How do hardware accelerators simplify motor control applications?
Hardware accelerators have been an essential feature of microprocessors for decades, providing performance for features that the processor is unable to execute efficiently itself. Many times, this circuitry has been dedicated to performing floating point calculations or complex mathematics to bridge a performance gap of the processor’s instruction set. The advantage they often bring is determinism to calculations that, without them, would complete in varying time periods. In the world of control systems, such as motor control, this reliable determinism is highly beneficial.
Brushless DC motors (BLDC) have changed the motor control landscape significantly, allowing smaller, lighter, and more efficient motors to be integrated into applications where brushed DC motors simply were not viable. Control solutions for such motors can be very simple, relying on a commutation circuit that consists of little more than a Hall sensor and a switch. However, for advanced and efficient operation, a microcontroller is required.
Microcontroller-based commutation requires that rotor position is correctly determined, often through measurement of motor phase voltages and currents. This needs to occur several times per electrical revolution, and significantly more per mechanical revolution, to guarantee smooth operation. This demands that the microcontroller can complete the necessary mathematical calculations based upon incoming sensor data prior to the next commutation point. As rotation speed of the motor increases, the window for completion of these calculations narrows. If the calculations are non-deterministic, it can be very difficult to develop an optimal solution that reliably supports high rotational speeds.
Vector or field-oriented control (FOC), first proposed back in the 1960s, relies upon a series of mathematical equations, known as Clarke and Park transformations, that challenge the instruction set of most processors. To combat this, Toshiba has developed a hardware accelerator that embeds the necessary 2 to 3 and 3 to 2 phase conversions, space vector modulation, and rotating coordinate conversions required.
This module, known as an Advanced Vector Engine (A-VE), is tightly integrated alongside an analog-to-digital converter (ADC) and a pulse-width modulation (PWM) block to provide a highly deterministic, relatively autonomous FOC solution. It is embedded alongside an Arm® Cortex®-M4, 32-bit processor, 512kB of flash memory and 32kB of SRAM as part of a microcontroller solution. In total, the M4K group of products from the TXZ4 series of microcontrollers can control up to three motors simultaneously. Motors can be supported at rotational speeds of up to 250,000 rpm with sensorless FOC.
To find out more about the technology behind the Advanced Vector Engine and how the TXZ4 series of microcontroller can be used to implement FOC motor control, take a look at our white paper available here:
