Number-crunching FPGAs take the strain
01 May 2012
Motor control algorithms are evolving and so too are the electronic platforms upon which they run. Ralph Bergmann takes a look at the FPGA, and how it can provide a single-chip solution to even the most complex of motor control applications
The author, Ralph Bergmann
It is estimated that electric motors now consume over 50 percent of all electricity generated; a staggering figure but one that is easily accepted given the proliferation of electric motors in our modern world. It is little wonder, too, that there is increasing pressure at all levels to improve the efficiency with which electric motors commutate physical work. Amongst semiconductor vendors, this area of industrial control constitutes significant opportunity which is being addressed through an burgeoning range of techniques.
But motors aren’t exclusive to industrial applications; the number of electric motors in commercial and domestic devices is increasing too, giving rise to demand for more efficient control. For ‘simple’ motors, control can be achieved using low-cost 8-bit microcontrollers but even in low-end applications demand is mounting for more sophisticated control algorithms that are able to get maximum efficiency and performance from even the lowest cost motors.
The proliferation of microcontrollers has promoted the ‘software-only’ approach to motor control, where general purpose peripherals such as analogue-to-digital converters and pulse-width modulators are fully integrated, allowing ‘closed loop’ control to be realised in a single device. However, this inherently ‘real time’ application puts a strain on the limited performance of low-end microcontrollers, particularly as control algorithm complexity increases.
One solution to this is to offload some of the heavy computations to dedicated hardware, in the form of a field programmable gate array (FPGA). This has the advantage of allowing motor control algorithms to evolve without changing the underlying hardware; it also supports deterministic execution even as algorithm complexity increases. If the microcontroller is implemented as a soft processing core within the FPGA, a single-chip approach can also be maintained.
The concept of a ‘software-only’ approach to control is evolving, particularly in applications that use brushless dc motors and permanent magnet synchronous motors. A technique that is growing in popularity, thanks to its efficiency benefits, is field oriented control, or FOC. This is a purely algorithmic approach to motor control and, as such, has high processing demands. However, this is offset by the potential energy savings it can deliver over the lifetime of the motor.
FOC (vector control as it is also known) is an extension of the proportional integral derivative (PID) method of control, where the magnetic field generated by the stator is varied to control the speed, flux and torque of the motor. As this is a computationally-intensive approach to control, it is beyond the capabilities of most low-cost general purpose microcontrollers, giving rise to a range of application-specific microcontrollers from a growing number of vendors.
Often these devices will use dedicated hardware blocks to achieve the necessary performance. The issue with this approach is the performance bar is effectively set by the hardware, meaning future improvements are limited. Conversely, by adopting an FPGA approach not only can the dedicated hardware acceleration blocks be accommodated but they can be augmented in future iterations.
Multiple motor control
Another trend within motor control is consolidation; using a single device to control multiple motors and often multiple motors with different topologies. For a general purpose microcontroller this would put massive pressure on the software’s execution but for an FPGA it is limited only by the hardware resources. For example, using the latest analogue-to-digital blocks in Xilinx’s Series 7 FPGA family, it is possible to create a fully autonomous motor control subsystem in a low cost FPGA.
As well as being able to control multiple and varied motors, manufacturers are also looking to add network connectivity to their control systems. To support this, Xilinx has partnered with QDesys to develop a sophisticated IP building block approach to motor control supporting advanced modulation schemes, sensor-less as well as sensor-based FOC.
It offers two 10/100/1000 real-time Ethernet interfaces supporting EtherCAT, ProfiNet, Powerlink, SERCOSIII and several legacy protocols. It also provides multi-channel analogue Sigma Delta data acquisition for measuring the stator’s currents and the motor’s bus voltages, with an isolation barrier between the power and control elements. Two power bridges support mixed configurations of permanent magnet motors.
Using more sophisticated motor control algorithms has the potential to lower the total cost of ownership for a range of industrial and commercial motor applications, as well as reducing energy savings that, in turn, translate to lower CO2 emissions. Through innovative design and the flexibility provided by the latest generation of FPGAs, it can also be achieved cost effectively for manufacturers.
Ralph Bergmann is with Xilinx
Contact Details and Archive...