical pose. Claims are matched to components by reference designator, translated through a footprint-datum catalog (mechanical/footprint-datums.toml) that maps each mechanical datum into footprint-local coordinates, and applied to the schematic as fixed, locked placements that the layout sync enforces on every run while HierPlace packs ECAD-owned parts around them. The .emn is discovered by convention (mechanical/<board>.emn next to the board's .zen or at the workspace root) or declared explicitly via [board.mechanical.idf] in pcb.toml. Includes an end-to-end example (examples/usbc_mcad) with a USB-C receptacle and mounting holes fixed by the enclosure. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

cursor

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 17001ce. Configure here.}

cursor · 2026-06-12T17:14:04Z

        return Ok(());
    }

+    pcb_mechanical::apply_mcad_positions(&mut schematic, zen_path, &resolution_result)?;


Release skips MCAD placement

Medium Severity

pcb layout applies IDF poses via apply_mcad_positions before process_layout, but release validate_build serializes the staged schematic to netlist.json without that step. Release DRC and layout sync then run against instances missing placement, so MCAD-owned parts are not fixed or updated during release the way pcb layout enforces.

^{Reviewed by Cursor Bugbot for commit 17001ce. Configure here.}

cursor · 2026-06-12T17:14:04Z

+    };
+
+    let datums = FootprintDatums::load_for_board(board_zen_path, &resolution.workspace_info.root)?;
+    let claims = idf::load_placement_claims(&emn)?;


Auto IDF parse fails layout

Medium Severity

When mechanical/<board>.emn is found by convention, apply_mcad_positions always parses it. An export without a .PLACEMENT section makes load_placement_claims fail and aborts layout, even if the file was only meant as outline data or MCAD placement is not in use yet.

Additional Locations (1)

crates/pcb-mechanical/src/locate.rs#L72-L105

^{Reviewed by Cursor Bugbot for commit 17001ce. Configure here.}

devin-ai-integration

Devin Review found 1 new potential issue.

devin-ai-integration · 2026-06-12T17:17:30Z

+    pub(crate) fn lookup(&self, idf_package: &str, footprint: &str) -> Option<&FootprintDatum> {
+        self.entries.iter().find(|entry| {
+            entry.idf_package.eq_ignore_ascii_case(idf_package) && entry.footprint == footprint
+        })


🚩 Datum lookup uses exact footprint string match, tying it to package URI format

The FootprintDatums::lookup at crates/pcb-mechanical/src/datum.rs:61-64 does case-insensitive match on idf_package but exact (case-sensitive) match on footprint. The footprint value comes from the instance's ATTR_FOOTPRINT attribute, which is a package:// URI like package://gitlab.com/kicad/libraries/kicad-footprints@10.0.3/.... The datum catalog must use the identical URI string. This is fragile if the version tag in the URI changes (e.g., @10.0.3 → @10.0.4), requiring datum catalog updates. The README mentions footprint_hash as an optional integrity check, which partially mitigates stale datums but doesn't address the matching itself.

Was this helpful? React with 👍 or 👎 to provide feedback.

devin-ai-integration Bot reviewed Jun 10, 2026

View reviewed changes

jonfodi force-pushed the fodi/mech_constraints branch 3 times, most recently from ceb841c to 83c66e7 Compare June 12, 2026 17:02

devin-ai-integration Bot reviewed Jun 12, 2026

View reviewed changes

This comment was marked as resolved.

Sign in to view

jonfodi force-pushed the fodi/mech_constraints branch from 83c66e7 to 17001ce Compare June 12, 2026 17:11

cursor Bot reviewed Jun 12, 2026

View reviewed changes

devin-ai-integration Bot reviewed Jun 12, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

MCAD-driven placement: apply IDF fixed positions to layout#876

MCAD-driven placement: apply IDF fixed positions to layout#876
jonfodi wants to merge 1 commit into
diodeinc:mainfrom
jonfodi:fodi/mech_constraints

jonfodi commented Jun 10, 2026 •

edited by cursor Bot

Loading

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

Uh oh!

Uh oh!

devin-ai-integration Bot Jun 10, 2026

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

devin-ai-integration Bot Jun 12, 2026

Uh oh!

This comment was marked as resolved.

Uh oh!

cursor Bot left a comment

Uh oh!

cursor Bot Jun 12, 2026

Uh oh!

cursor Bot Jun 12, 2026

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

devin-ai-integration Bot Jun 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

jonfodi commented Jun 10, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

devin-ai-integration Bot Jun 10, 2026

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot Jun 12, 2026

Choose a reason for hiding this comment

Uh oh!

This comment was marked as resolved.

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot Jun 12, 2026

Choose a reason for hiding this comment

Release skips MCAD placement

Uh oh!

cursor Bot Jun 12, 2026

Choose a reason for hiding this comment

Auto IDF parse fails layout

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot Jun 12, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

jonfodi commented Jun 10, 2026 •

edited by cursor Bot

Loading