Back to projects

Embedded actuator design project

Ultra-Flat Micro Servo Actuator

Designed and modelled an ultra-flat servo actuator combining compact mechanical packaging, spur/worm gearing, closed-loop motor control, PCB design, and embedded electronics for space-constrained robotics applications.

Motor control PCB design MATLAB/Simulink CAD packaging Geartrain modelling
Final CAD package of the ultra-flat micro servo actuator
Final CAD package showing the motor, compact spur/worm drivetrain, output stage, PCB, and ultra-flat casing concept.

Project snapshot

Problem
Most high-performance servos are too tall or bulky for compact mechanisms with strict vertical constraints. The project investigated whether high torque and closed-loop position control could be achieved in an ultra-flat servo form factor.
My role
Mechanical, electrical, and control-system co-design, including requirements definition, component selection, gearbox trade study, control modelling, PCB design, CAD packaging, and verification planning.
Tools
MATLAB, Simulink, EasyEDA, CAD, STM32, DRV8213 motor driver, AS5600L magnetic encoder, coreless DC motor, PCB design, and geartrain modelling.
Design targets
Approximately 10-11 mm height envelope, 120 deg travel, 4.8-5 V power stage, 3.3 V logic, RC PWM interface, high stall torque, low backlash, and closed-loop position control.
Outcome
Completed concept design, CAD packaging, gearbox selection, control simulation, compact PCB design, sensor PCB design, and verification plan. Simulation met the torque target but not the original speed benchmark.
Relevance
Embedded systems, motor control, compact PCB design, mechatronic packaging, requirements traceability, simulation, control systems, and design trade-off analysis.

Visual proof

My contribution

  • Defined the servo requirements and verification plan based on size, torque, speed, accuracy, backlash, power, and safety constraints.
  • Selected and evaluated the motor, geartrain, encoder, motor driver, MCU, and power architecture.
  • Designed a spur-then-worm drivetrain concept to provide high gear reduction, 90 deg output orientation, and low back-drivability.
  • Built MATLAB/Simulink models to evaluate gear ratios, step response, current limits, speed limits, overshoot, and torque output.
  • Designed a compact main PCB and secondary sensor PCB for the motor-control electronics and position feedback system.
  • Developed a verification plan covering PWM input, travel, step response, torque, backlash, thermal behaviour, current draw, and fault handling.
  • Documented the trade-offs between height, torque, speed, efficiency, current draw, thermal limits, backlash, and manufacturability.

Results and current status

This project reached the design, modelling, and simulation-validation stage. The final concept used a compact brushed coreless motor, spur/worm geartrain, magnetic position sensing, a custom PCB architecture, and a triple cascaded position-velocity-current control model. The simulation achieved a 60 deg step response in approximately 0.263 s with about 0.4% overshoot and 0.156 deg steady-state error, while meeting the stall torque target at approximately 1.63 N*m.

The design did not fully meet the original 0.14 s per 60 deg speed benchmark, which made the main trade-off clear: the actuator could meet the torque target in simulation, but response speed would need improvement through gear ratio selection, drivetrain efficiency, voltage/current headroom, and thermal performance. Physical prototype testing and tuning were identified as the next stage.

Technical details

Requirements and constraints

The project was constrained by an unusually flat mechanical envelope while still aiming for meaningful servo performance. The requirements connected physical package size to electrical limits, mechanical output behaviour, controller response, and planned validation.

Envelope
Targeted an ultra-flat package around a 10-11 mm height constraint, with the motor, drivetrain, output stage, PCB, and casing packaged together.
Motion
At least 120 deg output travel with closed-loop position control and controlled response to RC PWM commands.
Power
4.8-5 V power stage with 3.3 V logic, explicit current limiting, and planned brownout and fault handling.
Torque
Stall torque target based on high-performance servo requirements, with simulation estimating approximately 1.63 N*m.
Speed
Original benchmark of 0.14 s per 60 deg. Final simulation achieved approximately 0.263 s, identifying a clear design trade-off.
Validation
Verification plan covering timing, travel, step response, torque, backlash, current draw, thermal behaviour, endurance, and fault handling.
Mechanical architecture and drivetrain selection

The mechanical concept used a compact motor and high-reduction geartrain to fit torque multiplication, right-angle output motion, output support, and PCB packaging into a thin actuator body.

Motor choice
A compact brushed coreless DC motor was selected around speed, current, torque constant, diameter, mass, and packaging constraints.
Drivetrain
The selected concept used a spur-then-worm drivetrain to combine high reduction with a 90 deg output orientation.
Gear ratio trade study
Multiple spur and worm combinations were evaluated against transit time and stall torque, not only against static torque.
Back-drivability
The worm final drive reduced back-drivability, supporting holding behaviour without relying purely on continuous motor current.
Packaging
CAD packaging balanced the motor, output gear, worm, casing, bearing support, fasteners, PCB space, and connector access.
Control model and simulation

The MATLAB/Simulink model was used to evaluate whether the chosen drivetrain and electrical limits could deliver controlled position motion before physical validation. It was deliberately constrained by current, voltage, speed, and anti-windup behaviour.

Architecture
The simulation used a triple cascaded position-velocity-current control structure.
Limits
Current limiting, output-speed capping, voltage limits, and anti-windup behaviour were included so the model did not rely on unrealistic drive behaviour.
Simulation result
The simulated 60 deg step reached approximately 0.263 s with about 0.4% overshoot and 0.156 deg steady-state error.
Trade-off
The model met torque and accuracy targets in simulation, but the speed benchmark remained the main improvement path.
Electronics and PCB design

The electronics were split between a main motor-control PCB and a secondary sensor PCB. This separated power-drive packaging from output-position feedback while keeping the boards small enough for the actuator body.

Main PCB
Compact motor-control PCB designed around the MCU, DRV8213 H-bridge motor driver, power path, connectors, decoupling, and programming access.
Sensor PCB
Secondary PCB designed around the AS5600L magnetic encoder for compact output-position feedback.
Power architecture
5 V power stage and 3.3 V logic considered alongside current sensing, filtering, fault detection, and staged bring-up.
Bring-up plan
The electronics were designed with planned tests for power rails, programming, PWM input, encoder feedback, motor output, current limiting, and thermal behaviour.
Verification plan

The verification plan was used to connect design goals to specific future tests. This kept the project grounded in how the actuator would eventually be proven, not only how it was simulated.

  • Check RC PWM timing, command range, jitter, and direction with test signals and oscilloscope measurements.
  • Measure output travel, backlash, no-load 60 deg step response, overshoot, steady-state error, and repeatability.
  • Use a lever-arm torque rig to validate stall and continuous torque against the simulation estimate.
  • Measure current draw, voltage droop, temperature rise, and fault behaviour under load and stalled conditions.
  • Compare measured plant behaviour against the MATLAB/Simulink model, then retune the cascaded controller using real data.
Results, limitations, and future work
  • Physically assemble the actuator and complete staged PCB bring-up before claiming hardware validation.
  • Validate step response, torque, current, voltage droop, backlash, thermal behaviour, and repeatability on the bench.
  • Improve the speed result through gear ratio selection, drivetrain efficiency, voltage/current headroom, and thermal management.
  • Refine mechanical tolerances, lubrication, output support, casing fit, and manufacturability after prototype testing.
  • Tune firmware against measured actuator dynamics rather than only simulated plant parameters.