Motor control is a key function of any microcontroller (MCU) because the processing power needed to enable the application is inherent within the integrated circuits (ICs). But the breadth of motor-control applications and motor types is astounding. As a result, design teams need to carefully match the application with the right type of MCU to minimize implementation costs and ensure performance to the design specification. In some cases, low-end MCUs will suffice whereas other applications require higher performance controllers.
MCUs with an architecture optimized for the fast operational needs of digital signal processing (DSP) capabilities are commonly referred to as digital signal controllers (DSCs). DSCs are increasingly popular in motor-control applications because the math capabilities enable complex control scenarios on relatively-low-cost ICs.
Freescale has long been a leader in the DSC trend and provides a good point of reference as to how DSCs compare to other MCUs in motor control. The company positions its 16-bit, entry-level MC56F8006/2 DSCs in the middle of its motor-control portfolio. Generally the math capabilities make a DSC a good fit in applications where the processor must handle real-time capture of analog inputs. Once captured, the DSC analyzes the input and performs the real-time control algorithms. Higher-end MCUs, such as those based on the power architecture, can bring even more processing power to bear on such tasks, but DSC products are more cost effective because they can be scaled for the application.
The variety of products that use DSCs ranges from home appliances to power-sensitive applications and handheld power tools to medical systems. Freescale MC56F8006/MC56F8002 DSCs are appropriate for complex scenarios while the MC9SO8MP16 8-bit MCU family is targeted at entry level brushless DC (BLDC) motor control. The DSCs can serve in three-phase BLDC motor applications, in entry level field-oriented control (FOC) applications, and in permanent magnet synchronous motor (PMSM) applications.
DSC architecture matches motor application
A look at the MC56F8006/2 architecture (Figure 1) details reveals the reasons the family is a good match for motor control applications. The 56F800E processing core operates at a maximum clock speed of 32 MHz, but the timer and pulse-width-modulation (PWM) peripherals that are key in motor control applications operate at 96 MHz. The PWM module features six outputs. The DSCs integrate dual 12-bit ADC and can convert a sample in 3.03 µsecs at maximum clock speed. The ICs include three analog comparators that are also important in motor control.
Figure 1: The Freescale MC56F8006/2 DSC family includes the analog comparators, PWM, and data-converter peripherals needed to implement sensorless FOC control on the math-centric processor core. (Source: Freescale)
The math capabilities of the MC56F8006/2 are enabled by a 16x16-bit parallel MAC (multiplier accumulator) that is optimized for functions like matrix math. The architecture also includes four 36-bit accumulators for increased speed.
Moving further up the Freescale DSC line, the MC56F824x and MC56F825x also use the 56F800E core but push clock speed to 60 MHz making them ideal for many applications. The DSCs add other features such as 5-bit Digital-to-Analog Converter (DAC) references for the three analog comparators that can enable more precise control scenarios.
DSCs enable sensorless FOC applications
Microchip recommends its dsPIC DSCs for complex control scenarios such as sensorless FOC algorithms. The FOC approach is proving popular in applications where a motor faces a dynamically changing load, for example, a washing machine.
Traditionally, washing machines have utilized BLDC motor drives and six-step or trapezoidal control schemes. In such control systems, Hall sensors provide rotor-position information to the controller but accuracy is limited to the discrete sensor locations. The sensors don’t provide the continuous feedback required to make changes as washing cycles get more complex. Moreover, the load on the motor changes constantly and dynamically – especially in front loading machines where the weight of the wash load impacts the motor.
FOC designs that work with PMSM motors can sense the position of the rotor continuously by monitoring the back electromotive force (EMF) voltage that is present in the stator winding. The PMSM motors add the advantage of quieter operation and are generally considered more powerful relative to size compared with BLDC motors.
Figure 2 depicts a Microchip dsPIC30F-family DSC in a sensorless control design. There are no position sensors on the motor shaft. Instead, the design uses resistors inside the 3-phase inverter to make current measurements on the motor. The FOC algorithm, sometimes referred to as vector control, requires that the DSC perform complex math operations such as Clarke and Park coordinate transforms. Moreover the DSC ultimately must generate the PWM signals using techniques like Space Vector Modulation.
Figure 2: Microchip’s dsPIC30F DSC family takes input via resistors in the 3-phase inverter to continuously sense EMF and accurately determine rotor position. (Source: Microchip)
The dsPIC architecture is a good match for the application. The 16-bit family integrates a 17x17-bit single-cycle MAC. Moreover it includes two 40-bit accumulators to accommodate saturation bits. The DSCs also include a 40-stage barrel shifter. There are additional peripherals that serve to enable the application as well. For example, the dsPIC family includes a quadrature encoder interface along with the requisite PWM and data-converter peripherals.
The 32-bit DSC trend
While the Microchip and Freescale products are 16-bit devices, there is also a trend toward 32-bit DSCs for motor-control applications. Texas Instruments (TI) supports entry-level control scenarios with its 16-bit MSP430 MCU family, but more aggressively targets the 32-bit C2000 Piccolo family when the task at hand is FOC or vector control. TI doesn’t necessarily use the DSC moniker with the C2000 MCUs, but it is a fitting description. The processor core includes a 32-x32-bit single-cycle multiplier, and some family members include a floating-point processor called the Control Law Accelerator for more-precise algorithms.
There are a number of other suppliers offering 32-bit DSCs based on the ARM Cortex-M4 architecture, including Freescale, with its Kinetis family and the STMicroelectronics STM32 family. In addition, NXP Semiconductors was the first company to deliver a DSC based on the ARM Cortex-M4 architecture with its introduction of the LPC4300 family last year.
The Cortex-M4 design includes three separate MACs – one can operate on 32-bit data and the other two on 16-bit implementations. It is at the discretion of each DSC vendor to complement the core architecture with the PWM and data-converter peripherals required for motor control.
32-bit value proposition
The main advantage of a 32-bit processor is relative to the entire system design. Many projects may include multiple motors and some 32-bit processors can control three or more motors simultaneously.
Designers must take the full system into account when considering components such as MCUs and DSCs. In addition to motor control, a 32-bit MCU or DSC might be capable of hosting all of the features defined in a design specification – such as a touch-based user interface and communication capabilities. On the other hand, a 16-bit DSC might be perfectly capable of the motor-control task, but not able to handle the other system elements.
Conversely, the project at hand might require a distributed intelligence approach. Perhaps the motor-controller must be placed in a rugged environment and therefore requires a specialty IC that works at higher temperatures. Using a cost-optimized DSC might be the best option because it allows designers to leave the remainder of the system complexity for implementation in a protected environment using a standard, commercial-grade MCU. Moreover, the DSCs have headroom to support other tasks. As always in engineering, there is no universal answer, but it’s clear that today’s design teams have several viable options for motor control.