by Jason Goerges, Account Manager & Applications Engineer
Cameron Sheikoholeslami, Controls & Applications Engineer
ACS Motion Control, Inc., Eden Prairie, Minn.
Anyone familiar with the motion-control industry has heard the ambiguous term “high performance” describe many systems, but what does it mean to you?
High performance by itself is meaningless because it begs for a unit of measure — and it has none. With almost every product claiming high performance, how can a potential user differentiate between them? Some insight and knowledge of the key factors that determine motion-control system behavior shows you the right questions to ask that ultimately determine which motion-control system can meet your demands for high performance.
To understand how vendors define high-performance motion control in engineering terms, first divide the motion-control system components into two sections: the machine and the machine controller. The machine controller includes components that actively control the system. The machine itself consists of the components that are controlled or manipulated. That is, the motion controller is just one component of the machine controller. The motors, mechanics, and other machine components have a direct impact on machine performance, but to evaluate performance in terms of motion control, these components are considered “fixed.”
Key factors and their effects
When evaluating motion controls for a particular application, consider the following four key factors (I-IV). Vendors should explain and quantify the specifications that are relevant to their motion control system components. Compare these specifications among the various systems and it will become evident which is the “high-performance solution.”
I. Servo-control algorithms: Motion Controller and Drives
Motion controllers use measured variables for input to complex algorithms that precisely control actuator position, speed, acceleration, and payload. Drives and power supplies also use algorithms to control voltage and current.
Proportional–integral–derivative (PID) servo-system algorithms are the simplest, but more advanced algorithms often provide much better behavior such as field-oriented control and space-vector modulation. Furthermore, certain multiple-axis systems, such as gantries that consist of two motors and two feedback devices, can also benefit from multiple-input, multiple-output (MIMO) servo algorithms. Consider this:
Update rate: The servo loop (position loop, velocity loop, current loop) update rates for most controllers range from 100Hz – 20kHz. Generally, faster update rates correspond to higher servo speeds; however, rates above 20kHz give a marginal improvement. Also, if an algorithm runs a fast velocity loop (20kHz), but a relatively slow position loop (1kHz), it cannot filter high frequency signals and disturbances in the position loop. A servo algorithm with one high-frequency update rate for all loops indicates a high-performance design.
Advanced features versus servo update rates: Some vendors claim a high servo-update rate, but the rate drops significantly when the number of controlled axes increases. This limits the speed and accuracy. A motion controller employing a uniform and high update rate regardless of axis count favors a high-performance design.
Control: Most basic digital motion controllers employ PID, which is best suited for simple mechanical structures. For high mass/inertia systems, mismatched loads, and more complex structures such as robots, gantry and ultra high-resolution stages, a more complex algorithm is required. Additional filters (low-pass, notch, biquad, zero-phase), disturbance rejection, feedforward, and adaptive control algorithms are often required for high performance.
Current control: Servo algorithms control current in drives. Simple algorithms such as proportional-integral (PI) and trapezoidal commutation require little processing power. Higher performance drives employ sinusoidal commutation, field-oriented current control, and space-vector modulation for better response at high frequencies and additional bus voltage.
DSPs: Some digital signal processors include analog to digital converters (ADC) that the drive uses to measure the currents. They are cost effective, but relatively low quality solutions. The signal-to-noise ratio (SNR) and the real resolution are usually limited to 10 bits (even for a 12-bit DSP). It limits the position jitter and velocity smoothness. A high-performance drive controller has a 14-bit resolution ADC or more and a high SNR.
MIMO: To meet specs of the most highly coordinated motion systems, such as gantry tables and wedge designs, independent servo loops for the different axes are not sufficient. Individual network drive systems that use this approach are therefore limited, even when each drive has a ‘high-performance’ single axis servo algorithm. Instead, a multiple-axis integrated MIMO approach must be used to reach the highest speeds and lowest settling times.
II. Motion generation and machine process interaction: Motion Controller
The motion controller generates the position profile for the moving axes. Advanced controllers coordinate multiple axes and provide multiple motion modes including multi-axis vector moves, jerk control (3rd order profiles), higher order or sinusoidal profiles, on-the-fly motion command execution, interpolated motion, dynamic error mapping and compensation, non-Cartesian coordinates and general inverse kinematics, path blending and look-ahead to optimize machine throughput. Update rates and synchronization specs are also critical here. Consider these factors in determining performance:
Types of motion modes: Acceleration is built instantly in a second-order trapezoidal profile. In a higher order profile, acceleration is built-up gradually in a controlled manner. Most rigid bodies do not respond well to a discontinuity in acceleration because it produces high-frequency vibrations. Obviously, such vibrations affect settling time and throughput, and may damage delicate sensors and mechanical components. A motion controller designed for high-performance motion should let the user specify distance, velocity, acceleration/deceleration, and acceleration/deceleration buildup or jerk. This is commonly known as 3rd order profiles, which lets the stage achieve higher velocities and shorter move times.
Second-order profiles: Some controllers use a second-order trapezoidal profile and filter its output with a low-pass filer. This smoothes the profile, but the motion time is not predictable and difficult to calculate. Also, it limits the ability to activate processes that are related to the motion profile. It is not desirable for high-performance applications.
Linear, circular, helical, and other interpolation modes: These modes may be necessary as well as special profiles such as Input Shaping, and sinusoidal (or minimum energy) profiles, which may greatly improve overall behavior by reducing resonances of certain structures. High-performance controllers should have such features.
Methods to synchronize machine process and motion: The controller should support high-speed I/O to synchronize machine events to motion. Features including registration inputs and position/time compare outputs – position event generation (PEG) or position-synchronized output (PSO) – may be useful for coordinating machine processes. PEG outputs can accurately trigger a laser or camera, and registration inputs can be used in scanning applications where input position information augments moves.
III. Power Conversion and ADC Hardware: Drives and Power Supply
Amplifiers (drives) accept low-power input signals (usually proportional to desired motor current or acceleration), amplify them, and drive the motors. They create time-varying voltages across motor phase terminals to produce motion. Ideally, this process is simple and does not degrade performance. In reality, the opposite is true, and advanced techniques should be used to minimize degradation effects of this process. Look to these factors for determining high performance:
Drive control structure and hardware: Does the drive support digital current loops, field oriented control, sinusoidal commutation, and space-vector modulation? See the discussion on the first key factor for more information.
Digital current feedback data measured with adequate resolution and quality: Analog to digital (ADC) converter chips measure feedback current in the motor phases. These chips have specific resolution and signal-to-noise ratios. Low-performance drives use an ADC integral to the servo DSP chip to save cost. Such ADCs are effectively limited to 12 bits resolution. Furthermore, the SNR limits the actual resolution to 10 bits. High-performance drives use 14-bit or higher ADC chips and have SNR ratios of 60dB or more. A low-resolution device introduces a quantization effect in the controlled current and acceleration. It limits the bandwidth, the minimum jitter, and velocity smoothness that can be achieved.
Advanced or flexible drive technologies: The most advanced drives have algorithms that let them drive various motor topologies such as brush, brushless, stepper, and induction. This flexibility lets you optimize hardware without sacrificing design. Also, linear drives are needed occasionally where PWM jitter or noise cannot be tolerated. A high-performance control vendor should provide these options.
Power supply hardware—protection circuits and fault-handling capabilities: Drives designed for robust and reliable operation have built-in circuitry to monitor excessively high or low current and voltage. If these features are not part of a drive/power supply, it is not designed for robust, high-performance operation.
Regulating bus voltage: Amplifier input voltage variations that affect the bus voltage is one of the main sources of servo error. High-performance power supplies should provide methods for controlling the voltage.
ADCs and DACs: When data are converted from analog to digital or vice versa, quantization error and noise can corrupt it. This happens when: motion controllers send command signals to a drive through analog torque commands, drives read analog commands, controllers sample analog position feedback signals for SinCos encoders, and drives sample current to perform closed-loop current control. Resolution, SNR, and THD (total harmonic distortion) are some of the properties of these components that affect the system’s overall performance.
Integrating control components: When different components of a machine controller are physically separated, communications between them impede performance. Standard protocols such as EtherCat, CANOpen, and SERCOS that connect each component can make a clean control solution for connectivity and user friendliness, but this method does not allow for the most advanced servo features like MIMO control.
In addition, this approach requires intelligence at each control system component. A better approach for high-performance multi-axis control integrates all machine-control components, that is, controller, drives, power supply, and PLC all into one product. This allows for MIMO control, advanced servo, and drive control techniques of multiple axes from a single controller, and a clean solution requiring virtually no machine control component interfacing from the user. An example of this is in the ACS Motion Control MC4U, a complete integrated package with multi-processor computer architecture designed to achieve the highest performance. In addition, this approach saves cost by offloading all intelligence to the motion controller instead of requiring it at each drive node.
Encoders and options: A high-performance controller supports a variety of feedback devices: digital (AqB), SinCos, absolute (EnDat, SmartABS, BISS), laser interferometers, resolvers, and so forth. For SinCos encoders (used on most high precision linear stages), the quality of the feedback signal depends on signal amplitude, phase, and dc offset. High-performance controllers provide a method to identify and correct these types of error, which is critical for achieving excellent constant velocity performance. Furthermore, high-resolution stages generate high-frequency encoder signals. A high-performance controller should be able to process high SinCos input frequencies (such as 2.5MHz) at maximum speed. Also, controllers designed for high performance provide onboard SinCos interpolation. Higher interpolation factors mean higher resolutions. Some controllers can go to a multiplication factor of 65,536. Lower performance controllers will have lower factors, and sometimes the factor must be reduced as velocity increases. Such designs are not geared for high performance.
IV. Tools for Motion System Analysis and Design: Motion Controller / PLC and Drive
The software package must complement the motion controller that lets the user setup and tune the servo quickly and accurately. A tuning utility can range from a simple step response tool, to auto-tuning features, to a full electromechanical measurement and design suite.
Simple setup tools: How time consuming is the effort to get the system moving from when the components are first connected? Not everyone is an expert in control systems, and it should not be a requirement to get a system moving. Though an engineer with more experience with the controller will likely get it to achieve a higher performance, even the most novice engineer should be able to get the system moving. A high-performance controller should provide easy-to-follow guidelines for setup from start to finish. After each step is setup, there should be no need to return and adjust parameters later. Each step should also have a real-world meaning so that users cannot become confused after they learn the system and the information the controller needs. These steps should provide some intuitive terminology and processes that relate the setup process to the physical machine.
Quantify system performance: A high-performance controller should provide objective measurements and be easy for the user to measure system move and settle times. The constant velocity error should be easy to quantify. Frequently, a “soft oscilloscope” can do it as part of software tools. When the controller has an adequate update rate, the scope can objectively measure most system performance metrics.
Advanced tuning tools to optimize performance: What features are provided to measure system stability? One of the biggest mistakes that engineers make when setting up and tuning a motion system is not checking for a marginally tuned system. A marginally tuned system means that the motion system does not have large enough stability margins and can become unstable and oscillate under minimal environmental changes in the machine. It is necessary to measure these margins to verify that they are large enough, and a high-performance controller must have this capability. The easiest method to look at the stability margins is with a Bode plot of the entire electromechanical system. From the Bode plot, a user can extract the gain margin and phase margin of the system. A robust, stable system should have at least 6 dB of gain margin and 30 degrees of phase margin.
Tools to implement advanced tuning filters: A high-performance motion controller should provide an interface for the user to measure the current system response (Bode plot) and be able to calculate how a filter affects that response. This capability properly implements notch filters to remove resonances from the system. For example, the ACS Motion Control SPiiPlus MMI software package provides a frequency response analyzer (FRF Analyzer) that lets users measure the frequency response of the system and implements filters in an off-line mode. This allows the user to visually see and objectively measure how placing a filter will affect the system performance and stability. This is a critical capability because the user is no longer relying on a guess-and-check method of how to implement filters and optimize performance. Instead, the user can apply knowledge of the servo system to converge on the optimal tuning parameters.
ACS Motion Control, Inc.