Actuators: hobby servos vs BLDC vs stepper

Pick the wrong motor and your robot dies. Pick the right one and you can use a $5 part where someone else is paying $50. The decision tree is honestly small — three families cover 90% of hobby robotics, and they don't overlap as much as the marketing suggests.

The three families

Family	What it is	Best for	Cost
Hobby servo	DC + gearbox + position controller in one $5 package	Arms, grippers, low-precision joints	$3–25
Brushed DC	Analog motor, needs separate H-bridge + encoder	Wheeled robots, simple actuators	$8–50
BLDC (brushless)	Three-phase motor + ESC; FOC for torque control	Drones, quadrupeds, dynamic motion	$25–300+
Stepper	Open-loop position via discrete steps	3D printers, low-speed precise motion	$10–80

Hobby servo: $5 of integrated convenience

A hobby servo (SG90, MG90S, MG996R, DS3225, and friends) is a small DC motor + gearbox + position-feedback potentiometer + microcontroller, all in one package. You send it a PWM signal where the pulse width encodes the desired angle, and the internal controller drives there.

Strengths:

• Position control out of the box. No PID tuning, no encoder wiring.
• Cheap. SG90 is $1.50 in bulk; MG996R (metal gear) is $4–6.
• Easy to interface — every microcontroller can output PWM.

Weaknesses:

• No torque feedback. You command position; you don't know what force the servo applied.
• Stall current is brutal. MG996R can pull 2–3 A under load. Don't power six off your MCU's 5V pin.
• Limited speed. Most hobby servos move ~0.2 sec / 60° — fine for arms, slow for fast gestures.
• Limited durability. Plastic gears strip; cheap servos die after a few hundred load cycles.

Use them for: Robot arms (low-cost), grippers, animatronic mouths, antenna pointers. Quadruped builds with hobby servos work but are intrinsically limited — see Petoi Bittle for the upper bound of what plastic-gear hobby servos can do.

Brushed DC: the workhorse for wheels

A standard DC motor: two terminals, voltage in → torque out. To drive bidirectionally you need an H-bridge (TB6612FNG, DRV8833, L298N at the cheap end). For position or velocity feedback you bolt on a quadrature encoder.

Strengths:

• Cheap and simple electrically. Linear torque-vs-current.
• Well-understood; PID tuning is straightforward.
• Easily geared — Pololu sells dozens of metal-gearmotor SKUs from 5:1 to 1000:1.

Weaknesses:

• Brushes wear; lifetime is hundreds to thousands of hours, not unlimited.
• Brush sparking emits EMI that can confuse nearby electronics.
• Less efficient than BLDC at the same size.

Use them for: Wheeled robots, conveyor systems, anything that needs continuous rotation with simple speed control. Most commercial educational robots (TurtleBot, mini Pupper, etc.) use brushed DC for the wheels.

BLDC: the modern dynamic-robotics motor

Brushless DC motors are three-phase synchronous machines driven by an electronic speed controller (ESC) that switches the windings in sequence. With field-oriented control (FOC), you get smooth, torque-accurate, encoder-controlled motion at high efficiency.

Strengths:

• Higher torque-density than brushed DC.
• FOC gives direct torque control — you command N·m, get N·m. Critical for legged robots and drones.
• Long life (no brushes).
• Massive hobbyist ecosystem now: ODrive, SimpleFOC, MJBots Moteus.

Weaknesses:

• Expensive — even hobby BLDCs start at $25 for the motor + $50 for the controller.
• Setup is more involved. Encoder calibration, current limits, PID gains, anti-cogging.
• Most BLDCs come without a gearbox; planetary gearboxes for them are a separate purchase.

Use them for: Drones (FOC is overkill — most drone ESCs use simpler trapezoidal commutation), quadrupeds (where torque control matters), serious arms, anywhere you want backdrivability + dynamic behavior. The 2020s revolution in legged robotics is built on cheap-enough BLDCs + FOC controllers.

Stepper: the open-loop precision option

A stepper motor moves in discrete angular steps (typically 200 steps/revolution = 1.8°/step). With microstepping, ~3200 steps/rev. No encoder needed — you tell it "move 100 steps" and it does, as long as you don't exceed its torque budget.

Strengths:

• Precise positioning without feedback.
• High holding torque — keeps position with no sliding.
• Cheap drivers (A4988, TMC2209) are widely available.

Weaknesses:

• "Skipped steps" — if you exceed the torque, the motor falls behind silently and your position is wrong forever.
• Slow for their size. Generally limited to a few hundred RPM.
• Always drawing current to hold position; not very efficient.

Use them for: 3D printer axes, CNC, scanning sensor mounts, any "move there, hold" application that doesn't need fast or backdrivable motion.

What about specialty types?

• Coreless DC — same as brushed DC but with a hollow rotor for very fast acceleration. Used in micro-quads, finger actuators.
• Voice coil actuators — linear, very precise, very limited stroke. Camera autofocus, hard-drive head positioners.
• Servomotors with torque sensing (Dynamixel XM-series, ROBOTIS) — combine the convenience of a servo with the torque-control of BLDC. Expensive but excellent for arms.
• Hydraulic / pneumatic — high force-to-weight, mostly outside hobby scope.
• Series-elastic actuators — a spring in series with a stiff motor. Used in walking robots for shock absorption.

Decision framework

Ask three questions:

1. Position, velocity, or torque? Position → servo or stepper. Velocity → brushed DC or BLDC. Torque → BLDC with FOC.
2. How much force? Sub-newton hand things → micro servos. Few-newton arms → MG996R. Tens of newtons (legged robot) → BLDC. Hundreds (industrial arm) → BLDC + harmonic drive, or hydraulic.
3. How fast? Slow → anything works. Fast and dynamic → BLDC, no other choice.

The hidden cost: power

Whatever motor you pick, budget for power separately. A six-servo arm draws 5–15 A at peak. A mini-quadruped with twelve BLDCs needs 200 W of regulated supply. Almost every "my robot won't run reliably" thread on the maker forums turns out to be a power-supply budget too tight by 30%. Pick supplies sized for stall current, not nominal current.

Exercise

Pick a robot project you're considering. For each joint, write down: required force, required speed, position vs velocity vs torque control, and the resulting motor pick. Then add up the stall currents and pick a power supply. This 10-minute spreadsheet beats two hours of debugging brownouts.

Next

Encoders — the eyes that tell your motor where it actually is. Quadrature, magnetic, absolute, and the gotchas that make wheels report 17.3% off the ground truth.