What happens during a locked-rotor event
A mechanism jams and the motor stops turning with power still applied. What fails, in what order, and how long is there before something is damaged?
A locked rotor is not an abnormal electrical condition. It is the normal condition of a DC motor with no back-EMF: current equal to supply voltage over terminal resistance, torque at its maximum, output power at zero. Everything the supply delivers becomes heat, and it becomes heat in a winding that is no longer moving through the airflow it usually stirs.
The question a designer needs answered is not whether that is bad. It is how long there is, and what gives way first.
The thermal calculation
C · dΔT/dt = V² / R(ΔT) − ΔT / Rth
- ΔT
- winding temperature rise above ambient, K
- C
- thermal capacity of the winding, J/K
- Rth
- thermal resistance from winding to ambient, K/W
- R(ΔT)
- winding resistance, which rises with temperature
Two results matter. The first is that the resistance rise is self-limiting: as copper heats, the winding draws less current, so the steady-state rise settles well below what the cold-resistance calculation predicts. That is a real effect and it is why some small motors survive an indefinite stall rather than burning out.
The second is that self-limiting does not help in the first fifteen seconds. Both curves cross the insulation headroom at nearly the same time, because early in the transient the winding has not warmed enough for the resistance rise to matter. A stall protection scheme that has to act, has to act in seconds.
What fails, in order
-
Seconds: the commutator contact
Full current stands in one contact area with no relative motion, so there is no film renewal and no spreading of the heat. Precious metal contacts erode or transfer material. Carbon contacts overheat locally and can bond to the segment.
-
Seconds to tens of seconds: the winding insulation
The enamel on the magnet wire reaches its temperature index and begins to degrade. Degradation is cumulative. A winding that survived three stalls is not the winding it was before them.
-
Tens of seconds: the magnets
High armature current opposes the magnet's field while the magnet is hot, which is the condition for irreversible demagnetization. A partly demagnetized motor comes back with a lower torque constant, so it runs faster with less torque and draws more current for the same load. The change is permanent and it looks like a different motor.
-
Any time, mechanically: the drivetrain
Stall torque times ratio is applied to the gearhead for as long as the jam lasts. A gearhead sized for the application torque may be far below that figure, so a tooth, a press fit or an output bearing can fail before anything thermal happens.
-
Hours to weeks later: the consequences
A stall that does not fail anything immediately still leaves degraded insulation, a damaged commutator surface and cooked grease. The failure that gets reported is the one that happens weeks later, and it is usually diagnosed as a random failure.
Designing for it
| Approach | How it works | Cost or limitation |
|---|---|---|
| Current limit in the drive | Caps the current, so torque, heat and mechanism load are all bounded | Reduces available starting torque unless the limit is time-based |
| Time-limited stall detection | Detects no motion with current flowing and removes power within a few seconds | Requires feedback or a current signature the controller can trust |
| Thermal protector in the motor | Opens on winding or housing temperature and resets when cool | Responds to bulk temperature, so it is slow against a local commutator failure |
| Mechanical clutch or slip element | Limits the torque delivered to the mechanism | Protects the mechanism, not the motor. The winding still stalls |
| Design so stall is sustainable | Choose an operating point where the stall current is thermally acceptable indefinitely | Costs a larger motor, and it is the correct answer for a valve or clamp that stalls by design |
| Accept and specify it | State the maximum jam duration and count, and qualify the motor to it | Requires a real test, and the result belongs in the specification |
Diagnosing after the fact
A motor that has been stalled carries evidence, and it is worth reading before the unit is replaced.
- Compare cold winding resistance to the original value. A change indicates a damaged winding or a shorted turn.
- Compare no-load speed at a fixed voltage. A higher no-load speed with lower torque is the signature of a reduced torque constant, which means demagnetization.
- Inspect the commutator surface for a single burned or pitted segment pair. Damage concentrated at one angular position is a stall, not wear.
- Look at the grease. Grease that has been heated hard darkens and thins, and it will have migrated.
- Check the gear train for a single damaged tooth or a slipped press fit, at one angular position rather than distributed.
That distinction is the practical value of understanding this event. A returned motor with distributed wear is a life problem, and the answer is sizing or a different brush system. A returned motor with damage at one position is a control or mechanism problem, and no motor change will fix it.
Detecting a stall without adding a sensor
Most stall damage is preventable by removing power within a few seconds, and most compact mechanisms have no position feedback to detect the stall with. Three approaches work without one.
- Current threshold with a time limit. Stall current is several times running current, so a threshold is easy. The time limit is the important half: inrush at every normal start also exceeds the threshold, so the detector must require the condition to persist for longer than a normal start.
- Commutation ripple. A turning brushed motor produces a periodic ripple in its current at the commutation rate. When the rotor stops, the ripple stops. Detecting the absence of ripple distinguishes a stall from a heavy load, which a threshold alone cannot do.
- Back-EMF during an off period. On a PWM drive, the voltage across the terminals during the off portion of the cycle contains the back-EMF, which is proportional to speed. A stalled rotor produces none.
Whichever is used, the recovery behavior needs as much thought as the detection. A controller that retries immediately turns one stall into a repeated stall, and repeated short stalls do more cumulative damage than one long one, because each retry repeats the inrush at the same commutator position.