# Servo Drives Design Guide

Integrating servo drives into motion systems forces engineers to make decisions that will ultimately affect the entire machine build. Specification choices include picking between analog and digital servo drives; torque-mode or linear servo operation; PCB-mount or fully enclosed standalone construction; and partially to fully distributed drive topologies.

January 8, 2020

In this exclusive Design Guide, the editors of Design World review the functions and variations of servo drives as well as their connectivity and configuration requirements.

## Basics of servo drives and terminology

Servo systems consist of four main components — motor, drive, controller, and feedback typically from an encoder. The controller and drive work together to determine what the motor needs to do (the controller) and send the necessary electrical energy to the motor to make it happen — the drive. The drive can control torque, velocity, or position ... although in servo systems, the most common parameter to be controlled is torque.

Note that servo drives are also sometimes called amplifiers because they take the control signal from the controller and amplify it to deliver a specific amount of voltage and current to the motor. These are not to be confused with integrated designs (which include the motor, feedback, controller, and drive) also sometimes called drives for their axis-driving function.

There are several types of servo drives. A common variation is the torque-mode amplifier. These convert the command signal from the controller into a specific amount of current to the motor. Because current is directly proportional to torque, the drive is controlling the amount of torque that the motor produces.

In contrast, with a linear drive (in which current is proportional to force) there's direct control of the motor's force output.

Remember that servo-motor torque is directly related to current:

To use an analogy …

• As the brains of the system, the job of the controller is to take information from the feedback device and send the appropriate voltage signals to the drive.
• The drive acts as the nervous system and sends the necessary amount of current to the motor. This process of reading and responding to feedback makes the system closed loop … which is the defining characteristic of a servo system.

One of the most important tools for sizing a servo motor is its torque-speed curve. But often, the torque-speed curve is specific to a certain motor-drive combination. This is because the continuous and peak torque capabilities of the motor are affected by the thermal properties of both the motor and of the drive. Inefficiencies in the motor cause it to produce heat, which can degrade bearing lubrication and insulation around the windings. Excessive heat — typically caused by running a motor above its peak torque — can demagnetize the motor's magnets.

Although the drive has no moving parts, heat can damage its power transistors.

Note that continuous torque is the amount of torque the motor can produce indefinitely. Peak, or intermittent, torque is the maximum amount of torque the motor can produce, but peak torque can only be sustained for a short amount of time before overheating occurs.

The term servo drives also sometimes refers to integrated motors: Servo drives were originally stand-alone components, separate from the motor and controller. But in the past 15 years or so, numerous motor manufacturers have developed integrated drive-controller offerings … as well as integrated motor-drive systems … and even complete motor-feedback-drive-controller systems.

Slightly complicating matters is that these integrated designs (which include the motor, feedback, controller, and drive) are sometimes simply called drives for their axis-driving function. They tend to reduce wiring, make sizing and selection easier, and save considerable space and setup time.

### SERVO-DRIVE TERMINOLOGY VARIATIONS

Let's start with the term servo drive. Here's what a drive does in a servo system: It basically takes an input signal from a controller and amplifies that signal which is then sent to the motor. And in that description is the key. A drive serves to amplify a signal. Amplification is needed because control signals are too low (in terms of current) to power the windings of a motor, which require higher current levels.

So functionally speaking, signal amplification is what is going on inside of a servo drive. Hence, the reason a drive is sometimes referred to as a servo amplifier.

Now consider the term inverter. To understand this name for drives, we need to look at the electronic functions inside of a drive.

An electronic inverter converts dc power to ac. Drives contain inverters to generate the ac signals needed to drive a motor. So labeling something a servo inverter really only refers to one of the electronic systems in a drive … even though engineers may use it interchangeably with the word "drive" to refer to the same thing.

That brings us to the term servo controller. This name is perhaps the most problematic. That's because traditionally a controller is where the control signals for the motor originate.

This block diagram of a typical servo control system shows the traditional relationships between the drive, controller, and motor. Strictly speaking, a servo controller does not supply the currents for the motor. That's what a drive does.

The controller generates the control signals, which are then amplified by the drive and sent to the motor.

Even so, contemporary designs continue to integrate these older stand-alone components and functions into one unit.

So a drive of today — in addition to performing standard drive functions — may also be generating the control signals. Or a controller may have drive functionality within it. Either way, knowing what a drive functionally does (as well as having access to product specs) can help determine what is being referred to when any of these terms is used.

### DISTRIBUTED ARCHITECTURE FOR SERVO DRIVES

Traditional servo-system architecture consists of a power supply, a motion controller, and servo drives all housed in one location — typically a control cabinet located away from the machine. Then each motor connects to the control cabinet by two cables … one for power and one for feedback. This centralized architecture results in significant time and cost for routing, managing and connecting all of the cables. It also requires a larger control cabinet to house the multiple components and dictates the need for forced cooling (air conditioning) in the cabinet due to the heat generated by the numerous electronic devices inside.

With intelligent servo drives, however, it's possible to move the drives out of the control cabinet and closer to their motors. This is called a distributed architecture — sometimes also called a distributed control system or DCS. Typically, in a distributed architecture, more intelligence resides in the servo drive — up to and in some cases including the motion-trajectory calculations.

If the axes must be tightly coordinated, as is the case with interpolated motion, the motion trajectory is calculated by the controller. If the motions of the axes are independent, the trajectory calculations can be carried out within each individual drive.

Communication between the drive and the control is facilitated by a network bus such as DeviceNet, some form of Ethernet, or SERCOS.

The three primary drawbacks of centralized architecture mentioned above — extensive cable management, cabinet size, and cooling requirements — are mitigated with distributed architecture. With the drives located close to the motors and a central power module supplying all components, the amount of cabling required is significantly reduced. Moving the drives out of the control cabinet also trims the necessary cabinet size and at the same time, lessens the need for cooling.

Another benefit of the distributed architecture is higher reliability because fewer cables means fewer connection points … and shorter cables reduce the opportunity for electrical interference or noise.

While less cabling and the resulting cost and reliability improvements are attractive reasons to use distributed architecture, it's not always feasible to locate the drives near the motors. This could be due to environmental reasons, such as excessive heat near the machine or process, or because of physical space restrictions in the machine design.

Distributed control is also achieved by using integrated motor-drive systems, which further reduce wiring by doing away with the need for motor-drive cables.

Integrated motor-drives are the next step in component integration but may offer fewer options (for feedback or I/O, for example) than traditional systems. Integrated motor-drives are also sometimes limited in their power output due to the heat dissipation required to protect the drive.

Even so, for designers and machine builders looking to simplify machine layout, integrated motor-drive systems offer a compact modular solution for DCS arrangements.

## Selecting a servo drive

Sizing a motor for a servo application requires evaluating the move profile and torque requirements to determine the mechanical demands of the system, such as maximum velocity and acceleration, RMS and peak torque values, and load-to-motor inertia match. Once the motor is chosen, the next step is to select a drive.

On the surface, selecting servo drives may seem to be a matter of simply matching drive voltage and current output to the motor's requirements. But there are many factors that need to be considered to ensure the drive operates satisfactorily within the complete servo system. While some applications may require more specialized functions from the drive, here are nine factors that guide the selection of a servo drive for most applications.

Motor types commonly paired with servo drives: A servo drive can be used with any motor that operates in a closed-loop system — including stepper, induction, and asynchronous — but the two most common types of motors that are paired with servo drives are brushless DC motors and synchronous AC motors. Of these, synchronous ac motors are more common in motion control applications.

Commutation prompted by the servo drive: The type of commutation required by the drive depends on the type of motor being driven and the application's sensitivity to torque ripple. Brushless DC motors can operate with either trapezoidal or sinusoidal commutation, while AC synchronous motors use sinusoidal commutation. Trapezoidal commutation (also called sixstep commutation) is the simpler method, using three Hall sensors to determine the commutation sequence. But it produces high torque ripple.

Sinusoidal commutation virtually eliminates torque ripple and provides more precise control by continuously varying the current supplied to the motor windings. However, it requires a high-resolution feedback device to track rotor position.

Feedback: Servo systems require feedback in order to provide closed-loop operation and detect and correct any errors in the motor's actual position, torque, or velocity. This feedback can be provided by Hall sensors, resolvers, or encoders, although most high-end systems use a resolver or an encoder. Regardless of the feedback mechanism, the drive must be compatible with its signal to process it and communicate it to the controller.

Voltage and current: The most basic requirement of the motor-drive relationship is that the power from the drive — maximum voltage, continuous current, and peak current — must be sufficient to produce the mechanical output required by the motor — torque, speed, and position. Because the operation of the motor and the drive are co-dependent, manufacturers provide torque-speed curves that define the performance of specific motor-drive combinations.

Operating mode: Servo control loops within the drive are used for controlling torque, velocity, or position (or a combination of the three). While all servo drives incorporate a torque control loop and a velocity control loop, only digital drives can provide position control.

Analog or digital: Traditional servo drives were analog and converted ±10-Volt signals from the controller to current commands for the motor to control torque or velocity. In order to tune an analog drive, gain values and other parameters are set via potentiometers. With newer digital drives, the command can be executed by either digital or analog inputs, and tuning is done via software.

Along with torque, velocity, and position loops, digital drives can also manage higher-level functionality, such as path generation. Digital drives are also capable of monitoring internal functions of the drive (such as following error) and providing more detailed fault diagnostics.

Communication: In order for the drive and controller to "talk" to each other, they must have a common language. To accomplish this, servo drives are offered with a variety of communication protocols, from basic serial links, such as RS485, to more advanced protocols, such as traditional fieldbus networks (DeviceNet, for example) or Ethernet protocols. The choice of communication protocol will largely depend on the required communication speed, whether there is a master-slave arrangement between multiple drives, and in some cases, the type of controller being used.

Integrated safety: Functional safety is now mandated on many types of machines and is governed by the EN/IEC 62061 and EN/ ISO 13849-1 safety standards. To comply with these standards, safety functions, such as Safe Torque Off (STO) and Safe Stop 1 (SS1), are integrated into many higher-level drives.

Servo-drive mounting type: For industrial applications, servo drives are for the most part either panel-mounted or PCB-mounted. Panel-mounted drives often install into a control cabinet where they are protected from environmental hazards and can be externally cooled.

In contrast, PCB-mounted drives (also called plug-in drives) provide a compact solution for direct integration onto a PCB. These plug-in drives are commonly used in high-volume OEM applications.

## Where analog servo drives are used

The purpose of a servo drive is to convert low-power signals from the controller to high-power signals to the motor, instructing it to produce the desired torque or velocity. Servo drives (also referred to as servo amplifiers) can operate on either analog or digital input signals.

Analog servo drives receive ±10-V analog signals from the controller and convert these to current commands for the motor. The drive can control velocity or torque, and both the velocity and torque feedback loops are typically PI (proportional-integral) controllers.

A signal of +10-V indicates full velocity (or torque) in the forward direction, and a signal of -10 V indicates full velocity (or torque) in the reverse direction. A signal of 0 V indicates standstill, and other voltages indicate speeds (or torques) between full forward and full reverse, proportional to the level of the signal.

Unlike digital servo drives, analog types have no processing ability — that is, no electronics to make computations in the drive. This is actually a benefit in terms of servo response time, because the system doesn't spend time waiting for a digital processor to make the necessary computations and determine the response.

The tuning process for analog servo drives is also simple, with gain values and other parameters set via potentiometers.

With analog servo drives, high gains can be set, which makes for very stiff servo systems. This means that a small velocity or torque error will produce a large error signal. The result is very accurate control, even when there are significant changes in the load on the motor. For this reason, analog drives are commonly used when the motor's velocity or position needs to be precisely controlled. In fact, when position is the most important parameter, analog servo drives are often used in velocity mode with higher-level controllers having responsibility for the position control.

But the key benefits of analog servo drives are their low cost and setup that is more straightforward than that of digital versions.

## Strengths of digital servo drives

Digital servo drives operate over fieldbus networks that now dominate the market — and include a microprocessor to carry out computations — in turn to determine the output control signal based on a mathematical model of the system's behavior.

Most digital drives can accept feedback from tachometers, resolvers, encoders, and various types of switches or sensors. In addition to managing the torque, velocity, and position control loops, digital servo drives often include higher-level functionality … including operations such as path generation that were traditionally handled by the machine controller.

While analog servo drives are relatively inexpensive and simple to set up, there are benefits of using digital servo drives. First, a digital drive is tuned via software, rather than being tuned manually with potentiometers.

Most digital drives can also be auto-tuned or self-tuned, which is especially helpful when the load or inertia parameters are difficult to model or predict. This also simplifies the tuning process and provides a system that is much more responsive. Because all of the configuration and tuning setting are stored in the drive, it's also easier to replicate a specific setup across multiple drives.

Note that auto-tuning (self-tuning) is a process in which servo control-loop gains are set automatically. The drive excites an attached motor at various frequencies to sense the system's inertia and response, and then determines and sets the appropriate gains to ensure stability at all of the various frequencies.

With a digital servo drive, voltage pulses are sent to the motor at a much higher frequency — often 5 times or more — than with an analog drive. This allows the motor to respond to commands faster and provides smoother acceleration and deceleration. It also gives the servo system much higher holding torque.

Digital servo drives are versatile in their function. Most digital drives are capable of operating with an analog voltage signal, like an analog servo drive, and some can even accept step and direction signals to operate as a stepper drive. They can also be used when master and slave axes are required, with electronic gearing or an electronic cam between the axes.

## Difference between linear amplifier and PWM drive

Servo drives — also referred to as servo amplifiers — can be classified by the type of output stage they use:

• Pulse-width-modulated (PWM) amplifier output stage or
• Linear amplifier output stage.

Of the two types, PWM drives are more common in general motion control applications because they provide more power to the motor, have higher efficiency, and are generally smaller and lower cost. However, while they are sufficient for most industrial applications, PWM drives create both audible noise and electromagnetic interference (EMI). They also produce minute vibrations that can interfere with sub-micron level positioning.

So for applications that require extremely smooth motion, no audible noise, and little or no EMI, linear servo drives — commonly called linear amplifiers — are preferred over PWM drives.

Note that EMI noise can interfere with other electrical equipment in the vicinity of the drive and can be especially detrimental to the operation of sensors. Although filters can be used to remove the noise, they often reduce the speed of the sensing signals.

These differences in performance are a result of the operating principles that each drive employs.

A PWM drive delivers a specified amount of voltage to the motor by switching the voltage across the transistors on and off at a very high frequency — typically in the range of 20 kHz. When the voltage is switched on, the transistors are said to be saturated. This switching creates pulses, and the switching frequency controls the width of the pulses — hence the term pulse-width modulation. The ratio of on-time to off-time during switching determines the average voltage supplied to the motor.

In contrast, in a linear amplifier, the transistors are always on to some degree. That allows voltage to flow continuously through the transistors and to the motor, rather than being switched on and off. This makes matching the amplifier's output voltage to the motor and application requirements even more critical and, in some cases, more complex, than with PWM drives. Factors such as the motor's torque constant, back EMF, and resistance all play a role in determining the required output voltage from a linear amplifier.

Because there is always some voltage flowing through the transistors, linear amplifiers experience significant power dissipation, and in turn, relatively low efficiency — typically in the range of 50% … although it can be lower. (In contrast, PWM drives often have efficiencies of 90% or better.) Of course, the power is dissipated as heat, which means a linear amplifier needs large heat sinks to protect the transistors, increasing size and cost and often restricting their practical applications to those with power requirements of 100 W or less.

Despite these drawbacks, the lack of power switching gives linear amplifiers the benefit of very low audible noise and virtually no EMI. They also have a higher current loop bandwidth, and no dead band at the zero current crossing.

Drives that exhibit a dead band — as PWM drives do — cannot accurately provide the low levels of current required as the motor switches from positive to negative torque (or velocity) — at the zero crossing of the current waveform. This makes it difficult to control very small movements, which can cause dithering and long settling times. Unlike PWM drives, linear amplifiers don't experience a deadband, which allows them to provide much better velocity control and settling characteristics.

High current loop bandwidth, low EMI, and smooth motion characteristics make linear amplifiers the best choice for ultra-high positioning applications. They're often used in precision stages to drive linear motors or voice coil actuators in semiconductor, medical, and optical equipment.

One final note on terminology: The term linear amplifier has nothing to do with whether the motor being driven is a linear or rotary version. Linear amplifiers can be used on both linear and rotary motors — either brushed or brushless.

## Tuning servo drive systems

Tuning a servo system is a complex and iterative process. It typically requires tuning multiple control loops, each with its own gains (proportional, integral, and/or derivative) to be adjusted. In addition, tuning a servo drive usually requires adjustments to additional parameters including acceleration and velocity feed-forward gains and filters to reduce oscillations.

While manual tuning has been the predominant method for many years, most servo drives now incorporate functions that will automatically tune the system. Although in the beginning, auto-tuning functions were useful only when the load was rigidly coupled and the system dynamics were relatively simple, more complex algorithms and faster computing power have enabled the development of auto tuning functions that are sophisticated enough to address even the most complex systems, with minimal input or effort from the user.

Auto tuning is based on the same principles as manual tuning. That is, the performance of the motor is evaluated relative to a given command — and the servo drive automatically adjusts the gains until values are found that give the best performance. In most cases, the auto tuning process can also add filters to the control loop to suppress oscillations caused by resonance frequencies in the system.

As the motor and load are driven in the auto tuning process for this Yaskawa servo drive, the drive automatically determines the inertia of the system, measures oscillations, and sets, evaluates, and optimizes the control loop gains.

Adaptive tuning is similar to auto tuning but goes one step further and allows for a wide range of parameters that will provide stable control of the servo system. Adaptive tuning continuously monitors the system's performance and, if necessary, adjusts the control loop gains and filter parameters to compensate for unknown or changing load conditions during the system's operation. The key to adaptive tuning is that it runs continuously in the background of the control system to detect resonances by analyzing the frequency response of the torque loop.

One-parameter tuning typically refers to a tuning feature that can be used to fine tune the system's response after adaptive tuning has been configured.

The term one-parameter tuning comes from the fact that a single parameter is used to configure multiple system attributes — including gains, filters, friction compensation, and control of resonance.

In some programs, one-parameter tuning lets the design engineer manually adjust filters and gains as a way to more finely tailor machine response (after initial tuning) to various conditions and events. Several separate parameters can be optimized with a single value adjustment; separate modes can target responsiveness, stability, or things such as position-overshoot suppression. Here, some programs also allow setting of machine resonance filters to complement the specific machine assembly at hand. Then notch filters help minimize operation at or excitation of machine resonant frequencies — which in turn improve servo stability, responsiveness, settling time, and overall system efficiency.

## Bandwidth of a servo control loop

A servo drive can include any combination of three types of control loops—a position loop, a velocity loop, and a current loop. While each loop's purpose is to control a different aspect of the motor's performance, they are all characterized by a common parameter: bandwidth. The bandwidth, or response time, of the system is a measure of how fast it responds to the changing input command. In other words, the bandwidth of the control loop determines how quickly the servo system responds to changes in the parameter being controlled—position, velocity, or torque.

In servo drives, the bandwidth of a control loop is defined as the frequency at which the closed-loop amplitude response reaches -3 dB. At this point, the output gain (ratio of output to input) equals approximately 70.7% of its maximum, and the output power (power delivered to the load) equals 50% of the input power. (This article further explains the relationships between amplitude response, output gain value, and output versus input power.)

While higher bandwidth generally provides stiffer motor performance, decreases error, and improves transient response time, there are also drawbacks to high bandwidth in servo systems. Specifically, the higher the bandwidth, the higher the frequency at which the motor responds to disturbances, which typically requires higher accelerations and forces.

Power dissipation has a squared relationship to force, so any increase in bandwidth significantly increases power dissipation — in other words, heat. That in turn means a temperature rise in the motor. And because temperature is a limiting factor in motor operation, the motor characteristics may actually limit the allowable bandwidth of the servo drive.

Note: Other components in the system — including the resolution of the feedback device, the update rate of the drive, the motorload inertia ratio, and the rigidity of the motor-load coupling — also affect the maximum achievable bandwidth of the drive.

Servo drives often have a multi-loop structure with the current loop nested inside a velocity loop ... nested inside a position loop. Here, the response of the inner loop must be faster than the response of the outer loop, or the inner loop will have little or no effect. For servo control loops, the inner loop should have a bandwidth that is 5 to 10 times faster than the outer loop. In this scenario, the current loop bandwidth should be 5 to 10 times that of the velocity loop, and the velocity loop bandwidth should be 5 to 10 times that of the position loop.

## Pulse-duty versus continuous-duty servo drives

Servo systems are applied in a wide range of applications, from intermittent operations that require high torque output for quick acceleration and deceleration — such as pick-and-place from a conveyor — to processes that call for nearly uninterrupted operation with constant speed and torque requirements — such as printing, roll feeding, and labeling. Given these dissimilar application requirements and the effect that torque, current, and duty cycle have on servo system performance, it's easy to see why one type of servo drive or motor doesn't fit all applications.

A key factor in servo system performance is heating ... or more specifically, the ability of the motor and drive to dissipate heat to avoid damaging the motor insulation and drive electronics. There are a variety of causes for excessive heat generation, but aside from misapplied or poorly maintained products, operating at peak torque (and, thus, peak current) is one of the most significant factors.

Recall that the torque curve for a servo motor and drive combination includes two operating ranges: continuous torque and intermittent (peak) torque. The continuous torque range shows the torque the motor and drive can produce indefinitely at a given speed and is the basis for evaluating the RMS torque required by the application.

Peak torque is the maximum torque the motor and drive can produce at a given speed and requires maximum current from the drive. To avoid overheating, the peak torque value is allowable only for a short amount of time — typically a few hundred milliseconds.

So servo applications often fall into two categories:

• Servo applications that involve very rapid acceleration and deceleration … and therefore have high peak torque requirements
• Servo applications that require good continuous torque characteristics with moderate peak torque demands. The first type of application is referred to as pulse duty, and the second type is referred to as continuous duty. To address the disparity in performance requirements between these different applications, some manufacturers offer two variations of servo drives and motors: pulse duty versions and continuous duty versions.

Pulse-duty servo drives and motors are designed to perform well in applications that involve very rapid acceleration and deceleration rates, and in turn, have high peak torque requirements. Accordingly, pulse duty servo drives have a high current overload rating, while pulse duty motors have lower inertia than conventional designs, which reduces the amount of torque (and thus, current) required for demanding move profiles.

On the other hand, continuous-duty versions are designed to produce relatively higher torque at higher speeds on a continuous basis, with moderate peak torque capabilities. Thus, the torque curves for continuous duty drives and motors have larger continuous operation areas.

Continuous-duty motors and drives are suitable for printing applications needing near-constant operation at high speeds with moderate torque requirements.

Note: Root mean square torque (RMS torque) is a calculated value based on the torque required during each phase of the motion profile (acceleration, constant velocity, holding, deceleration, and so on) and the duration of each.

RMS torque is a time-weighted average — in other words, it is the amount of torque that (if produced continuously) would generate the same level of heating as the various torque levels and durations the motor experiences over its actual duty cycle.

The purpose of calculating RMS torque is to ensure that overheating doesn't occur during normal motor and drive operation.