PMSM technology, towards the advanced motor control

Sensorless Field Oriented Control (FOC) for high energy efficiency motors.

Two are the primary driving factors behind the adoption of advanced motor control systems based on Permanent Magnet Synchronous Motors (PMSMs) with sensor-less Field Oriented Control (FOC): improving the energy efficiency and strengthening the product differentiation. Although it has been demonstrated that a PMSM with sensor-less FOC hits both targets, the success needs a design ecosystem that provides a holistic implementation approach. A holistic ecosystem will allow designers to overcome the implementation challenges that have hindered its adoption.

PMSM technology, towards the advanced motor control. Why PMSM?

A PMSM is a brushless motor that uses the electronic switching. It is often confused with the Brushless DC (BLDC) motor, another member of the family of brushless motors that also uses the electronic switching but features slight manufacturing differences. The PMSM construction is optimized for FOC, while the BLDC motor is optimized to use a six-phase switching technique.  The optimization allows providing the PMSM with a sinusoidal Back-Electromotive Force (Back-EMF) and the BLDC motor with a trapezoidal Back EMF.

The position sensors of the rotor used with each of these motors are different, too. The PMSM typically uses a position encoder whereas BLDC motors exploit three Hall sensors for their operation. If the cost is a problem, designers can consider the eventual implementation of sensor-less techniques that eliminate the cost of the magnet, of sensors, of connectors and of the wiring. The elimination of sensors also improves the reliability since there are fewer components that can potentially breakdown in a system. When we compare a sensor-less PMSM with a sensor-less BLDC, the sensor-less PMSM that uses a FOC algorithm offers better performances, using a similar hardware design and with a comparable implementation cost.

The applications that will most benefit from a shift to PMSM are those that currently use a Brushed DC (BDC) or an AC Induction Motor (ACIM). The main advantages of the migration include a lower energy consumption, a higher speed, a more uniform torque, a lower audible noise, longer duration and smaller sizes, so making the application more competitive.    However, to reap these highlights deriving from the use of a PMSM, a developer must implement the most complex FOC control technique together with other specific algorithms of the application to satisfy the system requisites. In fact, even if a PMSM is more expensive than a BDC or an ACIM, it substantially offers more advantages.

Implementation challenges

The implementation of the advantages of the use of a PMSM, however, needs the understanding of the hardware complexities concerning the implementation of a cutting-edge FOC motor control technique and the required domain expertise. The figure 1 shows a PMSM sensor-less three-phase control system that uses an inverter with three-phase voltage source. The inverter control requires three pairs of PWM high-resolution signals that are interconnected and numerous analogue feedback signals that need the signal conditioning.  The system also needs fault tolerance hardware protection functions, designed by using high-speed analogue comparators for a fast response. These additional analogue components required for the detection, the control and the protection increase the cost of the solution, since they are not demanded for a typical scheme of BDC motor or for the simple Volt per Hertz (V/F) control of an ACIM.

There is also the necessary development time to define and to validate the specifications of the components for the control application of the PMSM motor. To face these challenges, designers can select a microcontroller that can offer a high level of analogue integration with the specifications of the customized device for the control of the PMSM motor. This will reduce the number of external necessary components and will optimize the Bill Of Material (BOM). Currently, highly integrated motor control devices are available, with high-resolution PWM to facilitate the implementation of advanced control algorithms, high-speed analogue peripherals for precision measuring and signal conditioning, hardware peripherals needed for the functional safety and serial interfaces for communication and debug.

The interaction between the motor control software and the electromechanical behaviour of the motor itself is as difficult. The figure 2 shows the standard block diagram of the sensor without FOC. To shift to an effective design concept, we need to understand the architecture of the controller and the instructions of the digital signal processor (DSP)for the implementation of math-intensive and time-critical control loops. 

To obtain reliable performances, the control cycle must be carried out within a PWM period. There are three reasons for which it is necessary to optimize the control circuit times.

1) Constraint: using a PWM switching frequency equal to 20 KHz or higher (time lapse of 50 µS) to suppress the acoustic noise from the inverter switching.

2) To achieve a control system with higher bandwidth, the control loop must be executed within a PWM period.

3) To support other activities in background like the system monitoring, the specific functions of the application and the communication, the control cycle must be performed even faster. Consequently, the FOC algorithm should aim at an execution in less than 10 µs.

Many manufacturers provide examples of FoC software with sensor-less estimators of the rotor position. However, even before the motor can start rotating, the FOC algorithm must configure various parameters, so that they correspond to the motor and to the hardware. A further optimization of control parameters and of coefficients is necessary to hit the targets of demanded speed and efficiency. This is attained through a combination of:

1. derivation of parameters by using the technical card of the motor and 
2. experimentation with a trial-and-error method. Developers will have to turn to the trial-and-error method when the motor parameters might not be always characterized with precision in the technical card of the motor or when designers will not have access to high- precision measuring instruments. This process of manual tuning requires time and expertise.

PMSM motors are used in several different applications, operating in different environments or with different design constraints. In the case, for instance, of a car radiator fan, it is possible that the fan blades rotate freely in the inverse direction owing to the wind when the motor is about to be started. Starting a PMSM motor with a sensor-less algorithm in this condition is a challenge and can potentially damage the inverter. A solution consists in detecting the rotation direction and the rotor position and using this information to slowdown the motor until the stop through an active braking before starting the motor. Likewise, implementing additional algorithms might be necessary, such as the Maximum Torque Per Ampere (MTPA), torque compensation and field weakening ^[1] etc.  
These kinds of additional application-specific algorithms are necessary to develop a practical solution, but at the same time they also add design complexity, extending developing times and complicating the software verification.

For designers, a solution to decrease the complexity resides in creating a modular software architecture that allows adding specific algorithms of the application to the FOC algorithm, without influencing the time-critical execution. The figure 3 shows the software architecture of a typical application of real-time motor control. At the core of the framework there is the FOC function, which has a rigid time constraint and many additional specific functions for the application. A state machine inside the framework interfaces these control functions with the main application. The architecture needs a well-defined interface among the function blocks of the software to make it modular and to facilitate the user-friendly code maintenance. A modular framework supports the integration of different specific algorithms of the application together with other routines of monitoring, protection and functional safety of the system.

Another advantage of a modular architecture is the separation of the peripheral interface layer (or hardware abstraction layer) from the motor control software, which allows designers to migrate their IP from a motor controller to another when the application characteristics and the performance requisites change.

Requisites for a complete ecosystem

Facing these challenges requires a tailored motor control ecosystem for sensorless FOC designs. The motor controller, the hardware, the software and the development environment should collaborate to simplify the implementation process of advanced motor control algorithms. To succeed in it, the ecosystem is requested to have the following characteristics:

1. A high-level instrument to automate the measurement of motor parameters, to design control loops and to generate source code allows designers without domain expertise to implement the FOC motor control and to write and execute the debug of complex codes that generally take a long time;
2. An applicative framework for FOC and different additional specific algorithms for the application shorten development and test times;
3. The motor controllers with deterministic response and analogue integrated peripherals for the signal conditioning and the system protection in a single chip reduce the total cost of the solution.

The figure 4 shows an example of ecosystem architecture for motor control that includes the application framework and a development suite for Digital Signal Controller (DSC) dsPIC33 of high-performance motor control.

The development suite is implemented around a FOC software development tool based on GUI that can measure the critical parameters of the motor and automatically adjust the feedback control gains. Moreover, it generates the source code demanded for a project created in the development environment by using a Motor Control Application Framework (MCAF). At the core of the stack of solutions there is the Motor Control Library, which allows implementing the time-critical control loop functions of the application and interacting with the control peripherals of the motor in the dsPIC33 DSC.  This GUI operates in combination with several motor control development boards at disposal to support the extraction of motor parameters and the generation of the FOC code for a broad range of LV and HV motors.

The shift to brushless motors has been motivated by the need of a high energy efficiency and product differentiation. A complete motor control ecosystem provides a holistic approach to simplify the implementation of sensor-less FOCS with PMSM and should be composed by controllers for dedicated motors, development boards for rapid prototyping and user-friendly FOC development software to automate the code generation.

(by Nelson Alexander)