Drone Board Integration

flux

Not safely as-is. The Instructables design may teach basic STM32F030 sensorless six-step commutation, but it provides no verifiable 4S voltage margin, 20 A continuous thermal data, MOSFET/gate-driver specifications, protection behavior, or PCB current-path validation.
For this board, use ST’s STEVAL-ESC002V1 as the ESC baseline instead:
  • 2S–6S LiPo
  • 20 Arms rated output
  • STSPIN32F0A integrated MCU/gate driver
  • Six 60 V, 2.8 mΩ STL140N6F7 MOSFETs
  • Sensorless six-step commutation with available firmware
  • BEMF and battery-voltage sensing
  • Hardware overcurrent protection
  • Published schematic, BOM, and Gerbers
Even that reference needs adaptation and testing for 20 A continuous / 30 A burst, DShot/CRSF integration, shared-PDB transients, cooling, and four channels operating simultaneously. The IWDF guidance reinforces separating the ESC switching loops from the FC/IMU/RF region, using heavy copper and thermal vias, local bulk capacitance, uninterrupted reference planes, filtered low-noise rails, and accessible test points.
My recommendation is to treat the Instructables circuit as a learning reference only and derive each of the four ESC channels from STEVAL-ESC002V1.
Develop ESC reference channel
Compare ESC reference designs
Document ESC adaptation
Define 8-layer stackup
Sign up to send
No schematic yet

Talk to Flux to get started.

Refine this doc
Ask about this doc
Architecture Decision Guide
1. Choosing Between 4S and 6S
A LiPo cell is about 3.7 V nominal, 4.2 V fully charged, and typically treated as depleted around 3.2–3.5 V under load.

Table

PackNominalFully chargedPractical design range
4S14.8 V16.8 Vroughly 12–16.8 V plus transients
6S22.2 V25.2 Vroughly 18–25.2 V plus transients
For the same mechanical output power, raising voltage reduces current: P = V × I. A motor system needing 600 W from the battery would draw approximately 40.5 A at 14.8 V or 27 A at 22.2 V before losses. Lower current reduces connector loss, copper loss, and voltage drop because conductor heating follows I²R.
4S advantages
  • Lower voltage stress on MOSFETs, gate drivers, capacitors, TVS devices, and regulators.
  • Easier first ESC design and generally more forgiving switching behavior.
  • Wider selection of 25–30 V-rated components, although propulsion transients still require significant margin.
  • Appropriate for smaller motors and a lower-risk educational prototype.
4S disadvantages
  • More current for equal power, producing higher PDB, connector, shunt, via, and copper losses.
  • Larger copper cross-section or parallel conductors may be needed.
  • Greater voltage droop at the flight controller during acceleration.
6S advantages
  • Lower current for equal power, improving distribution efficiency and reducing I²R losses.
  • Common for higher-performance aircraft where current and wiring mass matter.
  • Better opportunity to study high-voltage switching, transient control, and power integrity.
6S disadvantages
  • A 25.2 V charged pack leaves almost no safe margin for 30 V MOSFETs or capacitors. Propulsion stages normally move to at least 40 V-rated MOSFETs and suitably derated 35/50 V capacitors, subject to measured overshoot.
  • Switching-node ringing and avalanche/transient risk become more severe.
  • Avionics bucks need higher input ratings and careful hot-plug/surge design.
  • Layout and probing mistakes are less forgiving.
Recommendation
For the first integrated prototype, use 4S with a modest per-motor current target unless the intended motors require 6S. Design the architecture so a later 6S revision is possible. If the primary learning objective is demanding power integrity and the user accepts a harder bring-up, choose 6S but do not select any power-stage part until continuous current, peak current, switching frequency, cooling, and measured/estimated transient limits are defined.
2. ESC Architecture Options
An ESC commutates a three-phase brushless motor by switching six MOSFETs arranged as three half-bridges. The control system measures or estimates rotor position, applies PWM, protects the stage, and receives throttle commands such as DShot.
Option A: Dedicated MCU per ESC channel
Typical organization: one small motor-control MCU, one three-phase gate driver, and six MOSFETs per motor.
Strengths
  • Each motor has independent timing, fault handling, and firmware.
  • Closest to established 4-in-1 ESC practice.
  • Faults and processor loading are naturally partitioned.
  • Supports mature commutation approaches and high PWM update rates.
Weaknesses
  • Four additional MCUs, programming interfaces, clocks or clock strategy, firmware images, and support passives.
  • Greater BOM and board area.
  • BLHeli_32 is not generally an open architecture that can simply be adopted for arbitrary new hardware; licensing and supported-target constraints must be checked.
Best fit: a reliable, conventional 4-in-1 ESC and a strong firmware/layout learning project.
Option B: One shared motor-control MCU for all four motors
A capable STM32G4-class controller can generate many complementary PWM outputs and sample multiple currents/voltages.
Strengths
  • Lower processor count and potentially easier coordinated control.
  • STM32G4 timers, fast ADCs, comparators, and math hardware are designed for motor control.
  • Enables fully custom commutation or field-oriented control experiments.
Weaknesses
  • High firmware complexity and strict deterministic timing.
  • One MCU failure can disable all motors.
  • ADC triggering, DMA, timer allocation, and protection become dense.
  • Routing PWM and analog feedback between four spatially separated arms can increase noise susceptibility.
Best fit: advanced custom motor-control research after a single channel is proven.
Option C: Direct ESC control from the STM32H753 flight MCU
The H753 generates all gate-control PWM and performs flight control.
Strengths
  • Minimum MCU count.
  • Tight integration between control loops.
Weaknesses
  • Poor fault containment and severe real-time complexity.
  • Motor switching and ADC timing compete with flight-control tasks.
  • Firmware errors can simultaneously compromise stabilization and commutation.
  • Creates difficult pin, timer, interrupt, and safety constraints.
Recommendation: do not use this architecture for the first board. Let the H753 send DShot or another command protocol to a separate ESC controller layer.
Option D: Smart three-phase drivers or integrated power modules
Use a driver/module that integrates substantial gate-drive, current sensing, protection, or MOSFET functionality.
Strengths
  • Faster hardware development and stronger built-in protection.
  • Reduced gate-drive layout risk.
Weaknesses
  • May not support the voltage/current/performance required by a drone.
  • Less exposure to discrete high-current layout practices.
  • Cost, thermal performance, and sourcing can dominate.
  1. Build and validate one ESC channel with a dedicated motor-control MCU, three-phase gate driver, six external MOSFETs, current/voltage/temperature sensing, hardware overcurrent shutdown, and a local debug header.
  2. Validate it with a current-limited bench source and motor test fixture.
  3. Replicate the proven channel four times.
  4. Let the H753 send DShot commands; do not commutate motors directly from the flight-control processor.
A practical first choice is one small STM32G0/G4-class MCU per ESC channel, selected only after timer, ADC, firmware, and package requirements are finalized.
What LoRa is
LoRa is a physical-layer radio modulation based on chirp spread spectrum. Instead of transmitting a narrow, fast symbol, it sweeps frequency in chirps. The receiver can recover very weak signals, giving a high link budget and long range at the cost of reduced data rate and greater airtime.
Key settings include:
  • Bandwidth: narrower bandwidth usually improves sensitivity but lowers throughput.
  • Spreading factor: higher spreading factor increases time on air and sensitivity but lowers throughput and update rate.
  • Coding rate: adds error correction at the cost of airtime.
  • Transmit power and antenna gain: affect range but are limited by regulation.
LoRa is the modulation. LoRaWAN is a network protocol built on LoRa for low-rate sensor nodes and gateways. LoRaWAN is generally not the right protocol for a low-latency primary flight-control link.
Common frequency regions
  • United States/Canada: the unlicensed 902–928 MHz region, commonly called 915 MHz.
  • Europe: primarily 863–870 MHz, commonly called 868 MHz, with duty-cycle and sub-band constraints.
  • Other countries use different regional plans; the radio and firmware must match the country of operation.
  • LoRa modulation is also used at 2.4 GHz by some systems, but sub-GHz normally has better propagation and lower free-space path loss for the same conditions, while requiring a larger antenna.
ExpressLRS
ExpressLRS is an open-source, low-latency RC link that uses LoRa-family modulation and supports both roughly 900 MHz and 2.4 GHz hardware. It is much more appropriate for command/control than generic LoRaWAN. It can also carry bidirectional telemetry and MAVLink in supported configurations.
  • 900 MHz: generally stronger diffraction/foliage performance and a favorable link budget, but a physically larger quarter-wave antenna—about 82 mm at 915 MHz before practical tuning—and less spectrum for very high packet rates.
  • 2.4 GHz: smaller antenna—about 31 mm quarter-wave—higher update-rate options and widespread hardware, but usually more congestion and somewhat worse obstruction/foliage behavior.
Longest range is not determined by modulation alone. It depends on transmit power, receiver sensitivity, bandwidth/spreading factor, antenna efficiency and orientation, polarization, interference, Fresnel-zone clearance, and local legal limits. A badly placed 900 MHz antenna can perform worse than a well-installed 2.4 GHz link.
Recommendation for revision A
Use a connectorized or pre-certified ExpressLRS receiver/module, rather than placing a discrete LoRa transceiver and RF power amplifier on the first integrated board. Connect it to the H753 using a UART-compatible RC protocol such as CRSF and provide clean regulated power. This gives practical long-range capability while still allowing RF layout practice through:
  • A controlled 50-ohm feed or external antenna connector if required by the chosen module.
  • Antenna keepout and placement study.
  • Power filtering and EMI isolation from ESC switching.
  • RF coexistence testing.
Do not bury the antenna between battery wiring, motors, carbon fiber, ground planes, or high-current copper. Carbon-fiber frames are conductive and can detune or shield antennas.
Sources consulted:
4. Making the PCB Look Like a Four-Motor Drone Frame
There are two interpretations:
A. A central electronics PCB with four short lobes
This is feasible. A central X-shaped or plus-shaped board can place one ESC stage near each edge, shorten motor phase connections, and create clear functional quadrants. Separate structural arms still carry the motors.
Benefits
  • Intuitive physical partitioning: one ESC per quadrant, flight control at the center, RF/GNSS toward clean outer edges.
  • Short battery distribution paths if the battery connector and bulk capacitance are central.
  • Short motor-phase connections to edge connectors.
  • Strong visual and educational relationship between architecture and layout.
Challenges
  • Narrow necks between the central body and lobes can restrict current, return paths, and routing channels.
  • Internal corners concentrate mechanical stress; use generous radii rather than sharp re-entrant corners.
  • Plane continuity must be maintained beneath high-speed signals crossing any neck.
  • Board fabrication cost rises because the bounding rectangle contains unused area.
B. A full-size PCB whose long arms directly support the motors
This is not recommended for the first revision. Ordinary FR-4 is not an ideal primary drone structure: motor vibration, landing impact, arm bending, copper work-hardening, connector strain, and delamination risk threaten both mechanics and electronics. Eight-layer heavy-copper construction is also expensive and heavy. Long PCB arms become antennas and distribute ESC noise close to the IMU.
Use a central X-shaped 8-layer electronics board mounted to a conventional carbon-fiber or composite frame, not an FR-4 airframe. Keep the arms as short lobes that end at motor-phase pads/connectors or local ESC stages. Mechanically isolate the IMU near the board's center. Keep antennas outside the carbon-fiber shadow and away from ESC lobes.
Preliminary floorplan
  • Center: BMI088 and other inertial sensors, with symmetrical mounting and local low-noise power.
  • Inner center: STM32H753 and STM32F405, storage, clocks, and debug.
  • Four quadrants/lobes: one ESC channel each.
  • Central propulsion zone: battery input, TVS, bulk capacitors, current sensing, and star-like battery distribution.
  • One quiet edge: GNSS and magnetometer connector/module, away from phase copper and inductors.
  • Opposite clean edge: RF receiver/module and antenna launch/keepout.
  • USB and SWD: accessible edges, protected from propeller and battery lead interference.
5. Current Recommendation
  • Start at 4S for revision A unless the selected motor/propeller combination specifically requires 6S.
  • Target a moderate current class, then size from measured or manufacturer motor data rather than guessing.
  • Use a dedicated controller per ESC channel, with the H753 issuing DShot commands.
  • Use a connectorized ExpressLRS receiver, likely 900 MHz when legal and when antenna size is acceptable; keep the hardware region-specific.
  • Use a central X-shaped board with short ESC lobes mounted on a separate structural frame, not full-length FR-4 motor arms.
6. Decisions Needed Before Schematic and Outline Work
  1. Country/region of operation, which determines legal RF bands.
  2. Approximate drone class: propeller diameter, motor size/KV, and takeoff mass.
  3. Whether revision A should prioritize easier bring-up (4S) or maximum power-density learning (6S).
  4. Maximum board center size and mounting-hole pattern.
  5. Whether the receiver is removable/replaceable or permanently integrated.

  • 1. Choosing Between 4S and 6S

  • 4S advantages

  • 4S disadvantages

  • 6S advantages

  • 6S disadvantages

  • Recommendation

  • 2. ESC Architecture Options

  • Option A: Dedicated MCU per ESC channel

  • Option B: One shared motor-control MCU for all four motors

  • Option C: Direct ESC control from the STM32H753 flight MCU

  • Option D: Smart three-phase drivers or integrated power modules

  • Recommended development path

  • 3. LoRa, LoRaWAN, and Drone Control Links

  • What LoRa is

  • Common frequency regions

  • ExpressLRS

  • Recommendation for revision A

  • 4. Making the PCB Look Like a Four-Motor Drone Frame

  • A. A central electronics PCB with four short lobes

  • B. A full-size PCB whose long arms directly support the motors

  • Recommended mechanical concept

  • Preliminary floorplan

  • 5. Current Recommendation

  • 6. Decisions Needed Before Schematic and Outline Work

Hades Integrated Drone Controller

Hades Integrated Drone Controller thumbnail
An 8-layer HadesFCS-derived drone electronics platform combining an STM32H753 flight controller, STM32F405 navigation computer, RF/telemetry, four ESC power stages, and high-current power distribution, with emphasis on signal integrity and power integrity practice.

Properties

USB 2.0

BSD-3-Clause derivative

STM32 FreeRTOS

Drone Flight Control

8-layer integrated FC ESC PDB

Pricing & Availability

See prices from popular manufacturers for your project.

Controls