High-efficiency motor drives used to be items only found in factories. Led by strong impetus from the Energy Star program (which now includes a new “Super Star”category), energy-efficient motors drives are turing up in the high-volume consumer market.
Processor power can now be had for a reasonable cost, and advanced microcontroller-based systems now make it possible for consumer devices to use topologies seen only in high-end gear just two to three years ago. Moreover, designers are turning more towards energy efficient three-phase systems and permanent magnet and brushless dc motors whose cost has dropped substantially.
“The biggest thing we see is a significant amount of work on white goods, appliances, HVAC, even power tools,” said Tom Hopkins, director of applications engineering at STMicroelectronics, Schaumburg, Ill. “Energy ratings are driving this. You can really boost performance with a variable-speed or at least a two-speed compressor and a variable speed or two- or three-speed fan.” Indeed, UK-based IMS Research sees a doubling of variable-speed drives in home appliances by 2013. Currently, they go into just one of every eight items in this category.
“It's different today than only two to three years ago,” concurs Patrick Heath, Microchip Technology (Chandler, Ariz.) motor control marketing manager for the High Performance Microcontroller Div. “The cost of energy has gone up. Single-phase brush dc motors are cheap. A (more energy efficient) three-phase brushless (BLDC) or permanent- magnet type costs a lot more. But the cost of more efficient motors has come down significantly. The design, the materials, the type of magnets all reflect more efficient construction. There's more efficient control with better algorithms.”
The DoE, however, doesn't specify how to get the most out of a motor drive. Here, software has come to be an enabling tool for applications barely seen two years ago, such as energy-efficient washing machines. Designers are writing software for techniques such as field-oriented control (FOC), which is impractical to implement in hardware; and field-weakening (which lets motors reach peak speeds beyond their rated value). So-called “sensorless” drives — which do away with commutating Hall-Effect sensors and use the feedback from the motor windings to sense current — make for smoother and more reliable operation. “The price point for microcontrollers has fallen enough and the geometry of integrated circuits has shrunk to a point where you can start integrating advanced algorithms into much smaller applications,” says Chris Opoczynski, worldwide product marketing manger for motor drive and motion control at Texas Instruments Inc.
In addition, the world is moving towards digital power-factor-correction (PFC), perhaps the most critical hardware element for better overall system efficiency. It's a legislative issue, however. Most manufacturers are PFC-ready with appropriate products, but U.S. customers for those products appear to be lagging. By most accounts, it's the controller and the software that rule the day, over and above continuing efforts to maximize the performance of IC drivers, IGBTs, and power MOSFETs (likely the most expensive components in a motor drive).
“If you don't have a fairly sophisticated controller, you're missing information about your motor,” says Anders Frederiksen, motor control segment leader at Analog Devices Inc.(Norwood, Mass.), which specializes in signal-chain electronics. Its motor-integration system encompasses the BlackFin and Sharc processors, along with the VisualDSP++ development environment. “You can apply a switching sequence to control the power stage based on information from Hall-effect sensors. However, the motor dynamically changes with temperature and load variations. If you buy smart math to deploy real-time observers, you can better control the relationship between the current and voltage, which ensures you apply the right flux to the machine. So there is a tight link between the hardware and how efficiently you can control the motor. It's not just about the efficiency of IGBTs. It's a matter of improving the entire controller.”
Indeed, by many accounts, the power-stage driver may be the least of designers' worries. “Efficiency comes from all places - the PFC front end, the modulation scheme (largely determined by software), the quality of the output bridge, and the quality of input bridge rectifier,” says Joe Roy, field applications engineer at Fairchild Semiconductor (South Portland, Maine) which specializes in the power chain (Smart Power Modules) and generally applies an ASIC approach to motor control.
“But the control algorithm actually has far more to do with gains in efficiency by percentage than any other aspect of the system,” says Roy. “For example, you really wouldn't improve efficiency much on an induction motor by simply making the IGBT or MOSFET circuit very efficient. Simply substituting a brushless motor in that same application with a decent control algorithm could easily do better. The bulk of the losses are not in hardware (PFC is slightly a different story). The bulk of the losses arise from topology more than better IGBTs.”
Designers consider the remote debugging phase to be the most difficult part of a motor-drive design. Vendors that supply design tools for motor drives say they try to make their wares easy to use among engineers who may not be experts in motor operation. The vast majority of drive developers, they say, are engineers familiar with making power calculations but who know nothing about motors. And with the expansion of energy efficient motors into non-typical markets, there is a portion of developers who don't understand motors, motor drives, or power.
“In the past we might have seen a company buy a prepackaged black box drive and just use it,” says Chris Clearman, C2000 MCU marketing manager for Texas Instruments. “Now we're seeing guys wanting to do the drives themselves because they think they can. They think the microcontrollers are putting up enough collateral to make that happen. And then they want to go to the next step and create a high-performance, high-efficiency drive based on techniques that a dozen years ago were cutting edge and only a handful of people were working on. So we're definitely seeing a lot of this R&D technology move out to the masses.”
Consequently, TI and other chip makers in this field infuse their developer kits with a lot of extra bells and whistles designed for engineers who haven't had graduate classes in motor control theory. Most Web aids for today's typical “motor designer” are high-level tools, i.e., where computational analysis is transparent. They find that most engineers in this field aren't operating at the level of doing complete motor and drive simulations. But there are tools that make it possible to do simple analytical tasks.
“I don't know that (real-time simulation) doesn't have any meaning, it's probably a good idea in theory,” says Clearman. “The MathWorks does this sort of thing (simulation, graphical code development, block design, modeling, autocode generation). You can design a motor in their tool and then simulate different ways of controlling it. You can create your own PID (proportional-integral-differential controller) based on your simulation before you try it on your high-voltage, highly expensive setup. We find those tools to be very useful.”
More often, though, users lack the knowledge to do a simulation at this level. A simpler tool is Fairchild's SPM Loss Calculator, which has been an in-house product for years but was recently introduced on the Web. This tool, for component selection, helps the designer choose the appropriate power electronics to drive a given motor. It's designed for the bench engineer who can perform the typical array of power circuit calculations for power losses, thermal impedances, and so forth.
Beyond Web aids, motor-drive development remains largely a real-time, on-the-workbench endeavor. The most competent designers embrace multi-level approaches: from signal control through the power stage. A sampling of some of the more notable new products at the controller level include TI's Piccolo TMS320F2803x MCU, which features a 60-MHz Control Law Accelerator unique to F2803x-based devices. Its 32-bit floating-point accelerator operates independently of the C28x core for parallel execution of complex control algorithms. It can thus deliver up to five times the performance of existing devices for applications in motor control and digital power.
At the system level, TI touts two new development kits built around the company's Piccolo devices as the industry's first to enable optional digital power-factor correction and sensorless field-oriented control of up to two motors using a single MCU. The kits has a Piccolo controlCARD (with the DRV8402 power module, and 24-A current-delivery capability), IDE, software library, and detailed documentation on implementing FOC and PFC. The Motor Control and PFC Development Kit comes in two versions (either with one, or two, permanent magnet motors). STMicroelectronics' STM8-based motor control evaluation kit (STM32 processor) is also brand new. It's aimed at replacing STM7C-based systems for BLDC and induction motors.
Microchip offers include economical dsPIC-based development packages for control of three-phase BLDC motors for up to 48 V and 10 A (MCLV), and industrial motor applications (MCHV — motor control high voltage). In addition, the company's real-time data monitoring (RTDM) tool, in tandem with its data monitoring control interface (DMCI) within the company's MPLAB IDE, creates an alternative link between a host PC and the target device for debugging applications in real time. Using this set of software tools for getting data in and out of the target device lets developers run applications while providing the ability to tune variables and immediately see the effect without halting the application.
Things to come
Expect incremental improvements in algorithm development over the next two years, despite a well-established software bundle. Vendors also say they are devising better IC drivers and power MOSFETs for motor control, are improving modular power-density, and going to higher levels of integration (i.e., front-end components in a power module, the joining of motor control peripherals with the microcontroller). Most agree, however, there's not all that much to be gained in overall efficiency from better power-stage hardware. The design issue may well come down to how to lower system cost without losing efficiency. At the same time, economical integration of motor controls and ancillary circuitry will not be automatic.
“Integration doesn't necessarily mean better performance or improve bill-of-material costs,” cautions Joe Roy. “It is more important to do it well and economically rather than to maximize the integration, which brings with it less flexibility.”
Neither will traditional discrete components be dead in their tracks. “Hall-effect sensors for proper commutation aren't necessarily obsolete,” says Roy. “They're appropriate for some apps, not others,” he said, referring to the breadth of the motor market. “We can't assume the world will be sensorless in three years. In HVAC motors, for example, there's one new topology every eight to ten years. Even a perfect technology won't be as widely accepted in twenty to thirty years as you'd think it would be. Sensorless motors and switched reluctance motors won't replace everything. Because there are other factors involved.”
The micro: What's in a name?
The debate continues: whether to use a digital signal processor or a less expensive, general-purpose microprocessor or microcontroller. It depends on the application, but the differences between them are starting to blur. There's also a mix of semantics involved, various marketing strategies, and verbiage that users prefer over others.
“Twelve years ago we called some of these products DSPs because that was the heritage of the architecture,” said Chris Clearman, C2000 MCU marketing manager for Texas Instruments, whose processor chips run the gamut from basic to cutting-edge. “In the early 2000s, we started calling them digital signal controllers (DSCs), which was basically a hybrid between a controller and DSP. We now call them microcontrollers. One reason is that microcontroller architecture has changed. A 32-bit microcontroller has a dual-bus system like a DSP. A microcontroller today is much more like a DSP of 10 years ago.” Yet the confusion is destined to persist for some time; today TI's products focus around its so-called 32-bit high-performance DSP microcontrollers.
“The microcontrollers of today have a mathematical unit in hardware that allows single functionality (like a DSP),” agrees Anders Frederiksen, motor control segment leader at Analog Devices Inc. Frederiksen notes the term “DSP” was never one that resonated strongly with motor control practitioners. “In many cases, the more advanced microcontrollers are getting closer to what the DSP can deliver. They add floating point capability, multiple instructions, and better data engines that allow better data flow.”
More from the motor
Designers increasingly cite field-oriented control (FOC, i.e, vector control) as the best way to get a high degree of performance from motors in consumer goods. In addition, Microchip in particular touts its field-weakening algorithm as a way to effectively boost motor output without the need for higher motor ratings, a significant advantage when the motor must turn somewhat faster during peak cycle demands.
“If a donkey is moving in a circle and you hold a carrot in front in him, he's got incentive to move,” said Tom Hopkins (director of applications engineering at STMicroelectronics, Schaumberg, Ill.) about FOC. “In the same way, if you keep the flux in the motor stator 90° ahead of flux in the coil, you produce the most torque for a given amount of current. Vector control gives you smoother operation, less noise, constant torque (a traditional six-step drive will otherwise have 13% ripple in torque, ± 6%). And in some applications, it can be more economical.”
The alternative is to use somewhat more hardware, including three Hall-effect sensors in a six-position control system to determine rotor position for the brushless motor. On the other hand, so-called sensorless systems used with today's permanent magnet motors do away with Hall-Effect sensors and pick off the sinusoidal back-EMF voltage generated by the motor to determine the position of its rotor. All other factors being equal, the motor will be smaller and the system will run quieter.
Sensorless systems are also more reliable — you eliminate the cables or harnesses associated with the Hall-Effect sensors. The FOC algorithm requires considerable controller MIPS and some additional cost for analog/digital conversion and power-stage electronics. But FOC pays off in applications where the dynamic load and speed is constantly changing, as in washing machines.
Field-weakening, a somewhat recent addition to Microchip's libraries, lets a motor run to normally twice its rated speed beyond what its defining specifications would normally dictate. It's particularly useful for washer applications where, for example, the motor must run faster during the spin cycle to extract as much moisture from clothes as possible. Other applications include quick-freezing food, or for industrial applications where the system needs to recycle a conveyer belt to its original starting position at the end of the day. From the economics standpoint, applying the field-weakening algorithm spares designers the need to specify a higher-rated (and more costly) part.
In harmony with PFC
Power-factor correction (PFC) is designed to bring the current and voltage extracted from the mains in phase, thus cutting the reactive power or excess generating capacity utilities must otherwise provide to power supplies, motors, lighting products, and other inductive loads. The energy saved over the long haul ultimately cuts overall system costs as well.
Consider the input (ac/dc conversion stage) of a typical switching power supply, for example. The rectifier diodes in this stage conduct when the incoming ac waveform exceeds the nominal dc (filtered) bus voltage. That's a relatively narrow band of conduction, perhaps 15 to 20° of a rectified sinewave. The peak value of the current pulses is thus perhaps 5 to 10 times the equivalent average current value, and those pulses have high harmonic content. In practical terms, power-factor correction circuitry is designed to increase the conduction angle of the input rectifiers so the corrected input current is close to sinusoidal and in phase with the driving voltage.
“Say I'm going to supply 1 A continuously from a 10-V source (pulling 10 W), and say I'm putting that 1 A through a 1-Ω resistor,” explains Aengus Murray, director of iMOTION products for the Energy-Saving Product Group at International Rectifier (iMOTION first led the way in integrated motor-control). “So that's 1 A through a 1-Ω resistor, or 1 W burned up in that resistor. Now instead of supplying 1 A as a continuous dc current, say I have a switching converter that supplies 10 A for 10% of the time. That's also 1 W. But if I push that 10 A through the 1-Ω resistor, that's I2R, or 100 W × 10%, or 10 W burned up in that resistor. So from the utility point of view, when I'm driving current into a diode bridge and I only conduct for 15% of the sinewave, I'm squeezing all the power in a very narrow pulse. From my point of view you're burning a lot of power in my cable, absorbing power in a quite narrow period. So it's more efficient to deliver current as a sinewave.”
Harmonic standards thus measure how effectively a circuit pulls power from the grid, but that's aside from the actual harmonic content in the pulse (not a measure of the phase shift of current versus voltage). However, even after filtering, the pulse has the potential to feed forward in a motor system and affect motor operation. In that context, PFC has nothing to do with minimizing power-line EMI, explains Bilal Akin, C2000 MCU applications engineer at Texas Instruments. But PFC will help cut the cost of a motor drive.
“The good thing about PFC is that you can use it to regulate the dc bus,” says Akin. “That bus is connected to a motor drive, and during transients it can't tolerate much of a voltage drop. To keep the bus stable, you can put more capacitors in the system — the most expensive part of a motor drive. But using PFC, you won't need as many expensive capacitors.”
Analog Devices, Norwood, Mass., www.analog.com
Fairchild Semiconductor Corp., Sunnyvale, Calif., www.fairchildsemi.com
IMS (Intex Management Services Ltd.) Research, U.K., www.imsresearch.com
International Rectifier Corp., El Segundo, Calif., www.irf.com
Microchip Technologies, Chandler, Az., www.microchip.com/DSCMotor
STMicroelectronics, Switzerland, www.st.com
Texas Instruments Inc., Dallas, Tx., www.ti.com
Primer on field-oriented control, machinedesign.com/article/field-oriented-control-for-motors-1108