Current Monitoring & Detecting Motor Stalls

This document serves to educate about what a motor stall is, how they can be detected, and how to stop them to prevent damaging a motor.

Understanding a Motor stall

A Motors power can be caluclated with the following equation:

where $P$ is the power produced in Watts, $I$ is the current draw of the motor in Amperes, and $R$ is the resistance of the motor windings in Ohms. This formula is only appropriate in the case that the motor is 100% efficient. A more realistic equation would be as follows:

Where E, the motor effecincy is a percentage less than 100%. A manufacturer should provide an efficiency graph of their motor, for instance, here is one for the REV Neo Vortex:

A motor stall occurs when a motor is held stationary, preventing movement, but is requested to move. As you can see in the graph, when stationary, the motor is generating the most amount of torque (which if not stalled would cause it to move), and drawing insane amounts of current. At this moment, the motor is 0% efficient, so if movement is restricted, it results in all of the power going directly into heat. Therefore when a motor is stalled, current draw in the motor is proportional to heat as shown by the equation:

In general heat is bad for motors, when a motor gets hot it becomes less efficient, and if the motor becomes too hot, it may become damaged; such as the motor windings melting together causing a short.

Below is an image of a simulated stall condition using the WPILib simulator, where you can see, as the motor is stopped, the current proportionally increases as an attempt to rectify the situation.

In the simulation, the "motor" is being allowed to run for 5 seconds, then its velocity is forcibly brought to zero (simulating a stall condition) for another 5 seconds. This then causes the motor controller to try and rectify the "stall" by increasing its current output.

Detecting a Motor Stall (or any Current Spike)

Using a Debouncer

Using WPILib, we can use a Debouncer and the PowerDistribution class to detect a spike in the current. A debouncer, checks that the provided boolean value stays in the desired condition (true or false) for longer than a specified amount of time before it declares it as valid.

public class Robot extends TimedRobot {
 
    private static final double kStallCurrent = 35; // Amps
 
    private Debouncer m_debouncer;
 
    PowerDistribution m_powerDist;
 
    @Override
    public void robotInit() {
        // Time in seconds value must remain true to be considered true.
        double debounceTime = .25;
        // Debouncer is ensuring the value is staying true.
        DebounceType debounceType = DebounceType.kRising;
        m_debouncer = new Debouncer(debounceTime, debounceType);
 
        pdp = new PowerDistribution();
    }
 
    @Override
    public void teleopPeriodic() {
        // Get the current from channel 0 on the PDP/PDH.
        double currentDraw = m_powerDist.getCurrent(0);
 
        // Check the debouncer state, it will return false until (currentDraw > kStallCurrent)
        // has been true for the time provided in its constructor.
        boolean isStalled = m_debouncer.calculate(currentDraw > kStallCurrent);
    }
}

Considering Detection Time & Stall Current.

Some consideration is required when finding good values for your debounce time and stall current.

The first thing to tune should be the stall current. The stall current should be marginally above your idle current draw. In the example below the idle current draw is about 30 Amps, so a conservative value of 35 Amps was chosen to be considered as a stall.

Secondly, the debounce time, when a motor starts from rest (or changes speeds at all), a change in current will be seen. The problem with this fact is that, this is not a stall state, the motor is just getting up to speed. To get around this, a value should be chosen that is marginally above the time it takes for this initial current spike to die down. In the example below, it takes ~0.25 seconds in order to get the stopped motor up to speed, or close, but below the stall threshold. A value of 0.3 seconds was chosen as to mitigate false triggers while catching stalls as early as possible.

Both values should be tuned to be as small as possible in order to detect stalls as quickly as possible, but be cautious that you do not go too small and begin to have false positives.

Once a stall has been detected, your goal now becomes to stop it, whether this be to disable the motor, or start some other subsystem to "un-stall" the motor. Below is a simulated example of a motor stalling, and once it is stalled, it is set to neutral.

Usage

CHARGED UP

This strategy was used in the 2023 CHARGED UP season: code link (opens in a new tab).

Once the intake grabbed a game-piece, it would stall the intake motors. To prevent damaging both the game-piece or the motors, stall detection was implemented to run the hold game-piece Command, and to also give the drivers feedback so they would know a game-piece was successfully collected.

WPILib Simulation

The simulation code used for this document can be found on GitHub here (opens in a new tab).