PMSM TECHNOLOGY IN HIGH PERFORMANCE VARIABLE SPEED APPLICATIONS

John Chandler Automotion Inc., an Infranor Inter AG Company Ann Arbor, MI

Abstract - Many variable speed applications found in industry today make use of AC induction motor and drive technology.  AC technology is both cost effective and highly commoditized.  However, when machine designers are called upon to maximize performance in variable speed applications, they often select the permanent magnet synchronous motor (PMSM) over AC induction alternatives.  In performance driven applications, the physical capabilities inherent to the PMSM motor can tip the scale in favor of this more expensive technology. 

    This article examines the typical selection criteria and justification for PMSM technology. Multiple application examples are discussed including; high-speed centrifuges, turbo compressors, instrument grade spindles, and handheld tools.  Basic PMSM motor and drive operation is also examined.  Three practical drive techniques are presented to improve motor efficiency, precision and bandwidth in high performance applications. The discussed techniques include; center aligned modulation, mapped encoder feedback correction, and adaptive torque feed forward.

 

I.                    Introduction

 

The dramatic growth of AC induction technology over the past two decades can be directly linked to the evolution of digital drive electronics.  The advanced control algorithms required to effectively operate an AC induction motor over a broad speed range became practical through the introduction of specialized digital signal processors and micro-controllers.  One result of this evolution is that some traditional variable speed technologies have been displaced in the market.  However, Permanent Magnet Synchronous Motor technology (PMSM) continues to prosper as a viable and competing alternative to AC induction in performance driven applications.

PMSM technology is commonly advertised as “Brushless DC” or “AC Servo”. It is interesting to note that the difference between these two names stems from a difference in the drive’s control technique, rather than some physical difference in the motor. As in the case of AC induction technology, PMSM technology has also benefited significantly from the evolution of drive electronics.

Today, PMSM technology is dominant in many positioning applications, and it holds a unique niche in high performance variable speed applications.  Although the cost of PMSM technology in variable speed applications is generally greater than AC Induction, so to are its inherent physical capabilities.  For this reason, PMSM technology is more highly valued.

 

II.                  Measurements of performance

 

Physical constraints found in high performance applications usually lead to the consideration of PMSM technology.  In these applications, an evaluation of performance requirements like cycle time, operating speeds, and accuracy, all contribute to a final technology selection.  Table I provides a list of different physical constraints that commonly become limiting factors in demanding applications.  Depending on the nature of the application, some of these constraints will dominate technology selection, while others become secondary.  Table I also groups each of these restrictions into one of three basic measurements of performance.

 

 

III.                When to use PMSM

 

To illustrate how a combination of physical constraints can lead to selection of PMSM technology, it is helpful to consider an existing application where this selection has already occurred, and is validated by market acceptance.  A good example of such an application is that of electric nut-running.

In the automated assembly and manufacturing industry, electric tools known as "nut-runners" have displaced the use of some pneumatic products.  This trend continues today.  In part, it is driven by a market pull to improve the cleanliness of assembly environments, to improve the accuracy of control, and to reduce the ancillary costs associated with pneumatics.  However, the introduction of electric tools presented some unique technical challenges.

Electric nut-running tools require small diameter, high speed motors.  A small diameter motor is required so that tools can be hand-held by an operator, or so that multiple tools can be mounted in tight groups to accommodate existing bolt patterns.  High output torque is also required in fastening tools.  Although the requirement for high torque is typically addressed with mechanical gearing, the combination of this gearing, and the required cycle times, together lead to the requirement for a high speed motor. 

In addition to high speed and small diameter, the ability to rapidly decelerate is required.  Rapid deceleration is needed to avoid transmission of the motor's kinetic energy into a joint as a nut is tightened.  If the motor's flywheel energy is not dissipated quickly, the tool can over shoot the desired level of target torque on some joints.  To avoid the problem of over shoot, nut-running tools found in industry today employ low inertia motors running at 30,000 rpm that can decelerate to zero speed in less that one revolution. 

Electric nut-running is one of the earliest industrial variable speed applications to adopt PMSM technology.  Design engineers in this industry considered the performance and cost of competing technologies, and they selected PMSM technology as the best fit.   The fundamental performance metrics of efficiency, precision and bandwidth played a primary role in this selection.

 

IV.                Additional applications for PMSM

 

While PMSM technology may appear to be an obvious choice for the application of electric nut-running, many other variable speed applications exist for this technology as well.  In each case, the selection of PMSM technology over a competing alternative hinges on the perceived value of performance, and on the physical requirements found in each application.

For example, in the medical industry, a high value is placed on the available space in laboratory and hospital room environments.  Here, the high power density of PMSM technology can be a primary deciding factor in technology selection.  In operating rooms, peristaltic pumps are used to circulate blood or meter drugs.  In addition to being small, pump motors must also operate smoothly, quietly and regulate speed stiffly against a highly variable, periodic load profile. 

In laboratory environments, PMSM technology is used to operate centrifuges at high speed.  Rapid acceleration and deceleration of a high inertial load is required to minimize valuable processing time.  For bench top centrifuge products, the selected motor must be small, and it must operate efficiently throughout a broad speed range.

Cutting spindles provide another good example.  Throughout industry, gantry style cutting machines are used to shape or cut various materials.  The performance of these machines can be measured as a ratio of cutting tolerance to cutting speed.  The rigidity, mass and subsequent cost of all machine framing elements needed to achieve a given level performance can be reduced when a high power density motor is selected for the cutting spindle.

More variable speed PMSM applications can be found in the Chemical and Semiconductor industries.  Unlike AC induction motors, PMSM motors can be designed to operate efficiently when a comparatively large air gap exists between the rotor and stator.  Some chemical pumps and vacuum feed-through products exploit this capability by inserting a non-magnetic material into the air gap to create a pressure, chemical or environmental seal.  This technique improves end product reliability by eliminating the need for dedicated magnetic couplings or fluidic seals.  This technique can also reduce overall end product size and cost.

Today, in the emerging fuel cell industry, high-speed electric motors are required to drive air delivery devices like roots blowers, screw compressors and turbines.  Although AC induction motors can operate at high speed, the dynamic performance, efficiency and power density of PMSM technology provide compelling reasons for its use in both stationary power generation and in-vehicle applications. 

 

V.                  Driving the success of PMSM

 

The broad success of AC induction drive technology has actually helped secure the future for PMSM technology in high performance applications.  PMSM drive manufactures are now able to leverage the large-scale component integration that occurred during the commercialization of AC induction products.  Integrated IGBT modules, isolation components, and specialized control processors, developed for high volume AC induction motor applications, can also be applied in PMSM drives.  This sharing of components with AC induction drives has helped eliminate the early price/performance discrepancy that limited the growth of PMSM technology in variable speed applications.  And yet, the physical performance of PMSM technology, stemming from the use of permanent magnets and now exploited with advanced digital control, continues to keep it well differentiated in the marketplace.

 

VI.                PMSM drive topology

 

The circuit topology of a typical PMSM drive is shown in Figure I.  On the left-hand side of this figure, an input diode bridge rectifies AC line voltage.  Capacitors are then used to filter this rectified voltage.  Together, these passive components form a simple AC to DC converter.

PMSM motors can produce energy that is returned from the load to the drive.  This process is known as regeneration.  During regeneration, this excess energy will accumulate as charge in the DC supply capacitors, and the input diode rectifiers will block current from being returned to the AC Line.  To dissipate energy during regeneration, most PMSM drives in the ¼ to 10 HP range use the shunt regulator circuit depicted in Figure I. 

On the right hand side of Figure I, the DC to AC conversion circuit shown is known as a "full 3 phase bridge".  One pair of transistors in this bridge circuit is dedicated to each motor phase.   The "high side" transistor is used to apply positive voltage to a motor phase.  In turn, the "low side" transistor is used to apply negative voltage to a motor phase.  By controlling which bridge transistors are "ON", and which are "OFF", current can be directed into, or out of, any combination of the three motor phases.

Digital PMSM drives directly control all of the transistors in Figure I using pulse width modulation (PWM) techniques.  In a digital drive, analog feedback measurements of voltage and current are directly converted to digital values.  All motor control functions required for operation are then processed in software.  These functions fundamentally include; Modulation, Field Orientation, Current and Velocity loop control.  The consolidation of these functions, made possible by specialized processors, has significantly reduced the cost of PMSM drive technology.  Digital control has also improved the reliability of drives by reducing the number of discrete components found in early designs.

 

VII.              Current and Modulation

 

Torque production in a PMSM motor is a function of current.  For this reason, it is first helpful to understand how phase current can be regulated by a digital drive using Pulse Width Modulation (PWM) control.  It is also useful to understand this process because PWM produces motor heating that will limit performance in demanding applications. 

Figure II shows two of the six bridge transistors from Figure I.  A high side bridge transistor is driven with PWM, or "ON-OFF", control.  Rm, Lm, and Vemf, represent the combined phase quantities in two of the three motor leads.  For the purpose of this discussion, consider the low side transistor as a switch that is always closed.

When PWM is applied to the high side transistor, two components of current flow result.  The first component, i1(t), occurs when the high side transistor is "ON".  Applying voltage through the high side transistor causes the level of current to increase.  The phase to phase inductance, represented by Lm, limits the rate of increasing current.  While current is increasing, energy is also being stored in the inductor, Lm.

When the high side transistor is turned "OFF", a second component of current, i2(t), then appears.  The accumulated energy in Lm will force a "fly-back voltage", and it will forward bias a low side diode in the output bridge.  This second current, i2(t), is known as a "free-wheeling" current.  The combination of Lm, and the level of fly-back voltage present, limits the rate of decreasing current when the high side transistor is "OFF". 

The control function in Figure II measures the free wheeling current and then adjusts the amount of PWM "ON" time required to maintain some average level of current flow.  Figure III shows the resulting current waveform when PWM control is used.  The average level of current flow can be considered as torque producing current.  The ripple current that is shown in Figure III produces heating in the motor's stator.  It also produces heating in the rotor, which will be discussed later in this article.

 

VIII.            The PMSM magnetic circuit

 

Before discussing the more complex relationship between torque production and current in a PMSM motor, it is necessary to consider the magnetic circuit of this machine.  Figure IV shows a cross section view of a 4 pole PMSM motor.  The PMSM stator is essentially the same as an AC induction stator.  Phase windings are connected in either a WYE or Delta fashion, and they are spatially distributed in lamination slots.  Laminations are used to reduce "reluctance" in the motor's magnetic circuit so that flux can be conducted between the rotor and stator. 

The air gap, located between the rotor and the stator, determines the level of reluctance present.  Reluctance in a magnetic circuit is like resistance in an electric circuit.  By extending lamination "teeth" close to the surface of the rotor, reluctance is minimized, allowing a high level of flux to couple from the permanent magnet rotor into the stator. A high level of flux coupling is required to produce strong motoring action.

 In the design of AC induction motors the size of this air gap becomes a limiting performance factor.  This is because AC Induction motors produce torque by inducing magnetic poles on the rotor.  If the air gap in an AC induction machine is large, the level of induction is limited, and strong motoring action is not possible.  High energy magnets are used in PMSM motors to create fixed poles on the rotor.  For this reason, PMSM motors can be designed to support relatively large air gaps. 

 

IX.                Field Oriented Control & Torque

As with a DC Brush motor, torque production in a PMSM motor is proportional to current.  This linear relationship is one of the fundamental reasons why PMSM technology is favored in high performance applications.  However, like an AC Induction motor, the PMSM motor is also a three-phase machine.  As such, torque production is a function of both phase current amplitude and stator geometry. 

Modern PMSM drives use Field Oriented Control (FOC) to separate the problem of torque production into two parts.  First, the position of the applied field in the stator must be aligned with the rotor.  Second, the intensity of the applied field must then be controlled to regulate torque.

For simplicity, Figure V shows a vector diagram for a 2 pole PMSM motor.  The spatially distributed stator windings can be thought of as a stationary 3 axis coordinate system.  In this U-V-W coordinate system, each motor phase represents one axis, or vector direction.  Each axis in the U-V-W system is separated by 120 electrical degrees.

In Figure V, a second, "D-Q", coordinate system is assigned to the rotor.  This D-Q system is orthogonal.  The "D" axis is directly aligned with the PM field, and the "Q" axis is at right angle to the PM field.  When the rotor is spinning, the D-Q coordinate system is moving with respect to the stator's stationary U-V-W system.  The electrical angle between these two coordinate systems is measured as "theta".

A signal flow diagram for field oriented control of a PMSM motor is shown in Figure VI.  Digital drives that implement FOC, measure the rotor's electrical angle, theta, and they also measure the level of current flowing in each motor phase, U-V-W.  Given these quantities, a vector transformation is then used to calculate the level of current flowing in the stator that exists along each axis of the rotor's D-Q coordinate system.  These transformed current measurements are referred to as id and iq.

 

 

Torque production in a PMSM motor is optimal when no vector component of current is present along the rotor's "D" axis.  For this reason, a current control loop is used in FOC to force the value of id to zero.  At first glance, the function of this control loop may not appear obvious.  However, the id control loop produces a vector component of the applied U-V-W voltage that compensates for the stator's reactance.  This allows the iq control loop to regulate the "torque producing DC current" along the rotor's "Q" axis independent of the motor's operating speed, or shaft angle.  Figure VII shows the relationship between the synthesized iq current and the actual current flowing in one phase of the stator.

 

 

X.                  Velocity Feedback & Control

 

Finally, to discuss digital drive techniques that improve the performance of PMSM motors, a brief discussion of velocity feedback and control is helpful.  Figure VIII shows the signal diagram for a typical velocity control loop.  In most brushless drives, velocity feedback, v(t), is calculated from a position feedback device.  Many different methods are possible for this conversion.  Two common methods are to either differentiate the position feedback, P(t), or to measure the time, t, that it takes to traverse the incremental distance, dP.  In either case, the measured velocity, v(t), is compared to a desired velocity, and the difference is taken as velocity error. 

A high order control filter, typically a PID type, is then used to force velocity error to zero.  When the output of this control filter is feed into the "iq" current loop, it becomes a torque command signal. 

Any control filter used to calculate the torque command signal will introduce a time delayed response as it attempts to eliminate velocity error.  This means that velocity error will never be zero when either the desired velocity, or the load torque, is constantly changing in time.  For this reason, a torque feedforward signal is commonly added to the output of the velocity control filter to improve tracking performance.

 

XI.                Improving PMSM performance

 

Drive selection must be carefully considered to fully capitalize on the choice of a PMSM motor.  For example, in high speed applications motors are normally designed with low inductance to maximize power delivery at speed.  They are also commonly designed to operate from a high voltage to obtain high operating speed.  To prevent excess motor heating in this case, PMSM drives must regulate average current into a low inductance load, but they must also minimize current ripple.

Applications that require a high level of precision are very dependent on the selected drive.  The level of signal quantization that occurs within digital drives directly affects performance in this case.  Numerical control variables must have sufficient resolution to operate the motor smoothly, quietly, and with high gain.  Feedback signals must be processed optimally to achieve high accuracy and repeatability.

Applications that require rapid acceleration and deceleration need high bandwidth control.   Drives must execute control loops at high frequency to provide an acceptable level of stability.  To control a highly dynamic or cyclic load, a PMSM drive may need a specialized control filter or feedforward technique. 

The ability to adapt digital drives to specific applications through software development has greatly expanded the potential market use for variable speed PMSM technology.  Digital PMSM drives today employ specialized control techniques to maximize the performance of this motor.  Three such specialized techniques are presented here for consideration.  These techniques are provided as example methods that can be used to improve the efficiency, precision and bandwidth of PMSM motors.

 

XII.              Center aligned modulation

 

PWM ripple current creates two primary components of motor heating.  PWM ripple current produces copper loss in the stator and it also induces magnetic loss in the rotor.  This second effect, rotor heating, can be a critical limiting factor in high speed, high power density, or vacuum applications.  PM rotors can be demagnetized in these applications when current ripple is not minimized.

The current regulation technique shown in Figure III is used in many simple switching power supplies.  This technique, where only one transistor is modulated, is not suitable for controlling motor current during regeneration.  Other PWM switching techniques that do control current during regeneration also produce higher levels of ripple current.

Figure IX shows a PWM technique that is known as Center Aligned Modulation.  For simplicity, just two motor phases are shown.  In practice, all three phases are modulated accordingly.  The high side bridge transistors are driven with the PWM signals shown, and low side bridge transistors are driven with the compliment of these signals.  Individual transistors are switched at the frequency of the modulator.  However, all PWM signals are also centered about the modulation signal. 

This centering effect doubles the frequency of motor applied voltage pulses.  Also, the free-wheeling current that is present, when no voltage is applied, decays slowly like the technique shown in Figure III.  The combined effect of increased frequency and slow decaying current can reduce ripple current by a factor of 4 when compared to alternative PWM techniques. 

 

XIII.            Mapped encoder feedback correction

 

Small diameter motors are used in the optics industries where extremely precise velocity control and constant angular accuracy are required.  The modified velocity control diagram shown in Figure X can be used to improve precision in laser scanning and similar optical applications.  This technique is known as mapped encoder feedback correction.

Encoder feedback accuracy directly limits performance in these applications.  As motor diameter is decreased, the effects of encoder gradient error (i.e. edge distance error) and motor shaft run-out are amplified.  However, if these inaccuracies are repeatable, then they can be measured, converted to a table of correction coefficients, and permanently stored in the drive.  The table of correction coefficients is generally referred to as a "map".

Each coefficient in the map corresponds to an individual encoder edge within one revolution.  As the motor rotates, individual coefficients are indexed using position feedback.  Coefficients are used to correct the measured speed at individual edges.  If the edge distance is known to be long, for example, the calculated speed will be multiplied by coefficient that is greater than 1.  Conversely, if edge distance is short, the speed will be multiplied by a coefficient less than 1. 

Some encoders tested by the author exhibited gradient error up to 6%, but were repeatable to 0.01%.  In this case a high degree of velocity feedback correction is possible.   Mapped encoder feedback correction has been used to improve control accuracy in some applications by more than one order of magnitude.

 

XIV.           Adaptive torque feed forward

 

As previously discussed, a torque feedforward signal can be added to the output of a velocity control filter to improve tracking performance.  There are many different possible methods for calculating this signal.  For example, in applications that require rapid acceleration, the "desired velocity" input signal can be differentiated to produce a torque feed forward signal.  This method works well when the load is mostly inertial, and rigidly coupled. 

In some applications, velocity is held relatively constant but the load torque is rapidly changing.   An example of this situation can be found in peristaltic pumps that are used in the medical industry.  Another example can be found in CAM driven mechanisms that are used in packaging equipment.  In either case, if the load profile of the mechanism is also cyclic, then the technique shown in Figure XI can be considered.

Figure XI shows an adaptive method for calculating torque feedforward.  In this method, a table is established in the drive's memory to record the output of the velocity control filter as the load mechanism is operated through one machine cycle.  The output of the control filter is recorded in this table as a function of position. 

During the first machine cycle, no feedforward is used to supplement the control filter and the normal level of velocity tracking error is present.  On the second cycle, the previously recorded torque function is then used as a feedforward signal to supplement the output of the control filter and tracking error is reduced.  The recorded feedforward function from the first cycle will help eliminate any systematic or repeating error that is present in the second cycle.  This technique is most effective when rapid load torque changes are present and repeatable as a function of position. 

This method becomes adaptive when the process of recoding and playback is repeated continuously from one cycle to the next.  Each time the motor travels through a new cycle, the control filter output is averaged with previously recorded torque function.  In this way, the feed forward function will slowly adapt to changing load conditions over time and temperature.

 

XV.             Conclusions

 

The benefits of PMSM technology should be considered in high performance variable speed applications.  Specialized digital drives can be used to enhance the performance of PMSM motors.  As digital drives evolve, the cost performance ratio of this technology continues to improve. 

 

Want to read more articles on servo drives? Click here.