To control torque of DC motor systems precisely, you must regulate the current flowing through the windings, because motor torque is directly proportional to current ($T = k_t \times I$). For brushed DC (BDC) motors, this requires an H-bridge driving PWM with current-sense feedback. For brushless DC (BLDC) motors, the gold standard is Field Oriented Control (FOC), which uses a 3-phase inverter to map current vectors to the rotor's exact magnetic angle in real-time. If you just apply voltage and hope for the best, you are controlling speed, not torque.
Motor Type Comparison: Torque Curves and Control Needs
Not all DC motors handle torque commands the same way. The physical construction dictates the torque curve, which in turn dictates the driver complexity and cost. Here is how the three main DC motor archetypes compare for torque-critical applications in 2026.
| Motor Type | Torque Curve Profile | Control Complexity & Driver | Typical Cost (Module + Driver) | Best Load Profile |
|---|---|---|---|---|
| Brushed DC (BDC) | Linear drop-off from stall torque to zero torque at no-load speed. | Low. Requires H-bridge, PWM, and inline shunt resistor for current limiting. | $25 - $80 | Simple winches, basic conveyors, low-cost traction. |
| Brushless DC (BLDC) | Flat continuous torque up to base speed, then hyperbolic drop-off (field weakening region). | High. Requires 3-phase FOC inverter, rotor position feedback (Halls/Encoders), and Clarke/Park math. | $150 - $400 | Robotics joints, CNC axes, high-precision tensioning. |
| Coreless DC | Extremely linear, zero cogging, rapid thermal saturation if stalled. | Medium. H-bridge with high-resolution current sensing; highly sensitive to PWM frequency. | $80 - $250 | Surgical robotics, camera gimbals, aerospace actuators. |
The Verdict: If your load requires holding a specific tension or force without moving (like a web tensioner), choose a BLDC with FOC. BDC motors will overheat rapidly at stall because all the electrical energy converts to heat in the armature with zero back-EMF to mitigate it.
Sizing Rule of Thumb and Worked Load Example
The most common mistake on the bench is sizing a motor based on peak stall torque rather than continuous RMS torque. Rule of thumb: Size the motor's continuous rated torque to be at least 1.5x to 2.0x your calculated continuous load torque. This prevents thermal saturation and leaves headroom for acceleration spikes.
Load: Lifting a 20 kg (44 lb) mass at a constant velocity.
Drum Radius: 50 mm (0.05 m).
Step 1: Calculate Load Force. $F = m \times g = 20 \text{ kg} \times 9.81 \text{ m/s}^2 = 196.2 \text{ N}$
Step 2: Calculate Drum Torque. $T = F \times r = 196.2 \text{ N} \times 0.05 \text{ m} = 9.81 \text{ Nm}$
Step 3: Apply Safety Factor (2x). $9.81 \text{ Nm} \times 2 = 19.62 \text{ Nm}$ required at the drum.
Step 4: Factor in Gearbox. Using a 10:1 planetary gearbox with 90% efficiency ($\eta = 0.9$).
Motor Torque Required = $19.62 / (10 \times 0.9) = 2.18 \text{ Nm}$.
Selection: A NEMA 34 BLDC like the Moons' LE86S (rated ~3.0 Nm continuous) paired with a 10:1 planetary reducer perfectly fits this profile, providing a safe thermal margin.
Wiring and Terminal Identification for BLDC Torque Control
To execute FOC and accurately control torque of DC motor systems, the driver must know the rotor's position. While sensorless FOC exists, it fails at zero-speed high-torque holding. For full torque control from 0 RPM, you need Hall sensors or an encoder. Here is the standard terminal mapping for a sensored BLDC connected to an ODrive Pro or similar FOC controller.
- Phase Terminals (U, V, W): These carry the high-current 3-phase AC waveforms. Use 12 AWG or 10 AWG silicone wire depending on the RMS current. Never swap these without updating the motor calibration in software, or the motor will violently oscillate.
- Hall Sensor Signals (A, B, C or Hu, Hv, Hw): Low-voltage digital outputs (usually open-drain or push-pull). Use 22 AWG to 24 AWG shielded twisted pair. Keep these physically separated from the phase wires to avoid inductive noise coupling.
- Hall VCC and GND: CRITICAL WARNING: Most industrial BLDC halls require 5V DC. I have seen countless builders fry the internal hall ICs by accidentally wiring this to a 12V or 24V logic rail. Verify the datasheet before applying power.
Failure Signatures: Diagnosing Hum, Overheat, and Stall
When your torque loop misbehaves, the motor will physically tell you what is wrong before the driver throws a software fault code. Here is how to read the physical failure signatures.
1. The Audible Hum or High-Frequency Whine
Cause: Cogging torque or FOC angle offset. If the Park transform is calculating the $I_q$ (torque-producing) and $I_d$ (flux-producing) currents based on a slightly wrong rotor angle, it injects current into the wrong magnetic axis. This creates a vibrating torque ripple rather than smooth rotation.
Fix: Run the driver's automatic calibration routine (e.g., odrv0.axis0.requested_state = AXIS_STATE_FULL_CALIBRATION_SEQUENCE) to measure the phase inductance and hall sensor offsets. Ensure your hall sensor wires aren't picking up EMI from the phase cables.
2. Rapid Casing Overheat (Without Tripping the Driver)
Cause: Confusing peak torque with continuous RMS torque. A motor might be rated for 10 Nm peak (for 3 seconds) but only 3 Nm continuous. If your application demands 6 Nm to hold a static load, the motor will act as a space heater. The driver won't trip its overcurrent limit because 6 Nm is below the peak threshold, but the winding insulation will eventually melt.
Fix: Check the motor's thermal time constant and continuous current rating ($I_{rms}$). If you must hold high static torque, add forced air cooling or switch to a larger frame size with a higher thermal mass.
3. Hard Stall and Driver Fault
Cause: The load exceeded the motor's breakdown torque, or the mechanical linkage bound up. The FOC controller desperately ramps up current to meet your torque command, hits the hardware current limit ($I_{max}$), and faults out to protect the MOSFETs.
Fix: Measure the actual load with a torque wrench or load cell. If the load is correct, your mechanical advantage (gear ratio) is too low. Increase the gear reduction to multiply the motor's output torque.
FAQ: Advanced DC Motor Torque Control
How do you control the torque of a DC motor without an encoder?
You can use Sensorless FOC or simple open-loop current limiting. Sensorless FOC estimates the rotor position by measuring the Back-EMF (voltage induced in the un-driven phase) or by injecting high-frequency voltage pulses to measure magnetic saliency. However, sensorless methods fail at zero RPM. If you need holding torque at a dead stop without an encoder, you must use a brushed DC motor with a high-wattage shunt resistor for current feedback, or accept that a BLDC will lose position and torque authority at very low speeds.
Why does my DC motor lose torque at high RPM?
This is dictated by physics, specifically the Back-EMF constant ($k_e$). As the motor spins faster, it generates a reverse voltage that opposes your driver's supply voltage. The net voltage available to push current through the winding resistance drops. Since torque equals current, and current drops as speed increases, your available torque falls off hyperbolically. To maintain torque at high RPM, you need a higher bus voltage (e.g., upgrading from a 24V to a 48V power supply) or you must employ field weakening, which intentionally injects negative $I_d$ current to sacrifice efficiency for speed.
Can I use a standard RC ESC to control BLDC motor torque precisely?
No. Standard hobby ESCs (Electronic Speed Controllers) use 6-step trapezoidal commutation and are designed purely for velocity/throttle control. They lack the inline current shunts and the processing power required to run the Clarke and Park transforms necessary for FOC. If you send a 50% PWM throttle signal to an RC ESC, it will try to maintain a specific speed, not a specific torque. For true torque control, you must use a dedicated FOC driver like the ODrive, SimpleFOC shield, or a Texas Instruments InstaSPIN industrial module.
