Over the last decade, electronically controlled motor drives have become vital to automated machinery from traction and propulsion to servosystems, robotics, and automated machine tools. Developments in digital and power electronics have provided high degrees of control at affordable prices.

And, with low inertia and wide availability, the ac induction motor is the workhorse of the industrial motor drive industry. But, even with simple control needs, further improvements in induction motor performance and efficiency are possible using sophisticated digital-control algorithms to handle variations in parameters such as torque and load.

The dynamic requirements of today's motion control systems place significant computational demands on the CPU at the heart of electronically controlled motor drives. Although many 16- and 32-bit microcontroller units (MCUs) and 16-bit digital signal processors (DSPs) have been successfully used in motion control applications, they have certain limitations.

These roadblocks include: inability to provide a wide dynamic operating range over the various motor parameters; operating with variable data types; providing fast table look-up and interpolation features; and simultaneously servicing a fast multi-level interrupt structure. However, merging controller and processor functions into a single processing core, such as Siemens TriCore^TM architecture, overcomes these limitations for more efficient and economical control.

Motion control shortcomings. In many ac-induction-motor motion control applications, the combination of MCU and DSP chips or a hybrid chip with both cores is typically used to handle the system initialization, commutation, and feedback control-loop requirements. For example, the MCU typically uses the primary inputs of motor speed and rotor position to address a sinewave look-up table. By stepping through this table, the MCU determines the commutation times for the pulse-width modulation (PWM) signals, and then sinewave modulates them for driving the motor's phase windings. The MCU then adapts the desired current waveforms to reduce torque ripple resulting from harmonics in the motor winding back-EMF. The DSP performs autocorrelation, waveform filtering, and other computation-intensive functions, such as closed-loop proportional-integral differential (PID) algorithms.

While current embedded solutions address the drive's DSP system requirements, they are not the only demands on a high-performance drive system. Switching between DSP tasks and control tasks quickly and efficiently is one performance area where hybrid MCU-plus-DSP architectures experience memory bottlenecks, slow task-switching times, and poor interrupt latency. The latest merged-architecture MCU/ DSPs answer these shortcomings and provide a controller optimized for embedded DSP tasks as well as precisely tuned to the needs of real-time embedded task switching.

Meeting vector-control demands. For example, in certain applications, field-vector control, with the induction motor operating as a servo motor, replaces the traditional PWM technique for controlling the various phases of ac induction motors. AC vector control requires the coupling of fast task switching with DSP processing of time-critical loops. This operation cannot be controlled by the PWM process alone, so vector-controlled inverters have emerged. The space-vector-controlled inverter creates the voltage, current, and frequency necessary for field orientation to take place inside the ac induction motor.

The adroit merged architecture is well suited to this application. To perform the field orientation process, the flux vector drive needs to know the spatial angular position of the rotor flux inside the motor. Functions required for induction drive-control systems include:

PID transfer functions

DSP functions typically use finite precision arithmetic and allow a hardware multiply-accumulate unit (MAC) to handle such controller coefficients and system operating parameters as velocity, position, and acceleration. The MAC implements all the traditional DSP functions, such as signal filtering, equalization, and modulation. To ensure overall motor stability and robustness, MAC instructions must be carried out within the shortest execution time possible, preferably within one controller instruction cycle.

In addition, more sophisticated control algorithms provide improved system dynamics and precision which demand greater computing power and resolution. A common characteristic of all algorithms used is a basic structure based on a fast internal current (torque) control loop and outer control loops for speed and position.

Tomorrow's dictates. With the thrust in ac induction-motor drive technology toward increased processing throughput, lower cost, and improved system reliability, future programmable motor drive systems will demand more powerful 32-bit MCU/DSP derivatives. These will drive down the amount of code required to accomplish a task, and provide a more real-time responsive system than traditional control loops. Increased system flexibility will also require future MCU/DSP offerings to provide non-volatile, on-chip flash memory that can be programmed remotely to meet a wide variety of different drive system requirements.

Single-chip solutions are absolute imperatives for all future embedded motion control systems to reduce component costs, power consumption, design times, and system size.

What merged, single-chip control/ processing architecture may mean to you: