By 2035, the world will consume over 35 trillion kilowatt-hours of electricity annually, up from less than 21 trillion kilowatt-hours in 2015, accounting for nearly one-third of the current electricity demand for industrial production. Many of these motors are based on simple alternating current (AC) designs because they are relatively inexpensive and easy to drive. They are also inefficient in terms of energy use, especially at low speeds. However, these AC motors are not inherently wasteful. With the right form of electronic control, their efficiency can be significantly improved. Using control technologies available today, energy consumption at a given operating level can be reduced by up to 60%.
The simplest control technique currently in use is volts per hertz. This is conceptually simple and easy to implement on a basic microcontroller. The core algorithm fully utilizes the core characteristics of AC motor design. Every motor has a characteristic magnetizing current and a final magnetic flux and torque. These properties are related by a volt-per-hertz ratio. The motor rotates the mechanical load by switching stator coils arranged around a moving rotor. The switching between coils forces the rotor's magnetizing elements to rotate sympathetically to a stable state where the magnetic field remains balanced.
Increasing the coil switching frequency, in turn, increases speed. However, if the supplied electrical energy does not increase accordingly, the applied torque will decrease. Volt-per-hertz control provides a simple way to solve this problem by increasing the frequency through increasing the line voltage, allowing the torque to be maintained at a constant level. Unfortunately, this relationship is not particularly consistent at low speeds. Higher voltages are needed to maintain high torque at low speeds, but this reduces efficiency and increases the likelihood of coil saturation and overheating.
Field-oriented control offers a method for optimizing motor control, particularly at low speeds, and also allows for more precise positioning control of the motor. This expands the overall application range of AC motors, which helps reduce the cost and operating costs of industrial machinery.
In field-oriented or flux vector control, the relationship between speed and torque is broken down by volt-per-hertz (V/Hertz) control. The idea behind field-oriented control can be expressed using a DC motor model, where the current supplied to the rotor and rotor is independent. In this model, torque and the generated flux can be controlled independently. The field strength within the motor, generated by the current, determines the flux. The current supplied to the electromagnetic coils in the rotor controls the torque—as the magnetic field attempts to reach a steady state.
A DC motor uses a commutator on the rotor, which controls which coils on the stator are energized at any given time. The commutator is designed so that current is switched to mechanically aligned windings, thus generating maximum torque at that point. As a result, the windings are managed in a way that changes the magnetic flux to keep the magnetic field generated in the rotor windings orthogonal to that generated in the stator.
In AC motors, only the stator current is directly controlled. The rotor typically uses permanent magnets to provide its field. This means that flux and torque depend on the same current. However, field-oriented control provides the ability to manipulate them almost independently. In effect, the stator flux is dynamically controlled to provide the ability to manipulate torque independently. Typically, stator coils can be driven so that they either generate torque or apply force along the stator axis, a mode that does not affect rotation. These directions are the orthogonal axis and the direct axis, respectively. For motion, each coil is driven sequentially to generate a high orthogonal force.
Several mathematical transformations are used to provide the ability to change current and voltage to decouple torque and flux. Under field-oriented control, the current flowing through different parts of the stator is represented by vectors. Matrix projection transforms the three-phase time- and velocity-dependent system into two coordinate time-invariant systems. The coordinates are typically described using the symbols d and q, representing the flux and torque components, respectively. In the (d, q) reference frame, the applied torque is linearly related to the torque components.
In field-oriented control, electrical signals are received from the motor and incorporated into a (d, q) coordinate model. This model is typically calculated relative to the rotor, making it easier to calculate the required flow rate. A typical approach for this calculation is to pair the Clarke and Park transformations.
The Clarke transform takes currents from different phases (typically three) and uses them to estimate the current in a Cartesian coordinate system. The axes of these systems use the symbols α and β instead of the traditional x and y to reduce the possibility of confusion with the spatial coordinate system. These are then applied to the Park transform to provide the current vector seen in a rotated (d, q) coordinate system. Trigonometric functions provide the core of the transformation, requiring the use of a microcontroller or digital signal processor (DSP).
The flux and torque components of the current vector in (d, q) space are obtained from the current fed to each electrical phase and the rotor flux position via Clarke and Park transforms, which are described in most of the text using the symbolic theta algorithm. This structure is applicable to a series of motors. The inverse Park transform is used to generate the voltage output, which is then used in the algorithm to control the power of each of the three phases. The overall structure is shown in Figure 1.
Figure 1: Basic configuration of the transformation and control block used for magnetic field orientation control.
By simply changing the flux reference and obtaining the rotor flux position, the same core structure can be used to control both synchronous and induction motors. In a synchronous permanent magnet motor, the rotor flux is fixed because it is determined by the permanent magnets. An induction motor requires the creation of rotor flux to function, so this is incorporated as a non-zero value into the flux reference.
The key to successful field-oriented control is real-time prediction of the rotor flux position. This control strategy is complex. Inside an AC induction motor, the rotor's rotational speed does not match the speed of the magnetic flux driving its rotation. The rotor tends to lag, resulting in a discrepancy known as slip. In previous approaches, motor manufacturers used sensors to analyze the rotor position, but this incurred unnecessary additional costs. In practice, feedback from voltage and current generated within the motor can be used to compensate for slip.
Many systems use measured back electromotive force (EMF) to estimate rotor slippage. The magnitude of the back EMF voltage is proportional to the rotor speed. However, directly using this input can lead to problems at low speeds or when the rotor stalls, and estimating the initial position is not easy. Starting from an unknown rotor position can cause the motor to unexpectedly reverse a short distance or cause the motor to fail to start completely. Another drawback of simply sampling the back EMF is its sensitivity to stator resistance, which is prone to changes with temperature.
Indirect model-based schemes offer higher performance. There is a significant trade-off between computational overhead and performance, but overall, efficiency can be improved, especially at low speeds, by using more sophisticated model-based algorithms. Indirect model-based schemes estimate real-time values based on available sensor readings.
Similar to back EMF estimation, the core issue is determining the starting point of the motor. One solution is to begin with an estimate of the initial state from which a vector of predicted output can be derived, and then compared with the measured output vector. This difference is used to correct the model's internal state vector. However, noise can degrade the model's stability.
The Extended Kalman Filter (EKF) can compensate for the effects of noise and sudden disturbances. The architecture of the Kalman filter assigns higher weights to updates considered to have lower uncertainty than those estimated to have higher uncertainty. The filter operates recursively, such that each estimate requires only a new set of readings and the filter's previous states to generate a new state.
The Kalman filter employs two main phases: prediction and update. In the prediction phase, the filter calculates the next state of the system based on the previous state, which, in the case of a motion algorithm, provides the last known velocity and acceleration values. From this, the filter calculates a prediction of the current position.
During the update phase, the newly sampled voltage and current values are compared with their predicted values. The closer the input data is to the prediction, the lower the error probability. This error probability is fed back into the Kalman filter gain. At the algorithm level, the Kalman filter relies on some matrix multiplications and inversions. Therefore, the key to implementing an extended Kalman filter in motor control lies in high arithmetic performance, consistent with other aspects of field-oriented control.
To perform the numerous arithmetic operations required per second in real-time motor control scenarios, a high-performance MCU or DSP is needed. Texas Instruments' (TI) TMS320F2833x series of devices are specifically developed to handle the typical computational loads of AC motor applications and are supported by a variety of on-chip peripherals to facilitate integration with power conversion electronics.
The TMS320F2833x is built with a high-performance 32-bit CPU, supports floating-point operations, and conforms to the IEEE 754 standard for single-precision arithmetic. By implementing IEEE-compliant floating-point units, the TMS320F2833x simplifies algorithm development because it handles a very wide range of numbers and has built-in support for errors such as Not-a-N (NaN) and division by zero. The Harvard architecture, combined with dual 16×16 multiply-accumulate (MAC) units, provides high throughput for matrix and projection-based operations. For even higher precision, the units can be chained together to perform 32x32 MACs. On-chip peripherals include a 16-channel analog-to-digital converter (ADC) for sampling voltage and current feedback signals from the motor.
As a member of the C2000 series of DSP-enhanced MCUs, the TMS320F2833x is supported by the TI Digital Motor Control Library, which provides reusable, configurable software modules to implement various control strategies. The library consists of functions represented as blocks, providing not only control blocks for closed-loop operation but also transforms such as Clarke and Park, as well as peripheral drivers for functions such as pulse width modulation (PWM).
In motor control scenarios, the PWM output controls six power transistors, which collectively provide voltage and current to three electrical phases. A half-bridge transistor configuration is used in each phase. A common algorithm used for control in these cases is space vector PWM. This reduces harmonics compared to simple PWM techniques and employs eight switching states. There are six active states and two zero states, each a target state for eight corresponding space vectors. These states are arranged such that two sets of complementary states are active at any given time. One set is used for the three high-side power transistors, and the other for the low-side. The algorithm cycles through the states to switch power according to the requirements of the field-oriented control model. The TMS320F2833x includes PWM hardware suitable for software control using space vector switching. Six of the 18 total PWM outputs support high-precision control with a resolution of 150 ps. The result is a digital controller requiring relatively little external hardware to manage the power transistors, as shown in Figure 2.
Figure 2: Block diagram showing the control of the power phase by the PWM output of the F2833x.
in conclusion
With the help of microcontrollers featuring essential core and high-performance building blocks, along with TI's digital motor control library, designers are preparing to launch a new generation of high-efficiency AC motors.
