Smart Motion Makes For A Smarter Design

If it moves, jumps, rotates, or vibrates, it usually contains a motor. Electric motors come in all types of devices, from tiny hard-disk drives to hybrid vehicles to locomotives. Intelligent motor control can be employed over this wide range of devices, delivering improved efficiency, longer life, and better fault control compared to simply applying power to a motor. Key to greater use of intelligent motor-control systems are low-cost microcontrollers and digital-signal-processing chips that target this market.

Unfortunately, simply throwing a basic microcontroller at the problem is rarely an option. Digital design is hard. Analog design is harder. Combine the two disciplines and you've got the doubly difficult design task of digital motor control. Intensifying the problem is the need for a microcontroller with analog peripherals. On top of that, motor-control-oriented microcontrollers typically include a significant timing component.

Finally, there's the issue of power. Microcontrollers usually sink enough current for only the smallest of motors. Connect one directly to a one-horsepower (HP) electric motor, and the chip will burn out faster than a popping fuse. This is why most microcontroller-based solutions are coupled with external power transistors and often complex power control systems. We'll concentrate on the microcontroller aspects of the system here.

Motor-control microcontrollers must address a wide range of motor types (see "Motor Basics" at www.elecdesign.com, Drill Deeper 11291). A particular microcontroller tends to be tuned to a particular type of motor, such as a brushless dc (BLDC) motor. Given the need to support so many different motors with different performance characteristics, a wide variety of microcontroller solutions exists. Likewise, numerous microcontroller vendors supply motorcontrol microcontrollers (see "Need More Information?" p. 44).

SMART APPLICATIONS
Motors and electronics are ubiquitous. As embedded designers, we know that everything from kitchen appliances to MP3 players hosts them. Often, many intelligent motors are used in combination.

For example, take Segway's Human Transporter (HT) (Fig. 1). It uses a pair of Texas Instruments TMS320C2000 DSP platforms for closed-loop motor control and balance computation. This people mover is remarkably easy to ride because the electronics do the real work.

The Human Transporter has a simple design. It has only two moving parts—the wheels—with a fixed axle. The controller board is underfoot, and it includes the power transistors needed for motor control (Fig. 2). It works with the sensor platform, which contains a pair of gyroscopes and three accelerometers (Fig. 3). These sensors, which are checked 100 times/s, are as much a key to the Human Transporter's success as the software.

Dean Kamen, the Human Transporter's inventor, first developed the Independence IBot Mobility System for Independence Technology. This chair has more functionality than the Human Transporter, including the ability to climb stairs and raise the rider to the eye level of a standing adult. It's designed to give the disabled unprecedented mobility.

At the other end of the motor spectrum are tiny, mobile hard disks (see "Mobile Storage: Chips Served With Hard-Disk Salsa" at www.elecdesign.com, ED Online 10184). These devices employ one motor to rotate the disk and another to move the heads. In addition to the normal motor-control chores, the electronics in these tiny devices also use accelerometers to determine whether or not the drive is moving. That's because the heads must be locked down to prevent problems due to shock and vibration.

Designers of kitchen appliances with motors are turning to microcontrollers even for the least expensive products, due to their low cost and improvements to the product. For example, ramping speed up and down increases motor life and overall reliability. It's also easier to add motor-related features with no changes to the hardware.

Hybrid cars are a popular example for regenerative braking. Power recovery also is becoming more important for a range of products. For example, portable devices can extend battery life if it's possible to recover power when a motor is turned off.

SMART CONTROL
Motor-control-system configurations are relatively simple and consistent (Fig. 4). The differences typically-involve the number of phases, which dictates the number of power connections on the motor and the number of sensors within the system. Very simple systems don't employ a feedback mechanism.

Systems that employ feedback usually implement a PID ( Proportional, Integral, Derivative) feedback device, based on the relative rotational position of a motor's rotor. Optical sensors, Hall-effect sensors, and sensorless back EMF (electromotive force) systems are the most common feedback mechanisms.

Optical sensors use a rotating disk with slots that can be detected. The optical sensor digital outputs are fed into a quadrature encoder found on many microcontrollers. The encoder provides tachometer information as well as rotational direction.

Hall-effect sensors detect magnets mounted in the rotor or shaft. They provide positional and tachometer information via comparators.

For three-phase motors, sensorless feedback systems are the most popular (Fig. 5). The term sensorless indicates that, unlike optical and Hall-effect sensors, additional hardware isn't included in the motor. Sensorless systems take advantage of the fact that only two of the three windings in the motor are powered at any one time. That doesn't mean there's a lack of current flowing through the third winding, just that it's minimal and induced. The resistors used in this technique can be incorporated into the sensor inputs in the microcontroller. The detection circuit may simply check for the zero crossing. More sophisticated systems may employ analog-to-digital converters (ADCs).

Feedback provides a way to adjust and synchronize the pulse-width-modulation (PWM) outputs that drive the power transistors controlling the current to the motor. PWM generation for motor control usually requires more than just setting the high and low time periods, because the phase between pulses on different control lines is critical. At one extreme, for example, is Freescale's Enhanced Time Processing Unit (eTPU). The eTPU is essentially a programmable microcontroller designed for handling complex timing.

Another alternative is to customize the timing system to accommodate the motor-control requirements. For instance, it's usually preferable to have a short delay or dead time between the end of a pulse on one phase to the start of a pulse on another phase. This will prevent unwanted current flow through the windings of a multiple phase motor. Instead, the current flow for one part of the cycle stops before the next part of the cycle begins. Microcontrollers specifically designed for motor control normally implement this type of feature in hardware, minimizing processing overhead.

Motor control tends to be a repetitive process, with multiple sensor readings taken each cycle. To that end, Texas Instruments implemented a hardware-generated time stamp for captured information. This allows an application to determine when an event occurred so it can compute the values for next control settings.

Motor-control microcontrollers are a major part of the motor-control arena, but not the only solution. Motor-control chips like those from Performance Motion Devices and AMI Semiconductor are designed to interface with standard microprocessors using parallel, I²C, and SPI connections. These peripheral devices handle most motor-control chores by themselves, with the host setting the motor parameters. This approach offers a number of advantages, such as a simplified interface and the ability to put the motor-control functions on an adapter card. Moreover, there's no need for a motor-controlspecific processor. Chips are available for all types of motors, even stepper motors and micro stepper motors.

Hardware augmentation enables 8-bit microcontrollers to address a host of motor-control applications. Digital signal processors such as Microchip's dsPIC and Analog Devices' Blackfin are ideal, higher-performance motor-control applications. Still, even higher-speed chips tend to include hardware support that can reduce system overhead. This enables the processor to address additional, application related sensors.

SMART SENSORS
Many motor-control applications operate in isolation. It's either off or on at some desired speed. Recently, there has been a trend toward more integration with sensors related to, but outside of, the motor itself, such as the Segway and hard-disk applications mentioned earlier.

Part of this increase comes from new low-cost sensors based on nanotechnology, such as three-axis accelerometers (see "Accelerometer Offers Economical Low-G Sensing," ED Online 10221). In products like the Segway, they're used to maintain balance, while mobile hard-disk systems employ them to detect movement, like when the product is dropped. Temperature, current, and voltage sensors are often utilized as well. These will commonly find their way into fault detection (see "Whose Fault Is It Anyway?" Drill Deeper 11293).

Torque, load, and force sensors, such as those available from Lebow Products, are important because motor speed isn't always the only key factor. These types of sensors are often attached to the motor's shaft or even further down the power train.

Handling of specialized sensors often requires custom digitaltoanalog converters and ADCs. Many motor-control micro-controllers contain extra peripherals that can address this support. Others, like Cypress Technologies' PSoC, incorporate customizable hardware blocks. PSoC has both digital and analog blocks. Cypress Technologies offers a development kit that includes plug-in motors for working with its motor control blocks (Fig. 6). Its PSoC Express is a graphical programming environment that comes with the development kit (see "MCU Goes Graphical," ED Online 10215).

SMART TOOLS
Choosing a microcontroller with suitable motor-control capabilities is only the start, though. Software is still mandatory if things are going to work properly. Motor-control application notes are extremely popular, as are mechatronics development kits that include the microcontroller, development board, and one or more motors.

Zilog's Z8 Encore!MC Development Kit includes a hefty three-phase brushless dc motor along with a complete C-based development tool suite (Fig. 7). The carrier board contains the 8-bit microcontroller, which is smaller than one of the drive transistors on the base board.

In addition to app notes, motor-control microcontroller vendors now include run-time libraries and interactive development tools specifically for motor control. This tends to be easier for development kits that are known quantities.

For example, a typical application combines a graphical front end running on a PC that communicates with the microcontroller using its debug interface, which is often the serial port. The front end permits designers to change any of the run-time parameters to see how they affect the motor's operation. Such applications are especially effective for highlighting features like ramping, which tend to be more difficult to experiment with when simply using a run-time library.

Fluke addresses another aspect of debugging with its 1587/MDT Advanced Motor and Drive Troubleshooting Kit. The kit combines a Fluke 1587 Insulation Multimeter, a Fluke i400 Current Clamp, and a Fluke 9040 Rotary Field Indicator into a single unit. The Fluke 1587 Insulation Multimeter is a true-rms digital multimeter (DMM) and an insulation tester. The Fluke 9040 Rotary Field Indicator enables the user to easily check the rotation of three-phase motors.

Trying to wring out the most from a motor and microcontroller takes a good understanding of both along with the software. For most applications, though, the standard run-time libraries are often more than sufficient, allowing the average embedded developer to handle motor-control development without too much effort.

Intelligent motor control is becoming more critical as time goes on. Motor-control microcontrollers have stepped up to the plate, offering small size and low cost. With software support improving significantly, minimizing the necessary expertise in motor-control applications, the question is why not use one.