Project Start Conversation

flux

The HY2120 Kelvin-sensing work completed, but I could not retrieve and reconcile its final report before execution ended.
Confirmed work from this session:
  • Added dedicated Kelvin branches from:
    • U6 VSS → RS1 BATT_NEG-side land
    • U6 CS/R38 → Q2 source land
  • Removed the previous remote sense branches.
  • Preserved component placement, USB routing, MOTOR_B Kelvin routing, high-current copper, and the 0.10 mm spacing rule.
  • Latest verified board state:
    • 0 airwires
    • 0 blocking DRC errors
    • 17 intentional redundant-copper groups retained
  • The last reviewed qualification still had two unresolved release gates:
    • Proving isolation of the Q2 Kelvin pickup via from Bottom/Mid-Layer-2 load-current copper.
    • Conservatively bounding the internal RS1–Q1–Q2 copper resistance and confirming positive margin below the HY2120’s 170 mV minimum overcurrent threshold.
The final executor was specifically working on those two gates, but its result was not available for reconciliation. Therefore, the board is not yet approved for manufacturing, and no new manufacturing exports were generated.
Remaining work:
  1. Retrieve and review the final Kelvin-sensing report.
  2. Confirm the 10.565 A / 500 ms hot overcurrent margin.
  3. Run an independent final ERC/DRC and protection-path review.
  4. Verify JLCPCB BOM stock and via/stencil requirements.
  5. Update firmware and bring-up documentation.
  6. Regenerate Gerbers, BOM, CPL, STEP, PDF, and IPC-D-356 files.
Execution stopped because your ACUs ran out.
D

duke0331

Sign up to send

Files

Schematic

Layout

3D View

Routing Recovery Bounded Results
Outcome
Partial cleanup completed; routing completion failed. Do not manufacture this state. All eight requested connections were attempted individually twice and remain open. No component placement, board outline, stackup, schematic, USB routing, or clearance rules were changed.
This report covers the continuation beginning with 8 airwires and 44 redundant nodes in 18 groups, not the earlier bulk-recovery batch.
Scope status

Table

Required itemStatusEvidence / limitation
Inspect eight airwires and 18 redundant groupsCompletedLive routing-state and DRC matched the starting counts.
Selective demonstrably redundant cleanupCompletedRemoved one DRC-proven floating group and five exactly coincident, unprotected duplicate vias; did not bulk-delete other redundant groups.
BMS_CELL2_SENSE C3:P1 → U2:VC7FailedTwo actual searches; both blocked-no-path.
DRV_CPH U3:CPH → C20:P1FailedTwo actual searches; both blocked-no-path.
DRV_DVDD U3:DVDD → C22:P1FailedTwo actual searches; both blocked-no-path.
DRV_VREF C23:P1 → U3:VREFFailedTwo actual searches; both blocked-no-path.
GH1_DRV R19:P1 → U3:GH1FailedFirst search timed out; reverse-direction second search returned blocked-no-path.
MOTOR_B Trace b704 → U3:SH2FailedFirst search timed out; reverse-direction second search returned blocked-no-path. Existing current-path copper retained.
MOTOR_ENABLE R15:P1 → U3:nSLEEPFailedFirst search timed out; reverse-direction second search returned blocked-no-path.
MOTOR_FAULT_N R16:P1 → U3:nFAULTFailedTwo actual searches; both blocked-no-path.
Fresh routing/DRC checksCompletedEight airwire errors and twelve redundancy warnings remain; no other requested routing violations returned.
USB physical connectivity/preservationCompletedAll four USB data nets absent from final airwire list; protected connector breakout traces remain present and untouched. This is not a fresh impedance certification.
MOTOR_B load copper preservationCompletedRead-back matches starting geometry; no MOTOR_B copper was removed or changed.
MOTOR_B Kelvin continuityFailedSH2 remains disconnected. The historical Mid-Layer-2 branch is not present in the live MOTOR_B node inventory inspected at the start or end.
Exact obstacle-by-obstacle proof for every failureIncompletePathfinder returns no blocking object IDs or collision geometry. Local restrictions are documented below, but they do not establish an exhaustive geometric impossibility proof.
Changes and counts
  • Deleted 6 traces, 5 smart vias: 11 persisted copper-node deletions total.
  • Six floating LIDAR_RX_CONN traces: 42d74d47-861e-4e9b-adbd-c142213532f2, 4822c23c-031f-4b34-9af2-1557ac71bb0b, 5f7f4d4a-3db3-4a6a-a063-48ac90b1fb29, 7ac37837-c151-4099-8ebe-b7bb12915cc4, 7d22ebb5-c6f3-4d65-ad86-00da22da6419, f89e7e4a-a452-4aea-9616-1295f389edfc.
  • Five removed vias had exact same-net, all-layer duplicates retained at the same coordinates. Removed unprotected instances only:

Table

NetPosition, mmRemoved UID
GPS_RX(-11.6758, -7.4023)e67a10c4-646a-47da-83d5-8bed7fc5395f
SYS_GND(10.0500, 0.2250)71f86fe9-6c68-48c0-8d50-c09e356b2df6
SYS_GND(10.0500, -2.7000)b1343b6c-35bf-48aa-8472-10d1dff127d8
AUTOPROT_CHG_DRV(5.3400, 22.2250)a6c1fe74-6e40-4b3d-9663-539628da559c
AUTOPROT_CHG_DRV(8.0400, 17.4250)adf9489d-f1a6-4c61-b5ef-e1002a5a78fd
  • 0 successful targeted connections, 0 new routed traces/vias retained.
  • Neckdown regeneration reported 2 generated neckdowns, but increased airwires 8 → 14, introducing three BMS_CELL_MID_SENSE and three U1 SYS_GND airwires. The single generation commit #5618c580 was atomically reverted to #a565292b, restoring 8 airwires while keeping the selective cleanup. Net new neckdowns retained: 0. Do not repeat unrestricted regeneration as a finishing strategy without investigating this behavior.
  • Component moves: 0. Rule changes: 0. Full-board autoroute calls: 0.
Bounded targeted attempts
First attempts used 30 s search caps, 0.15 mm signal width, and 0.254 mm for the MOTOR_B Kelvin connection. Second attempts reversed terminal order and used 120 s caps at the same widths. The second BMS_CELL2_SENSE attempt occurred after the neckdown regeneration; the generation was subsequently rolled back after the additional airwires were discovered. No third attempt was made on any net.
The exact final search result for each was:
No path exists: the search explored every reachable cell — change the board (move parts, rip or reroute traces in the way, add routing layers) and retry.
This proves failure of the available targeted pathfinder in the tested configuration, not an exhaustive proof that no manual or active-route rip-up solution exists. Three first attempts reported time-limit at 30 s: GH1_DRV, MOTOR_B, MOTOR_ENABLE. All other first attempts and all second attempts reported blocked-no-path.
Geometric observations and safe next actions
Coordinates below are board-world millimeters. They are measured local access restrictions or nearby copper, not asserted complete cut-set proofs. No active functional routes were deleted under the authorization to remove only stale redundant copper.

Table

NetPin location and inspected neighborhoodSafe next action requiring expanded repair scope
BMS_CELL2_SENSEU2:VC7 approximately (12.000, 6.273), between BAT at x=11.500 and VC6 at x=12.500. Existing BMS_CELL_MID_SENSE custom copper extends to x=12.323 and down to y=4.975, alongside the VC7 escape corridor; BAT trace runs south at x=11.500.Inspect/manually plan a VC7 escape and via while preserving adjacent cell-filter geometry; determine whether a local portion of active BAT/mid-cell fanout must be reconstructed. Do not rerun global neckdown generation.
DRV_CPHU3:CPH (-8.950, 14.250). VCP runs from (-9.346,14.750) diagonally to (-9.725,14.371), then west; CPL approaches the neighboring pad at y=13.750.Review a short CPH/CPL/VCP via escape together; approve surgical rerouting of the active charge-pump fanout if needed, not a long detour.
DRV_DVDDU3:DVDD (-7.750,12.050), between AVDD (-8.250,12.050) and nFAULT (-7.250,12.050); center ground pad bounds the inward side.Inspect a local south-side escape/via against existing AVDD and ground/decoupler copper; obstacle identity beyond this neighborhood was not established by the tool.
DRV_VREFU3:VREF (-8.950,12.750); active CPL runs vertically at x=-9.727 from y=12.509 to 13.570. Its right copper edge is approximately x=-9.600, while the VREF pad outer edge is x=-9.300: approximately 0.300 mm gap, smaller than 0.150+2×0.100=0.350 mm needed for the tested trace between two foreign boundaries.Consider approved reconstruction of the local CPL escape, retaining a short flying-cap loop, to open VREF access. Other escape possibilities need manual verification.
GH1_DRVU3:GH1 (-5.050,14.750). MOTOR_IN1 runs vertically at x=-4.376928, y=13.882576…16.776354, width 0.254 mm. Its left edge x=-4.503928 is only approximately 0.196 mm beyond the pad's x=-4.700 outer edge.Approve moving this active MOTOR_IN1 segment farther outward or to another layer, then plan a short GH1 escape; do not move the MOSFETs or gate resistor without approval.
MOTOR_BU3:SH2 (-7.750,15.950), between GH2 and GL2 escapes at x=-8.250 and -7.250. Destination current copper remains Trace b704, (-14.000,22.260)…(-14.000,25.300).Plan a dedicated sense-only via/fanout from SH2 around the top gate-drive pad row; reconnect to the existing drain/current island using a Kelvin branch, without replacing load copper. Exact blocking object not proven.
MOTOR_ENABLEU3:nSLEEP (-5.050,12.750), immediately below MOTOR_IN2's y=13.250 escape and above IDRIVE routing near y=12.050; via neighborhood includes (-3.745,12.587) and (-4.399,12.337).Inspect/reconstruct an approved portion of the active IN2/IDRIVE fanout to create a small signal escape if necessary; exact cut-set not proven.
MOTOR_FAULT_NU3:nFAULT (-7.250,12.050), between DVDD and SNSOUT south-row pads; exposed ground pad bounds the inward side.Inspect a coordinated south-row nFAULT/DVDD escape rather than independently repeating routing. Exact blocking object not supplied by pathfinder.
No placement move or relaxed clearance is authorized or applied. The appropriate next step is an obstacle-level manual routing review with permission for narrowly scoped active signal-fanout rip-up/reconstruction, not another identical targeted pass or blanket copper deletion. Preserve USB and load-current copper throughout.
Final checks and preserved constraints
Full live DRC at 2026-09-18T17:12:09.764Z returned:

Table

FindingCount
Airwire errors8
Redundant-copper warnings12 groups / 33 nodes
Other returned DRC errors0
A further focused check at 2026-09-18T17:15:18.840Z returned no findings for overlapping copper, overlapping traces, floating copper, dangling traces, under-connected vias, pad-access clearance, protected intrusions, keepouts, or invalid layers. This means zero reported findings in those checks; it is not an independent measured minimum-distance or manufacturing sign-off.
The active Auto-Layout: Universal Keepout rule remains minKeepOut=0.1mm, keepOut=0.1mm; Recovery copper hard spacing floor remains minKeepOut=0.10mm. D1 local minimum remains 0.1 mm. Neither spacing nor trace-width rules were relaxed. Board remains 40 × 75 mm, Standard 4 Layer.
USB verification
  • USB_DP_CONN and USB_DN_CONN each retain the two J4 data contacts and their D1 connection; USB_DP and USB_DN each retain D1-to-U1 connectivity.
  • No USB airwires were reported before or after this batch.
  • Read-back contains the explicitly protected 0.25 mm connector breakout traces on both polarities, plus the existing continuation copper. No USB node was targeted for deletion, routing, or rule changes.
  • Existing routing includes redundancy; preserving it does not certify its end-to-end 90 Ω performance. No new impedance or signal-integrity solution was performed.
MOTOR_B verification
  • Preserved live 1.6002 mm Mid-Layer 1 current-path segments, 1.0 mm Top Trace b704, 0.54 mm Bottom segments, and the existing via at (-15.5015,25.7715).
  • Geometry matches the initial read-back; no MOTOR_B mutation succeeded.
  • The Kelvin connection is not intact electrically: U3:SH2 remains one of the eight airwires. The historical 0.254 mm Mid-Layer-2 Kelvin branch described in prior notes does not appear in the current net inventory; do not report it as restored or intact.
Intentionally retained warnings
Preflight warnings (exact codes)
Four ROUTING_OVERLAPPING_PINS_SAME_NET at J4 remain: two VBUS pairs and two GND pairs. The router's wording is that the pins "are overlapping and belong to the same net" and "will be considered as one single pin." These fixed connector overlaps were not modified.
Five ROUTING_DUPLICATE_VIA warnings were resolved by the five verified duplicate-via deletions.
Two informational ROUTING_PIN_NEEDS_VIA_AT_SMD remain for GND_40cac9af083b252ce2e96561c4ba6154 and EPAD_af741d1ff2cf656be674e0f6d069054d.
Redundant groups retained
Live redundancy reports four USB-net groups, not merely the historically cited three protected groups. All USB copper was retained, whether explicitly protected or not. The other eight groups were also retained because no specific node was demonstrated to be the obstruction requiring deletion under this request's selective-cleanup constraint.

Table

NetGroup identifier / first nodeNodes
USB_DN_CONNTrace 134e (134e6c53…)8
USB_DP_CONNauto-layout-944ef6d7… Trace 369eec739c1f757b4bc0f8e75d6bcb3e5
USB_DN_CONNauto-layout-944ef6d7… Trace c540e4bcbb4c97eebd933a8c837729901
USB_DP_CONNTrace e5af (e5afe431…)1
SYS_GNDauto-layout-01867a4c… Trace 04cfd2469396747f4c9fab8086e9b12b3
SYS_GNDauto-layout-01867a4c… Trace 0a01d1aa8cd1563f6c3c97232391c4d79
SYS_GNDauto-layout-01867a4c… Trace e43914930dad9dba9cea090beb5afc731
SYS_GNDauto-layout-675aa2a9… Trace 850b8da6bec1e56d0bd9b78b051a659a1
GH2_DRVauto-layout-944ef6d7… Trace 36691bf77a1fd025471b77e39ee929d81
BMS_ALERT_Nauto-layout-944ef6d7… Trace caf40deb6009d99e646653c2a6ef92061
GL2_DRVauto-layout-944ef6d7… Trace 71207c69b0093d51ab5f3dc65a564c601
SYS_GNDauto-layout-944ef6d7… Trace 4a7bb742a348ce8beaf58d652a2d6d1a1
The review's generic item descriptions are "Trace 134e has an issue", "Trace e5af has an issue", and "Trace auto has an issue"; the review category is Redundant Traces — Warning.
Tool limitations
  • Editor-focus request failed because no requester frontend session was found.
  • Internal feedback submission about neckdown regression and missing obstacle diagnostics failed; no ticket/report was filed.
  • Exact per-net geometric blocker proof remains incomplete. Do not turn the pathfinder's generic failure into a fabricated specific obstruction or assert that placement must change.

  • Outcome

  • Scope status

  • Changes and counts

  • Bounded targeted attempts

  • Geometric observations and safe next actions

  • Final checks and preserved constraints

  • USB verification

  • MOTOR_B verification

  • Intentionally retained warnings

  • Preflight warnings (exact codes)

  • Redundant groups retained

  • Tool limitations