Drone Board Integration


The live reconciliation is still running through the pin-capability, PMU calculation, and ERC fix loops. I’m waiting for that pass to settle before independently checking the final results.
Thought for 6 mins 13 sec
Awaiting 1 background task
Received notification from "Hades H753 Redesign" that it finished — picking the work back up.
The H753 redesign executor completed and the live project changed to 138 components and 107 nets, consistent with the single-MCU cleanup and PMU/interface work.
Still required
  • Independently reconcile the executor’s detailed report against the live schematic.
  • Confirm complete STM32F405 removal and validate every H753 peripheral assignment.
  • Verify PMU, external-shunt, TMP36, ESC-header, and camera-trigger calculations.
  • Run a fresh ERC fix/verify pass.
  • Confirm the ArduPilot hwdef, power budget, and bring-up files were correctly updated.
I cannot complete those verification steps because no ACUs are currently available. The executor’s completion should therefore be treated as unverified, not final schematic approval.
Reconcile redesign changes
Resolve ERC issues
Verify MCU migration
Verify interface calculations
Review project documentation

J14
J13
U7
U12
J2
J1
U10
J12
D1
Rev-A Schematic Audit
Follow-up results
Completed
  • Replaced the defective public-library BMI088 with project-local part BMI088-CORRECTED-PIN2-GND. Physical pin 2 is now GND_2, Pin Type Ground, and footprint pad 2 remains mapped to physical terminal 2. All sixteen U7 connections were restored by function and the prior no-connect ERC violation is cleared.
  • Reported the original public BMI088 library defect for correction.
  • Re-audited VBAT_RAW, J4–J7 raw ESC branches, reverse-polarity/eFuse path, AVIONICS_VBAT, 5 V, 3.3 V, sensor 3.3 V, and USB VBUS behavior.
  • Found and corrected a genuine protection error: DMT6008LFG had only ±12 V VGS rating while TPS26630 B_GATE is specified up to 14 V. Q1 is now TI's datasheet-recommended CSD19537Q3 with 100 V VDS and ±20 V VGS rating; drain/source/gate connections were restored.
  • Raised the generic LMR33640 inductor saturation requirement from 5 A to 6.5 A because TI requires Isat not below the 6.2 A maximum high-side current-limit envelope.
  • Added current and maximum-current metadata to AVIONICS_VBAT, 5 V, 3.3 V, and SENSOR_3V3. Raw propulsion VBAT remains intentionally unspecified pending PCB current-density analysis.
  • Created the detailed Power Budget.
  • Ran complete ERC, passive-component, and schematic-legibility review categories after all replacements. At 2026-08-17 16:11 UTC, no OPEN, WARNING, ERROR, or UNCERTAIN checks were returned.
Power conclusions
  • Conservative 3.3 V digital peak: 1.024 A versus MP2332 2 A rating.
  • Conservative sensor/analog peak: 60 mA versus MPQ20056 250 mA rating; estimated LDO dissipation 0.102 W.
  • Conservative 5 V peak including explicit GNSS, CRSF, ESC-signal, CAN auxiliary, and expansion allowances: 2.361 A versus LMR33640 4 A rating.
  • Reflected protected-battery current: approximately 1.118 A at 12.0 V and 1.307 A at the 10.5 V UVLO boundary.
  • TPS26630 current limit is 1.84 A to 2.16 A with the 9.09 kΩ setting, so the 2 A nominal setting is appropriate.
  • SMBJ18CA's 18 V standoff is above 16.8 V fully charged 4S voltage; 20 V minimum breakdown and 29.2 V rated clamp are compatible with the protected devices. Sustained overvoltage is cut off by the approximately 18.5 V eFuse OVP.
  • USB VBUS remains ESD-biased and sensed only. It is not tied into the 5 V rail, so it cannot back-drive the LMR33640 or battery path.
Scope status
  • Completed: BMI088 definition correction and U7 replacement.
  • Completed: Power-path audit, worst-case budget, eFuse validation, MOSFET correction, regulator/inductor/capacitor voltage/current review, USB isolation review.
  • Completed: Electrical and automated legibility validation; ERC is clean.
  • Failed — schematic spatial rearrangement: The current agent tool surface does not expose a schematic placement/arrangement mutation. Available placement functions target the PCB and were not used because this task is schematic-only. The rendered schematic remains widely scattered and does not meet the requested visible eight-region organization, despite clean automated legibility checks.
Approval decision
NOT APPROVED to proceed to PCB placement yet. The electrical architecture and power headroom are approved, but the schematic phase is not complete because the visual organization done criterion remains unmet. A schematic-arrangement-capable pass must group the eight requested regions, place support passives adjacent to served pins, and then re-run the same clean review set.
PCB-stage constraints after schematic organization
  • Size VBAT_RAW and each J4–J7 branch from actual ESC current, copper weight, temperature rise, via strategy, connector/pad ratings, and fault containment. The schematic does not establish propulsion-current capability.
  • Enforce the external connector assumptions documented in the power budget or revise the budget for the selected GNSS, CRSF receiver, CAN accessory, ESC telemetry electronics, and expansion loads.
  • Validate LMR33640 and MP2332 thermal copper, switch-node loops, inductor DCR/current rating, and effective MLCC capacitance under DC bias.
  • Keep sensor LDO and IMU return paths isolated from switching-current loops; place sensor decoupling at the pins.
  • Preserve dual SWD/SWO access and confirm boot/reset/debug accessibility after placement.
  • Route USB D+/D- as a 90 Ω differential pair with short ESD stubs and continuous return; USB remains data/sense only unless a reverse-safe power mux is deliberately added.
  • Populate CAN termination only when this board is physically at a bus end.
  • Follow-up results

  • Completed

  • Power conclusions

  • Scope status

  • Approval decision

  • PCB-stage constraints after schematic organization