What Is Cogging Torque and Why Does It Matter?
Cogging torque is the periodic torque variation that occurs in permanent magnet motors due to the magnetic attraction between the rotor magnets and the stator teeth. As the rotor turns, each magnet passes over each stator slot, creating a repeating “snap” — a preferred resting position where the reluctance of the magnetic path is minimized.
In a motor running at 3,000 rpm with 12 poles and 18 slots, the cogging frequency is 3,000 × (12/2) × 18 / 60 = 2,700 cycles per second — far above audible range. But at low speeds (below 100 giri/min), cogging causes visible jerking motion that degrades positioning accuracy in:
- Robotic joint actuators requiring sub-arcminute positioning repeatability
- Machine tool servo axes where surface finish depends on smooth low-speed movement
- Medical device drives (surgical robots, infusion pumps) requiring constant low-speed torque
- Direct-drive rotary tables where gearboxes cannot smooth out torque ripple
- EV steering motors (EPS) where cogging is felt as steering “notchiness”
For these applications, cogging torque testing is not optional — it is a mandatory acceptance criterion that distinguishes production-grade motors from low-cost alternatives.
How Cogging Torque Is Generated: The Physics
A permanent magnet creates a magnetic field that seeks the path of minimum reluctance through the stator iron. Reluctance varies with rotor angle because stator slots (air gaps) have higher reluctance than stator teeth (iron). As the rotor rotates, the energy stored in the magnetic field changes cyclically — and the derivative of magnetic energy with respect to angle is torque.
Key parameters that determine cogging torque magnitude:
- Slot-pole combination: Motors with a high least common multiple (LCM) of pole count and slot count have lower cogging — more, smaller cogging cycles average out
- Magnet skewing: Rotor magnets skewed by one slot pitch nearly eliminate fundamental cogging but add manufacturing cost
- Slot opening width: Narrower slot openings reduce cogging at the cost of winding difficulty
- Air gap uniformity: Eccentric rotors amplify cogging; tight bearing tolerances are essential
- Magnet arc span: Optimized magnet arc (typically 0.8–0.9 times pole pitch) minimizes cogging
Test della coppia di cogging: Equipment Requirements
1. Dynamic Torque Sensor — Resolution Is Everything
Standard shaft-mounted torque sensors used for efficiency testing (precisione ±0.1% Fs) are inadequate for cogging torque measurement. A servo motor rated at 10 N·m with 0.5% cogging torque produces only 50 mN·m of cogging — below the noise floor of a standard sensor.
Required torque sensor specifications:
- Rated range: 0–0.5 N·m to 0–5 N·m (sized to cogging amplitude, NOT rated motor torque)
- Resolution: 0.1 mN·m or better
- Larghezza di banda: ≥1 KHz (to capture high-frequency cogging harmonics)
- Reaction type: Flange-mounted reaction torque sensor preferred — no slip ring noise
2. High-Resolution Angle Encoder
Cogging torque must be plotted as a function of rotor angle with sufficient resolution to resolve individual cogging cycles. For a 6-pole-pair motor, there are 6 cogging cycles per mechanical revolution — each cycle is 60 mechanical degrees wide. To capture the waveform shape accurately, at least 100 measurement points per cycle are needed: 60° / 100 points = 0.6° per point.
Required encoder specifications:
- Resolution: ≥18-bit (262,144 counts/revolution), equivalent to 0.0014°
- Precisione: ±10 arc-seconds or better
- Tipo: Absolute encoder preferred — eliminates index pulse uncertainty
3. Rotation Drive
The motor under test must be rotated at a controlled, very low speed — typically 1–10 giri/min — while the test system records torque versus angle. Two approaches:
- External drive motor: A separate servo motor drives the MUT shaft at constant speed. The load dynamometer is disconnected or operates at near-zero torque. This is the most accurate method.
- Self-rotation at minimum current: The MUT is commutated at minimum current below rated speed. Less accurate due to electromagnetic torque interference, but simpler to implement.
Cogging Torque Test Procedure
Fare un passo 1: Mount the Motor Vertically
Cogging torque measurements at mN·m level are affected by rotor weight. A 500g rotor with 50mm eccentricity produces 0.245 N·m of gravitational torque — five times larger than the cogging torque being measured. Montaggio verticale (output shaft pointing up or down) eliminates this effect completely.
Fare un passo 2: Demagnetization Check
Before the cogging test, verify magnet health by measuring the no-load back-EMF at 1,000 rpm and comparing to specification. Partially demagnetized magnets reduce cogging amplitude AND rated torque — both failures, but cogging reduction might mask the demagnetization during the cogging test alone.
Fare un passo 3: Temperature Stabilization
Cogging torque has a slight temperature dependence (magnet remanence decreases with temperature). Conduct the cogging test with the motor at a stable ambient temperature (20–25°C), before any load testing that would heat the magnets.
Fare un passo 4: Multi-Revolution Averaging
Rotate the motor through at least 3 complete revolutions and average the torque-angle data. This averages out measurement noise and reveals the true periodic cogging waveform. The standard deviation of averaged points should be less than 10% of the peak cogging amplitude.
Fare un passo 5: Analisi armonica
After acquiring the torque vs. angle waveform, apply FFT analysis to identify cogging harmonics:
- Fundamental cogging frequency: Number of slot-pole interactions per revolution
- Higher harmonics: 2nd, 3rd, 5th harmonics of the fundamental
- Asymmetry: Non-symmetric waveform indicates manufacturing defects (unequal magnet strength, eccentricity)
Cogging Torque Acceptance Criteria
| Applicazione | Coppia di cogging (% of Rated) | Typical Rated Torque | Absolute Limit |
|---|---|---|---|
| Standard industrial servo | <2–5% | 5–50 N·m | <1.0 N·m |
| Precision servo (machine tools) | <0.5–1% | 5–50 N·m | <0.25 N·m |
| Collaborative robot joint | <0.5–2% | 50–300 N·m | <1.5 N·m |
| Direct-drive rotary table | <0.2–0.5% | 100–1000 N·m | <0.5 N·m |
| EPS (electric power steering) | <1% | 5–15 N·m | <0.1 N·m |
| Medical device drive | <0.5% | 0.5–5 N·m | <10 mN·m |
Cogging Torque Reduction Techniques
When a motor fails the cogging torque acceptance test, the following techniques can reduce cogging — either in new design or through motor modification:
Design-Level Reduction
- Magnet skewing: Most effective. Continuous skew by one slot pitch reduces fundamental cogging by 90%+. Step skew (3–5 magnet segments with angular offset) achieves similar results at lower manufacturing cost.
- Slot opening optimization: Reducing slot opening width from 3mm to 1.5mm can halve cogging torque with no change to magnets or windings.
- Asymmetric magnet placement: Slight angular offset of every other magnet pair disrupts the periodicity of cogging without skewing.
Control-Level Compensation
- Feedforward cogging compensation: A lookup table of cogging torque vs. angle is stored in the servo drive. The drive adds a compensating current injection at the measured frequency. Can reduce position error from cogging by 80–90%.
- Notch filter: A notch filter in the velocity loop at the cogging frequency attenuates velocity ripple without affecting the control bandwidth significantly.
Domande frequenti
Is cogging torque the same as torque ripple?
NO. Cogging torque is the position-dependent torque variation that exists even with zero current (the motor unpowered). Torque ripple is the variation in output torque under powered conditions, caused by a combination of cogging torque, current harmonics, commutation switching, and back-EMF harmonics. Cogging torque is one component of torque ripple, but not the only one. A motor can have low cogging torque but high torque ripple due to poorly tuned current control.
Can cogging torque be measured with the motor powered?
It is possible but more complex. With the motor powered, electromagnetic torque is superimposed on cogging torque. To separate them, the test must be conducted at very low current (vicino allo zero) and the data must be carefully processed. The preferred method is to rotate the unpowered motor externally — this directly measures cogging torque with no electromagnetic interference.
What is the cogging torque frequency for a 12-pole, 18-slot motor at 100 giri/min?
Cogging frequency = (pole pairs × slots) / 60 × speed = (6 × 18) / 60 × 100 = 18 Hz. The motor produces 18 cogging cycles per second at 100 giri/min — well within the measurement bandwidth of a standard data acquisition system.
How does temperature affect cogging torque?
Neodymium magnets (NdFeB) lose approximately 0.1–0.12% remanence per °C above 20°C. At 80°C operating temperature, magnet remanence drops by approximately 6–7%, reducing cogging torque by a similar percentage. For acceptance testing, always conduct cogging measurements at a controlled temperature and specify the test temperature in the report.
Conclusione
Cogging torque testing requires specialized measurement equipment — high-resolution torque sensors and angle encoders sized for the cogging amplitude, not the motor’s rated torque. The measurement procedure — vertical mounting, thermal stabilization, multi-revolution averaging, and FFT analysis — is as important as the hardware. Getting it right separates motors that perform on a data sheet from those that deliver smooth, precise motion in demanding applications.
EconoTest cogging torque test systems cover servo motors from 0.5 N·m to 500 N·m rated torque, with dynamic torque sensors to 1 mN·m resolution and 23-bit absolute encoders.
→ Contact our team to configure a cogging torque test system for your motor range.
