Vector Control Competes with Electric Servos
This article (Part 1 of 2) compares technologies and looks at selection guidelines. Part 2, in September, will explore applications. Once upon a time…electric servo systems using permanent magnet (PM) synchronous motors represented the ultimate in high-dynamic position and torque control.
This article (Part 1 of 2) compares technologies and looks at selection guidelines. Part 2, in September, will explore applications.
Once upon a time.electric servo systems using permanent magnet (PM) synchronous motors represented the ultimate in high-dynamic position and torque control. For a long time, it was 'the only game in town.' Several technology advances of the past decade have improved the capabilities of closed-loop flux vector control-based on the popular asynchronous (ac induction) motor. Today, vector control (VC) is in a position to compete with traditional servo system in a number of situations.
While it's generally agreed that VC offers a viable alternative, debate remains about levels of performance. Much 'depends on the response and cycle time requirements of a given application,' says Roy M. Anderson, product line manager, high-performance ac & dc drives, Rockwell Automation/Reliance Electric (Cleveland, O). He notes that to do positioning, the more sophisticated vector control systems embed a motion controller into the drive. This trend is just getting started in industrial VC drives.
A VC drive can operate as an amplifier, following an external motion controller command as well, and would be configured as a pure torque regulator for best performance. 'As long as the cycle times are 30 per minute or less, vector control can make a very viable solution, while easily reaching response times of 1,500 rad/sec,' states Mr. Anderson, 'matching the response of low-end servos.'
This radar chart from Yaksawa Electric shows the traditional view of servo vs.
vector control capabilities. Size of bounded graph area indicates performance.
Mitsubishi Electric also uses a very similar graphical comparison of technologies.
The two methods are converging on the electronics side. Both drive types have the same power section. Both rely on closed-loop operation with feedback. (Sensorless vector control is not discussed here.)
'Basically, 'vector' and servo drives each use vector control,' says William Sinner, product line manager, high-performance ac drives, Rockwell Automation/Allen-Bradley (Mequon, Wis.). The main differences come from the applicable motors. In a servo solution, current from the drive goes only into torque production, simplifying the 'vector' control to one current component. 'Motor flux produced by the magnets is available instantly, making low-speed operation smoother and more responsive,' he says.
In a vector solution, drive current to the motor is divided between flux- and torque-producing components at all speeds. How well the drive discriminates these components determines a VC system's overall performance (see sidebar). And before torque can be produced, flux must first be established in the induction motor. 'The time delay can be appreciable in applications needing high cycle rates or extremely fast response,' adds Mr. Sinner.
Overshoot of commanded speed and settling time into position are typically greater for an ac induction motor than a brushless dc (bldc) servo motor. 'The servo motor's low inertia and permanent magnet field results in crisper, more responsive performance; however, that performance isn't needed in most industrial applications,' says Mr. Anderson.
Some companies specialize in both servo and vector control, making little distinction between them. 'On the first-pass, a high-performance drive of one type is the same as the other,' says William Erickson, staff engineer, Indramat Div., Mannesmann Rexroth (Hoffman Estates, Ill). 'I can pretty much give you a vector system that will perform equivalent to a PM brushless servo system in most power and torque ranges (see sidebar).' Subtle differences will determine the final choice.
The same algorithm can be used to control either system. Position loop closure is not inherent to vector control, but externally provided to the algorithm. Some people point to this as a differentiator. Yet, this is similar to commutation schemes for brushless motors, where rotor orientation must be known relative to the stator to control torque, explains Mr. Erickson.
Siemens AG Automation & Drives (Erlangen, Germany), sees the lines between servo and vector drive versions blurring from historic origins. Advances in power electronics, microelectronics, and fieldbus technology are among forces responsible, according to Dr. Albert Wick, manager of R&D. A growing number of suppliers now offer one drive able to run induction as well as synchronous motors. Also, additional functions-such as closed-loop position control-are being integrated into the drive package at reasonable cost.
Dr. Wick told Control Engineering , 'Today, several manufacturers offer ac drive converters with a high-quality vector control, where the control characteristics approach those of a servo drive.'
A servo by any other name.
Based on a classical definition of a servo-'any prime mover with a feedback device'- Baldor Electric Co.'s (Ft. Smith, Ark.) view is that vector drives have a place in positioning applications. For positioning, VC drives are used with programmable position controllers and standard
He lists some advantages of VC drives to include reliability, capability of rated torque to zero speed, precise speed control (at low end and up to several times base speed), constant power output above rated speed, and various programmable features.
However, Mr. Mazurkiewicz rates acceleration performance of dc brush- and brushless-type servos-with their faster positioning, smaller unit sizes, and lower motor inertias-ahead of vector control. (See technology comparison table.) Baldor's VC systems, through 60 hp (45 kW ) at 460 V ac (or 30 hp at 230 V), use induction motors specially designed for positioning. They are smaller in diameter and longer for lower inertia than standard induction motors.
Ron Koehler, director of product engineering at Yaskawa Electric America (Waukegan, Ill.) notes that earlier vector controls served high-precision speed regulation rather than precise position control. But servo applications comprise more than one performance level. Mr. Koehler sees an influx of VC drives into the 'low-performance' as opposed to the 'high-performance' servo market.
A separate controller (can be external) provides the position loop. Its torque reference scan rate must be matched to the drive's input, says Yaskawa. New generation vector controls have scan times &2 msec, with some below 1 msec. Servos, on the other hand, have analog input scan times of 250 msec.
'Position accuracy capabilities of a vector control drive, however, still lag behind servo control. Vector control can achieve an accuracy of about 0.1 mm compared to 0.001 mm with a servo controller,' adds Mr. Koehler.
Among differences noted between these drive methods at Mitsubishi Electric Automation (Vernon Hills, Ill.) are starting or low-speed short-duration torque, typically 300-400% for servos compared to 150% for vector control; speed control range (5,000:1 vs. 1,500:1); and speed bandwidth (250 Hz vs. 32 Hz). This reflects a traditional view of higher dynamic performance coming from a PM servo system. Stephen Racine, senior product manager, variable frequency drives, calls VC 'a low-cost, lower demand approach to servo controls. Either method can do the job in many cases, but vector control is preferred when motor and drive are separately purchased and for ease of maintenance.' Bruce Herman, servo product engineer, adds, 'PM servos offer adaptive tuning-a benefit that can maintain higher performance as system changes occur over time, such as revised production or output needs.'
'Vector control can replace servos in higher power applications,' comments Ray W. Rosati, chief engineer-Electronic Products at Warner Electric Motors & Controls Div. (Bristol, Conn.). 'Vector control offers performance similar to brushless servos, which have more limited available power ratings and production (hence are more costly).' Consider VC for torque requirements from 4 Nm through 100s of Nm, speeds to 1,800 rpm, and power ranges of 0.75 kW through 100s of kW.
However, VC carries such caveats as 'less demanding applications' and 'lower bandwidths.' In Warner's view, servo systems still dominate the high-acceleration, lower power, and higher speed (to 6,000 rpm plus) market. 'Brushless versions are usually the system of choice for extremely fast, precision motion,' adds Mr. Rosati.
'In reality, a vector control system is a closed-loop servo system. The only variance is in the performance comparison between them,' remarks Tom England, director of marketing for Kollmorgen (Radford, Va.).
Due to technology differences of the motors mentioned earlier, VC 'induction servos' are typically used in higher power applications. An induction motor system's typically lower cost compared to a PM solution-especially in the higher power ranges-provides a practical alternative with comparable performance.
At Kollmorgen, VC is applied in systems rated 12 kW and up that do not call for high dynamic re-sponse-'typically velocity bandwidths of less than 50 Hz.' Mr. England says, 'Vector servo is particularly effective in process control applications not requiring high bandwidth response.' He cites continuous flow control and some web handling applications as examples.
ABB Industrial Systems (New Berlin, Wis.) presents a special case because its servo alternative ac control method-DTC (direct torque control)-is a sensorless solution. Developed in Finland, a major advantage of DTC is its power range of 1.5-4,500 kW, compared to the typical 15 kW for PM brushless servos, explains Chuck Hollis, director of industrial sales, ABB Drives & Power Products.
'Applications with many fast reversals, changes in velocity, and very accurate positioning should be left to electric servos. With DTC (vector) drives, an induction motor can achieve performances never thought of in the past, and should be considered for many uses now served by servos,' says Mr. Hollis.
For example, the ACS600 DTC drive (2.2-315 kW) has an operating mode specifically for positioning. Integrated positioning software (with cycle time and refresh rate of 1 msec) eliminates the cost of an external motion controller. A simulated
Drive Technology Comparison at 1 hp (0.75 kW)
Higher speeds possible
7.6-in. OD (193 mm)
4-in. OD (102 mm)
3.5-in. sq (89 mm sq)
* -Base speed: motor can operate at higher speeds
Comparing the Motors
AC servo motor
30% less losses
0.8 power factor
Better efficiency, thermal losses occur mainly in stator
Thermal losses occur in rotor
No blower allowed
Blower is optional without blower
70% lower inertia
Torque >60 Nm available
Up to 1,000 Nm
Up to 30% lower
Low-cost gear motor combinations available
Emergency backup operation
Can be directly powered from 3-phase supply
Source: Atlas Copco Controls
How to select
General guidelines vary among manufacturers. At Rockwell/Reliance a servo drive is preferred for high cycle rates (>30/min); low-inertia loads (&10X motor's rotor inertia); and power ratings of 10 hp (7.5 kW) or less. The opposite attributes apply to vector drives: Low cycle rates, high inertia loads, and power ratings over 2 hp. Rockwell/A-B's Mr. Sinner suggests a checklist to match application performance requirements to the solution's feature set; for example:
Consider velocity bandwidth, process cycle time, and low-speed operation needs;
Look at signal throughput times (is the reference a network or an analog signal?); and
Is position control required? Is switching needed between speed and torque modes on the fly?
Indramat's Mr. Erickson takes a pragmatic approach to selection. If the PM servo system provides enough torque and speed, he goes that route, if not, the choice falls to a vector servo. 'That covers 90% of the cases; in the other 10% I look at inertia effects,' he says. Controllability problems (oscillations, instability) can arise if system load inertia dominates, while efficiency suffers if motor inertia dominates. A significant overlap exits and other details enter the final selection.
Use of very high-resolution feedback is one factor behind Indramat's enhanced vector servo systems. 'For servo performance, you need different feedback,' explains Mr. Erickson. His benchmark is >100,000 counts/rev so that resolution doesn't become the system's limiting factor. He puts the low-end of the servo range at 20-25k counts/rev, which satisfies a majority of lower power/torque applications where cost is critical. Typical industrial VC systems use coarser feedback.
How about the cost of highly precise feedback? For a a motor in the 5-10 kW range or a full system it's not significant (about 2% differential per axis), according to Indramat. However, it would likely be prohibitive to use with a small motor.
Load inertia, more importantly, inertia 'reflected' back to the motor is a further item to consider, suggests Baldor (and other manufacturers). Smaller load-to-motor inertia ratios are best when practical, but opinions differ on how close to match inertias (see CE , Sept. '97, p. 103). For more on inertia guidelines see an 'Online Extra' to this article at www.controleng.com .
Despite similarities, product implementation can often be distinct. Siemens' Dr. Wick offers some distinguishing characteristics for each method. Vector control is generally used for single-axis drives, available over a wide output range (well above 200 kW), and simple to operate. Motor and drive are often purchased from different manufacturers and the combined cost is significantly lower (40-70% of a servo drive).
For their part, servo controls come in both single- and multi-axis versions, have precise, high-dynamic performance (closed-loop torque control), and are generally available up to 160 kW output. Moreover, closed-loop positioning control and often the drive and numerical control function is integrated into one unit.
Yaskawa Electric's Mr. Koehler suggests calculation of the acceleration torque needed, as a starting point to decide if vector control suits a positioning task. The faster the acceleration rates, the higher the acceleration torque requirements. Yaskawa Sigma Series servos handle 300-400% torque; this compares to 150-200% with a VC drive.
Other basic selection factors include speed range, system inertia, speed and torque control bandwidth, analog input scan times, and encoder resolution. Typical resolution value cited by Yaskawa for vector control is 1,024 ppr with an encoder frequency response of 300 kHz; a comparable machine tool (servo) application would have 8,192 ppr/500 kHz.
Similarly, Warner Electric suggests checking if sufficient peak torque is available to accelerate/decelerate the system inertias (motor plus load) in the required time. 'A good starting point would be an induction motor with inertia similar to that of the load,' adds Mr. Rosati.
Bruce Butler, area sales manager for Control Techniques (U.S. headquarters, Chanhassen, Minn.) concurs about the blurring of VC and servo technologies. However, he concludes that application details must be reviewed case by case to arrive at the best solution.
His criteria for choosing a servo solution include high accel/decel times, cycle rates over 450/min, load/motor inertia ratios of 10:1 or higher, and if multiaxis linear or circular interpolation is desired.
Mr. Butler further explains that motor speed tends to favor a servo approach. Induction motors can work well above base speed, but their torque output drops rapidly at higher speeds. Servo motors have standard base speeds up to 6,000 rpm.
Cost, size perspectives
Ready availability and lower cost of induction motors favor VC systems. In larger sizes, 20-150 kW, they dominate all other motor types made for general applications. Also, VC drives from a growing list of producers are offered in sizes down to 0.37 kW, but this isn't their competitive range. Few PM servo machines are made above 50 kW. One new announcement is availability of Mitsubishi's PM brushless motors up to 55 kW at 460 V ac.
When both technologies compete for the same application, an economic break point exists that varies from one company to the next. For further cost perspectives and examples see our 'On-line Extra.'
Still, it's the PM servo system that enjoys wider acceptance today and is most numerous in smaller unit sizes. Looking ahead, however, both servo and vector technologies appear destined for expanding futures.and possibly to live happily ever after.
Comparing a servo motor and an induction motor of drastically different sizes is a way to show their real performance capabilities. At Indramat (Hoffman Estates, Ill.), staff engineer William Erickson ran 'small signal step response in velocity' comparisons of an 11-kW PM brushless servo motor and a 225-kW induction motor. Such tests are one measure of the systems' dynamics.
Results of the comparison show nearly equal dynamic response in reaching commanded speed-10 versus 12 msec response for servo and induction motors, respectively-(see diagram, points 1 and 2). For this demonstration, the settling time is not considered significant, and no special effort was made to optimize it.
'A few key things are needed to get this kind of performance,' says Mr. Erickson. To make it work, he integrates into the solution high-speed digital signal processors and high-resolution feedback (up to 4 million ppr, after quadrature detection). 'Technology makes little difference, performance within 2-5% is possible with either approach' he adds.
Atlas Copco Controls (Penthaz, Switzerland; Pittsburgh, Pa.) is closely involved with vector control and servo methods. Bernard Schneider, product manager SAM drives, has experimentally compared PM servo and induction motors for response to a torque step of 8 Nm. The motors had the same inertia, inertial load, resolver feedback, and other provisions. Results of the study showed both motors had 'full compensation' 4 msec after the torque step, that is, 'no more noticeable error.' Mr. Schneider concludes, 'It's possible to control an induction motor as accurately as an ac servo motor.provided the drive can control the flux-producing current component without any interference from the torque-producing current.'
Also see a table from Atlas Copco comparing selection criteria for these motor types.
Performance comparison of a bldc servo motor and an induction motor (3.3 Nm vs. 4.07 Nm stall torque), without connected loads at Rockwell Automation/ Reliance (Cleveland, O.), provides further insight into positioning applications. For a 100 msec velocity command, the servo motor showed superior speed and position response. At a 300 msec command-considered reasonable for many industrial uses by Rockwell-the induction motor had very linear accel/decel profile, minimal overshoot, and 0.1 rev following error.