Do not start auto-route yet. Bounded preparation reduced live preflight anomalies 63 → 45 → 30 over three checks / two corrective passes. Every check retained 110 components / 71 nets. All remaining preflight findings are engine Warnings, not engine Errors, but the engine says the affected pins will be excluded from routing: they are routing-readiness blockers, not waived advisories.

No auto-route or surgical routing was started. No component position, rotation, side, board dimensions, silkscreen, schematic wiring, MPN, package property or footprint geometry was changed. Only eleven new width rulesets and deterministic neckdown copper were authored. Manufacturing profile and physical stackup were preserved.

Source requirements are from existing power-budget-design-notes and project-specification, not freshly certified protection tolerances. The former has stale L5/L6 entries; live BOM takes precedence for inductors. The specification's old 70-net count is stale; live preflight is 71.

Width screening used IPC-2221 external-conductor relation I = 0.048 × ΔT^0.44 × A^0.725, with A in mil², 35 µm copper and 10 °C rise. These are conservative planning widths, not IPC-2152 thermal validation, guaranteed fault-current ratings, or voltage-drop signoff. Apply these widths to outer-layer power conductors; inner-layer power routing needs fresh sizing. Short fine-pitch neckdowns require local copper spreading, RMS/temperature and via-current review after routing. Wide traces do not validate the converter's loop geometry.

Low-current widths set to 0.20 mm, without reducing the 0.20 mm manufacturing floor:

All other nets retain existing effective defaults. No schematic Current properties were changed. Bootstrap/bias widths above apply to short local connections, not external load supply rails.

All rules only set traceWidth; none changes clearances or geometry.

U6 is TPS61088RHLR, source part b6114485-fe8d-4e20-b639-69efba540c98, placed at (0, −14) mm. Its electrical mapping agrees with datasheet pin assignments: VCC 1, SS 10, NC 11/12, AGND 20 and PGND 21.

Do not diagnose from the 2D rectangle alone. Current layout_2d renders PGND as 1.6 × 1.6 mm, but published source scene_tree exposes a custom SVG path:

The evidence favors a custom-pad bounding-box / clearance-rule artifact, not a proven physical pad short. This is not yet a definitive router-conversion diagnosis: the exact contour passed to the routing engine is not exposed here, and published-source geometry alone does not certify the consumed pad conversion. Broad pad clearance rules are also incompatible with the observed gaps: Pad clearance (NPTH proxy) matches every pad with keepOut 0.40 mm/minKeepOut 0.30 mm; Pad-to-pad spacing sets pad(0.50 mm); Hole-to-trace spacing also matches every pad/via. Those rules were preserved, not silently relaxed below the applied manufacturing profile.

The cited datasheet identifies its land-pattern drawing as RHL0020A. The initial extraction supplied numerical thermal-pad and center-spacing claims but failed support validation; those numbers were rejected. A narrowed query did not establish the custom fingers' dimensions. Thus a rectangle replacement, enlargement, shrink, or custom-path rewrite would be an unsupported package correction. No pad edit was made. A source/CAM contour review and a manufacturing-profile/clearance decision are required before declaring U6 safe. Internal diagnostic feedback was attempted but submission failed; no ticket was filed.

The active footprint names match live Package metadata and the prior BOM lockdown report; no fallback chip-package substitution was observed. Nevertheless, the 3.35 × 4.20 mm MAPI drawing envelope and L5 8.50 × 6.60 mm envelope are not proof of actual body-size correctness. L3 datasheet extraction verifies one body-width dimension as 4.1 ±0.2 mm, but did not resolve the orthogonal body length or numeric pad recommendations. This is enough to flag the small-looking envelope for review, not to certify or replace the land pattern. L5 numeric body/pad callouts also remained untranscribed. No reliable prior screenshot/version dimensional baseline was provided, so an actual historical shrink is unverified.

L3's datasheet also explicitly calls for no vias or traces underneath the inductor. This manufacturer layout constraint needs review before final routing; no new body keepout was guessed without complete dimensional evidence. Do not assume the generic footprint implements it automatically.

There are 1 grouped overlap warning + 7 width-exceeds-pad warnings + 22 inaccessible-pin warnings. All remain engine Warning severity.

U6 AGND, NC_1, PGND, SS, VCC across GND / AUX12_SS / AUX12_VCC. See custom-pad diagnosis above; do not move U6 as a supposed solution to an intra-footprint representation issue.

These require effective local pad transitions / escape copper after rule and footprint validation. The global power widths must not be reduced merely to clear these warnings. Existing R4 pad minTraceWidth 0.15 mm override was preserved; it did not make this pin routable, and it is not blanket authorization to route below the manufacturing floor.

These survive concrete width/neckdown correction. The preflight does not identify an exact obstructing neighbor for each pin, so they are fixed-placement escape blockers, not proven placement-only defects. Manufacturing pad/clearance rules and intra-package escape geometry must be resolved before proposing any component move. No speculative direction/distance is claimed.

Fresh selected get_checks review at 2026-09-15T23:01:10.383Z reports six Switching Power Stage warnings and no returned findings for selected overlapping-copper, pad-access, stackup, missing-footprint and neckdown reviews. This narrower review does not overrule the live preflight's 30 pin exclusions.

The six explicit placement-dependent advisories are:

Bounded preflight preparation is finished, overall routing readiness is not. Preserve the 46 generated neckdowns and justified widths. Resolve the U6 custom-contour conversion/clearance conflict, manufacturing-profile suitability for fine-pitch escape, inductor dimensional/under-body constraints, and the exact residual local transitions before starting auto-route. A different manufacturing profile requires explicit confirmation because it can replace existing rulesets. No additional routing or placement mutation was attempted after the third preflight check.