Sensorless AC Drives Fill Price/Performance Niche
KEY WORDSMotors, drives, & motion controlAC variable-speed drivesAC induction motorsAdaptive controlControversial right from their name, "sensorless" ac drives actually use current and voltage sensors to achieve motor control. Even beyond terminology, it's not obvious what sensorless ac drives can or can't do, and where they fit into the big picture.
Frank J. Bartos, CONTROL ENGINERING
Motors, drives, & motion control
AC variable-speed drives
AC induction motors
Controversial right from their name, 'sensorless' ac drives actually use current and voltage sensors to achieve motor control. Even beyond terminology, it's not obvious what sensorless ac drives can or can't do, and where they fit into the big picture.
Sensorless ac drives fill the middleground between high-performance closed-loop control and simpler open-loop (or V/Hz) control of ac induction motors. Pricing is commensurate with performance delivered. Most sensorless ac drives are based on one of various flux vector methods, hence loosely called sensorless vector (SV) control (see Basics sidebar).
Rockwell Automation's Force Technology uses a simplified terminal-voltage model for motor reference. The model is based on parameters insensitive to temperature changes during motor operation.
Why sensorless control?
In short, sensorless control achieves near-closed-loop performance, without a motor feedback device, and lower life-cycle cost. Virtually all drive manufacturers offer SV control models.
For Schneider Electric/Square D Co. (Raleigh, N.C.), the appeal of SV drives lies in enhanced performance obtainable 'with minimal additional cost.' Susan Bowler, Schneider Electric drives marketing manager, lists favorable features as low-speed operation, torque, responsiveness, and positioning capabilities. 'Because sensorless ac drives' performance is approaching that of servo drives, they are finding broader application in areas requiring more precise load positioning,' she says.
Schneider Electric's third-generation Altivar SV drives feature autotuning, which eliminates the need for users to enter motor data. With autotuning, dc pulses sent from the drive interrogate the motor for characteristics. The closest motor model in the system is chosen for control and supplied with current and voltage feedback. Thus, the drive continues to optimize motor performance as motor characteristics change over time, explains Ms. Bowler. 'Knowing the type of motor, the control algorithm calculates optimum voltage to apply to the motor to establish maximum torque generation for a commanded speed reference.' The motor model also accounts for thermal effects.
Baldor Electric Co. (Fort Smith, Ark.) sees several benefits in SV control versus V/Hz control. Among them are improved low-speed operation, better speed regulation with varying loads, and more starting torque for loads with high friction or inertia. Such attributes are moving SV technology toward 'the control of choice' for general-purpose, constant torque uses. 'For variable torque applications, V/Hz inverters still have the lion's share of the open-loop market,' says Baldor product manager Roddy Yates.
Baldor refers to its SV products by the more logical name of 'Encoderless Vector Controls.' Using only motor current feedback, these stator-flux-type drives estimate motor voltage from the PWM waveform, explains Mr. Yates. 'Control algorithms also estimate motor slip, rotor speed, and other operating parameters,' he adds.
SV drives offer real improvement over V/Hz drives, but that still doesn't make them capable of closed-loop performance. Mr. Yates offers some performance insights on various makes of SV drives from testing done at Baldor. (See Online Extra article at www.controleng.com.)
Russ Kerkman, engineering consultant at Rockwell Automation (Mequon, Wis.) notes that SV drive performance is proportional to the number of motor parameters measured. Simpler drives rely on phase-current measurements alone, while an electronic 'observer' makes voltage estimates. Bus voltage also is sensed for safety reasons.
At low speed
One significant issue with sensorless control is the low-speed limit, below which torque control becomes impractical. Dr. Kerkman cites typical useful speed ranges up to 60:1 (1 Hz min.); higher performance SV drives provide up to a 120:1 range from base speed (or around 14-15 rpm for a 60 Hz motor). 'It's possible to achieve control approaching zero speed, but not common. Most applications do not require it,' he says.
It gets complex and costly to go much below 0.5 Hz. Use of more sensors, careful matching of motor to the drive, tuning tasks, and longer commissioning time are involved. High-end SV drives add voltage sensing to current sensing to help with low-speed operation. Dr. Kerkman thinks that adding voltage sensors can account for 40-50% of the cost of a standard control board.
From the perspective of Siemens Energy & Automation (Alpharetta, Ga.), a majority of ac drive sales 'defaults' to SV control. Flux vector control (FVC) drives appeal to the few demanding applications needing tighter speed control and zero-speed torque control, explains Dave Kirkpatrick, AC Drives product manager, Solutions Business Unit of Siemens E&A. Complexity and cost of FVC drives (encoder, cabling, etc.) also lead to low sales volumes. Open-loop (V/Hz) drives account for more unit sales, but less than SV, he says.
Low-speed capability limits also depend on such factors as inertia and load changes. Operating at 1 Hz is 'fairly comfortable' for torque control; around 0.5 Hz is possible, depending on the application, but much below that is difficult for SV control, explains Mr. Kirkpatrick.
According to Dr.-Ing. Dieter Eckardt, manager of system technology, Standard Drives R&D, at Siemens Automation & Drives (Erlangen, Germany), the choices for low-speed control are 1) use a very complete motor model to help observe events and to lower the critical speed range, 2) switch to open-loop control for this portion of the control, or 3) extend parameter identification using 'test signals' that measure the field or rotating frequency via current signatures. This method is still in the research phase.
In ABB s Direct Torque Control method, parameters calculated in the motor model are continuously updated with operating values from current and voltage measurements.
Direct torque control
ABB's answer to sensorless vector control is what it calls Direct Torque Control (DTC). Introduced in 1995, DTC incorporates separate loops for speed and torque control (see diagram). 'DTC was developed from the start as a sensorless control architecture. It's a 'native' sensorless torque control method, not vector control,' says Kalyan Gokhale, manager, AC Drives R&D at ABB Automation New Berlin, Wis.).
DTC eliminates the pulse-width modulator stage found in a typical vector drive, because explicit current regulators or inverter voltage commands do not exist in the control. Instead, two (hysteretic) loops are closed using motor flux and torque estimates, explains Mr. Gokhale. Also, there is no intermediate torque- or flux-producing current loop.
'Output of the hysteretic control loops is the most optimum inverter state to achieve the desired flux and torque quickly and accurately,' he remarks. Control loops run every 25
Only motor nameplate data are needed as input to DTC. ABB premium ac drives employ the DTC technique exclusively. More SV product details at www.controleng.com.
Tougher than closed-loop
In some ways SV control is more difficult to implement than closed-loop control, say several manufacturers. Working without motor shaft information means the motor model must be extra accurate.
'Accuracy of the model is key to avoid high torque ripple caused by modification of the drive's firing control to errors in motor model prediction,' says Rodney A. Fickler, manager strategic marketing at GE Toshiba Automation Systems (Salem, Va.). With enhanced motor models that can adapt their estimates to changes in motor operating conditions, GE Toshiba reports torque ripple initially over 7% was cut to less than 2%, 'for a particular value of slip or load condition.'
Ability of the model to adapt is most important as the motor nears zero speed, where model errors increase significantly. Some specs of the drive can help gauge the motor model's adaptation accuracy. Mr. Fickler suggests to look for torque regulation accuracy typically in the 1-2% range and steady-state speed accuracy of 0.1% of rated speed.
However, even the latest modeling and adaptation methods can't match closed-loop performance. Mr. Fickler (and others) agree that encoder feedback is needed to achieve full torque and highly accurate torque regulation at or close to zero speed (typically meaning under 5% base speed) or to improve speed regulation to 0.01% rated speed.
'Inferring motor conditions rather than acting on them takes time,' remarks Mr. Fickler, who recommends examining speed of response when selecting an SV drive. He notes speed of response differences as high as 15:1 between 'sensored' and sensorless drives. 'This must be considered when dealing with high-performance applications,' he adds.
Motor modeling and adaptation advances have brought the further benefit of cost reduction to ac drives. See more about this and other cost aspects of SV drives in the Online Extra article www.controleng.com.
John A. Cline, manager of engineering at GE Fuji Drives USA (Salem, Va.) agrees that from a computational sense SV control is more complex than closed-loop FVC, though not as accurate.
Moreover, it's the quality of the math models and the motor parameters used for inputs that determines performance of vector control. Mr. Cline considers the autotune feature of SV drives as 'critical for arriving at accurate motor parameters.' He points out that motor parameters change dramatically during normal operation.
On-line tuning is integral to dynamic torque-vector control (DTVC) used in GE Fuji's AF-300 G11 drive. But autotuning goes well beyond easing drive startups. DTVC has a routine that follows the motor's actions, looking for parameter changes due to temperature or load changes. 'Motor parameters are continuously evaluated during operation and used in unique mathematical models to adjust both voltage and current for optimal motor control even at low speeds,' adds Mr. Cline.
At Mitsubishi Electric, Advanced Magnetic Flux Vector Control (AMFVC) represents the latest sensorless technology. It refines a prior method developed in 1993 to exploit 'increased torque and rotational stability, at low speeds, for ac induction motors without the need of an encoder.' AMFVC offers 'improved speed control and torque emulation,' according to Wayne Kantarek, marketing manager, AC Drives, at Mitsubishi Electric Automation (Vernon Hills, Ill.).
Mr. Kantarek agrees that benefits of SV control extend to most constant torque applications especially ones requiring high starting torque and low-speed smoothness. He thinks SV drive performance has reached the point where users can apply ac control's advantages over dc control.
Mitsubishi's SV control starts with a model of the motor's internal characteristics. 'The better the motor model accuracy, the better the motor's performance will be under inverter control,' states Mr. Kantarek. Autotuning allows the inverter to sample motor parameters every several milliseconds. Then, the inverter separates output current into excitation phase and torque producing phase components. Respective compensating voltages are factored in to keep the induction motor's primary flux at a stable value, he explains. Further math operations take place in the slip frequency calculation section of AMFVC.
Limits and possibilities
Yaskawa Electric looks at sensorless vector control as a valuable method for ac induction motor control, but not as capable as the closed-loop flux vector method. Mahesh M. Swamy, chief R&D engineer at Yaskawa Electric America (Waukegan, Ill.) describes some limitations of SV control from a practical context. Torque control at zero speed is not possible, but 100% torque is available in Yaskawa SV drives, starting at 1% rated speed. Speed control depends on the accuracy of calculated slip speed, which in turn depends on temperature-sensitive motor parameters. To improve results, the drive's slip calculator has a term to compensate for temperature, but its fixed value does not change with motor temperature, explains Dr. Swamy.
Operating a drive in torque control mode is generally unavailable, since SV control needs a speed reference to generate a torque reference. Processing power is another factor, since SV control is more calculation-intensive than closed-loop vector control. 'Choice of high-speed processors is important for sensorless vector control,' he says.
Another SV drive limitation is inability run multiple motors with independent speed control. 'Multiple motor applications, that share one variable-speed drive still have only one viable solution, the V/Hz inverter,' says Baldor's Mr. Yates.
'Increased performance at lower cost' is the view ahead for SV control at GE Fuji. Improved motor models will be the enablers, made possible by faster CPUs. However, Mr. Cline says, 'In spite of the improvements, a majority of ac drive applications will still be fully satisfied with simpler, V/Hz controls.'
Schneider Electric's Ms. Bowler envisions further advances in SV control, particularly in low-speed performance at higher power levels. Other improvements will come through faster microprocessors, DSPs, and integrated peripherals that enable faster reaction time to speed and load changes.
GE Toshiba's Mr. Fickler thinks performance differences between sensorless and closed-loop ac drives will diminish in the next few years. 'However it's not certain that sensorless ac drives will be the solution for all applications,' he concludes.
For more suppliers, go to www.controleng.com/buyersguide; for more info, use the following circle numbers, or go on line at www.controleng.com/freeinfo: ABB www.abb.com/motors&drives 246
Baldor Electric www.baldor.com 247
GE Fuji Drives www.geindustrial.com/
GE Toshiba Automation www.getoshiba.com 249
Mitsubishi Electric www.meau.com 250
Rockwell Automation www.rockwellautomation.com 251
Schneider Electric www.squared.com 252
Siemens E&A www.sea.siemens.com 253
Yaskawa Electric www.yaskawa.com 254
Online: Go to controleng.comfor more on: Sensorless ac drive developments
Products and companies
Control Techniques 255
Danfoss Drives 256
Eurotherm Drives 257
TB Woods 259
For further reading, see CE, Sept. 1996, pp 99-106.
'Sensorless' control basics
Despite the catchy moniker, sensorless ac drives rely on current and voltage sensor circuits to make up for the lack of motor-shaft feedback used in closed-loop control. Sophisticated motor models and math-intensive algorithms transform the sensor inputs via ever-faster processors to derive the flux and torque components needed for ac induction motor control.
Most sensorless ac drives are the sensorless vector (SV) type. Direct torque control (DTC) is another alternative (see DTC diagram).
SV drives come in different flavors. At the high end are products derived from closed-loop (field-oriented) flux vector control and based on inferred rotor flux information (see SFOC diagram). Less capable SV drives are based on stator flux inference and simpler control algorithms, explains Rockwell Automation engineering consultant Russ Kerkman.
Estimating slip frequency or the rotating frequency of the rotor is the difficulty for SV technology. One or the other quantity is essential for control. It's really a two-part problem, explains Dieter Eckardt, manager of system technology, Standard Drives R&D, at Siemens Automation & Drives in Germany. 'At high speed, the motor field can be calculated directly from the back electromotive force (emf). However, at low-speed it becomes difficult to calculate stator flux in the neighborhood of zero stator frequency. At zero frequency, stator flux is, theoretically, not observable,' he says.
Case Study Database
Get more exposure for your case study by uploading it to the Consulting-Specifying Engineer case study database, where end-users can identify relevant solutions and explore what the experts are doing to effectively implement a variety of technology and productivity related projects.
These case studies provide examples of how knowledgeable solution providers have used technology, processes and people to create effective and successful implementations in real-world situations. Case studies can be completed by filling out a simple online form where you can outline the project title, abstract, and full story in 1500 words or less; upload photos, videos and a logo.
Click here to visit the Case Study Database and upload your case study.