"Right the First Time", Feature Article, June 2008

right the first time

New tools can help mechanical engineers avoid common pitfalls in designing complex mechatronic motion control systems.

by Brian Mac Cleery
and Nipun Mathur

If you find yourself contemplating modern mechatronics systems and commenting to colleagues about how things seemed simpler back in the old days, then you are definitely not alone.

The relatively well-understood gears and cams that once served as the mechanical brains of motion control systems are quickly going the way of the cassette tape, replaced by digital technology. Mechanical systems are increasingly controlled by sophisticated electric motor drives that get their digital intelligence from software running on an embedded processor.

No longer the exclusive domain of mechanical engineers, getting elec-tromechanical designs right requires multidisciplinary teamwork and superb communication skills among team members. A decision like choosing the screw pitch for a lead screw actuator has a ripple effect through-out the design and can impact the performance of other systems in subtle ways.

New tools make it possible to simulate the behavior of complex electrical and mechanical systems before building a prototype.

To help facilitate a more integrated design process for electromechanical systems, software developers are partnering with electrical and control design companies to add motion simulation capabilities to CAD environments to create a more unified mechatronics workflow. One example is the partnership between National Instruments and SolidWorks, which are integrating motion programming software, normally used to run an embedded control system, directly into the CAD package for system simulation.

Integrating motion simulation with CAD simplifies design because the simulation uses information that already exists in the CAD model, such as assembly mates, couplings, and material mass properties. Adding a high-level function block language for programming the motion profiles provides easier access to control those assemblies.

In the past, trying to hand-calculate position versus time arrays for coordinated multi-axis motion trajectories like arc or contour moves was prohibitively difficult. Industrial motion software simplifies such simulations. It also yields more accurate results, because force and torque data are highly sensitive to the shape of the velocity profiles, which are typically trapezoidal or S-curve with defined constraints for maximum acceleration and jerk.

This concept is sometimes called "virtual machine prototyping." It brings together motion control software and simulation tools to create a virtual model of an electromechanical machine in operation. Virtual prototyping helps designers reduce risk by locating system-level problems, finding interdependencies, and evaluating performance trade-offs.

Complements of the management: In a mechatronics project, each decision has a ripple effect on other aspects of the design.

Simulations enable everyone to work on development before the first prototype is up and running. For example, engineers can use force and torque data from simulations for stress and strain analysis to validate whether mechanical components are stiff enough to handle the load during operation. They also can validate the entire operating cycle for the machine by driving the simulation with control system logic and timing. Or, they can calculate a realistic estimate for cycle time performance, which is typically the top performance indicator for a machine design, and compare force and torque data with the realistic limitations of transmission components and motors.

This information can help identify flaws and drive design iterations from within the CAD environment before a company builds a physical prototype. Are you violating vendor-specified limits of mechanical transmission components? Is the ratio between the load inertia and the motor shaft inertia within a recommended range? Can the motor produce the peak and RMS torque and velocity you require? Will the machine vibrate on its mounting brackets? Will the end effector vibrate at the end of the move operation? All of these questions can be answered with virtual machine prototyping.

Simulations also simplify evaluating engineering trade-offs between different conceptual designs. For example, would a SCARA robot be preferable to the 4-axis Cartesian gantry system? Given the complexity and interdependencies of mechatronics systems, relying on back-of-the-envelope calculations is a risky proposition. Simulations are faster and more accurate, and you can rerun them whenever you make design changes.

Picking Parts

Consider an analysis of the torque load for the bottom lead screw actuator in a 4-axis pick-and-place machine. If you violate the limits specified by the manufacturer, the mechanical transmission parts may not last for their rated life cycles.

Using simulation software, you can find the mass of all the components mounted on the lead screw and determine the resulting center of mass by creating a reference coordinate system located at the center of the lead screw table and calculating the mass properties with respect to that coordinate system. With this information, you can calculate the static torque on the lead screw due to gravity caused by the overhanging load.

Turn of the screw: 3-D software can verify that torque loads in this 4-axis pick-and-place machine do not exceed recommended limits.

In this case, let's say that the overhanging load consists of a rotary indexing table, two more lead screws to provide additional motion axes, and a pneumatic gripper arm. Next, run a motion simulation to take the machine through its pick-and-place operation fast enough to meet the customer's cycle time requirements. With a few more clicks, you can view the simulation results to obtain data for the dynamic torque produced by the motion profiles. Evaluating the dynamic torque induced by the motion is important because it tends to be much larger than the static torque load.

For example, say that the mass of the upper assembly is 2.25 kg and the center of mass is 38.9 mm away from the center of the lead screw table. With these two values, you can calculate the static torque exerted by gravity to be 0.86 N·m. The maximum allowed torque specified on your Thomson lead screw data sheet about this axis is 290 N·m. Next, run a motion simulation to determine the torque loading caused by the execution of the motion profiles, which yields a maximum dynamic torque value of 2.05 N·m. You see from your analysis that your machine is well under the maximum allowed torque limits for the slide.

You have examined this machine under maximum extension, where the maximum torque is applied, but with the gripper arm unloaded. You can quickly reproduce this same analysis for different conditions, such as having a loaded gripper or increasing the velocity and acceleration limits on the motion profiles.

The Right Motor

In principle, motor sizing is pretty simple: Just pick the best motor to match the speed, inertia, and torque requirements of your application. However, the U.S. Department of Energy estimates that about 80 percent of all motors in use in the United States are oversize. Although modest oversizing compensates for rising friction over the life of the machine, going too far decreases performance because of higher inertia, lower motor operating efficiency, and higher energy costs. This is even more important when the motor is part of the payload, as in many multi-axis motion control systems. Approximately 96 percent of the lifetime cost of a motor comes from the energy it consumes, rather than the cost of buying the motor itself. Choosing the best components for the design affects both the performance and environmental impact of the machine over its lifetime.

Several Web-based motor-sizing tools have been designed to help sort through the thousands of choices, and some include motor data from multiple vendors. However, these tools are only as good as the inputs you provide, and they typically are limited to single-axis, straight-line motion profile analysis.

You can augment the tools by using features built into your CAD package to make more accurate estimates of your speed, inertia, and torque requirements. A few mouse clicks in a typical 3-D CAD environment, for example, provide the rotational inertia of your payload with respect to the motor shaft. Typically, you do this by creating a reference coordinate system at the center of the motor coupling and measuring the rotational inertia with respect to that location. Then, to calculate the inertia reflected to the motor, multiply the rotational inertia by 1/N2 where N is the transmission ratio for the gear. If the ratio between the load inertia compared to the motor shaft inertia is greater than 6:1, consider switching to a lead screw with a different pitch to change the transmission ratio, substituting a different motor, or adding a gearbox.

Using realistic multi-axis motion profiles to drive your simulation can provide more accurate torque and velocity requirements, which depend on the acceleration characteristics of your motion profiles and the mass, friction, and gear ratio properties of the transmission. When evaluating simulation results, compare RMS values for torque and velocity with the rated performance curve for the motor, and compare peak values with the maximum velocity and intermittent torque limits.

For stepper motors, you typically want to derate the manufacturer's torque-velocity curve by as much as 50 percent to make sure your peak torque requirements don't cause a stall condition.

Component Stiffness

Analyzing stiffness and selecting the right material thicknesses to do the job might be considered the bread-and-butter stuff of mechanical engineering. However, when the prime mover is an electric motor under the control of a complicated power electronic chopper drive, governed by closed-loop electronic control systems with software-configurable PID control gains and real-time motion trajectory generators following trapezoidal velocity profiles—well, you can't blame an ME for asking a few questions.

Mechanical compliance issues that lead to unintended vibrations are one of the most common causes of problems in the motion control industry, and they are not easy to fix in controls software. Sometimes, you can avoid a complete redesign by stiffening a few components or adding a little extra dead time to the end of the move operation, but this is definitely not a situation most MEs want to discover when the system first comes on line.

Most of the time, a mechanical team considers compliance issues when it designs the assemblies, but incorrect assumptions about operational forces and torques may lead to problems. If accurate motion trajectories aren't considered until after the mechanical design is finished, any reasonable project manager cannot blame the mechanical team. But try explaining that to management when the fingers start pointing.

It's better to avoid this situation in the first place. So consider how, in mechatronics systems, compliance issues take two main forms: rotational compliance and linear compliance.

Evaluating trade-offs: New tools can simulate motion of this Cartesian gantry system in the CAD environment.

Rotational compliance is affected by the flexibility of mechanical transmission components, such as the connecting rods and couplings that connect motors to their rotating loads. Each rotating part acts like a spring with a particular stiffness, and the entire drivetrain acts like a series of springs connected in series. A phenomenon with symptoms that manifest themselves in a similar way is backlash, which is caused by the clearance between such mating components as gear teeth and appears primarily during direction changes. These clearance gaps mean that when a gear train is reversed, the driving gear must be turned a short distance before the assembly starts to rotate.

Both of these phenomena can wreak havoc on the closed-loop PID control system—the proportional-integral-derivative feedback devices used in industrial control—making it difficult or impossible to tune, or causing the system to literally hum during operation. If the system is detuned by reducing the PID gains to try to avoid the problem, you can kiss the cycle time performance good-bye. Because the best way to fix the problem is typically a mechanical change, your frustrated, bloodshot-eyed control engineer is eventually going to end up pointing his finger back at the ME.

Linear compliance problems are caused by the flexibility of mechanical assemblies, such as the gripper arm in a pick-and-place machine. The length of the moment arm, the weight of the payload, and the speed of the motion profiles all play a role. Because standard transmission components like lead screws tend to be quite stiff if you stay within the manufacturer's specifications, linear compliance issues typically manifest themselves in the custom-designed mechanical assemblies you attach to the actuators.

The most common control system workaround is to add extra dead time to the end of each move operation-to literally wait for the vibrations to settle. In some cases, running the motion profiles faster can actually reduce symptoms by shifting the excitation frequency of the motion trajectories above the natural frequency of the mechanical system.

You can think of the motion trajectories as frequency-domain stimulus signals for your mechanical assembly. Motion control systems often produce torques at fairly low frequencies that can induce vibrations during machine operation. Changing the velocity and acceleration values on the motion trajectories shifts these frequencies up or down.

Complex mechatronics enable the Nexans Spider remotely operated underwater vehicle to prepare the sea floor for pipelines running 3,000 feet under the ocean.

From a mechanical design perspective, you can use the force and torque data from motion simulations to help guide your stress/strain analysis. For example, you can plug in the peak torque values and see how much deflection to expect in rotating or linear parts using linear static analysis. This can also give you the equivalent spring constant with respect to an assembly mate.

Once you have the equivalent spring constant, you can add a spring to your motion simulation and see how much the assembly will oscillate during operation. Because you can reduce a series of springs connected in series down to an equivalent spring constant, you can approximate the overall compliance of a drivetrain as a single rotational spring located at the motor shaft coupling.

Although motion simulators are typically rigid body dynamics tools, adding an equivalent spring to the couplings can enable you to approximate compliance effects for simple transmission systems. For example, you can use this approach to examine how much the rotating transmission compliance will reduce the positioning accuracy of the end effecter or how much the machine will vibrate on its mounts. Dynamic analysis simulation tools take this a step further by evaluating the natural modes and frequencies of the entire machine assembly to locate the components that will exhibit the largest displacements and stresses based on input excitation torque.

They Don't Build 'Em...

The Boston-based Aberdeen Group recently conducted a survey of about 160 companies involved in the design of mechatronics systems to learn what separates the best performers from the rest of the pack. Primarily, the best performers implement processes to improve collaboration and communication among the members of the design team, such as using electromechanical simulation and design analysis tools. In fact, best-in-class companies were 5.3 times more likely than the industry average to use simulation tools.

According to the Aberdeen Group, it is also critical to work with the customer early in development and track requirements throughout the development cycle. Successful mechatronics design teams conduct frequent design reviews with the entire cross-functional team and actively try to identify design decisions that affect more than one group. Throughout development, the entire team should be notified if any changes are made to the motion control system because these changes are likely to have a ripple effect. Using these strategies, best-in-class companies were able to meet product launch dates 21 percent more often and budget targets 25 percent more frequently than the industry average for mechatronics projects.

One thing is certain in this fast-changing world: Things don't seem to be getting any simpler. As the old saying goes, you've got to have the right tool to do the job. The mechatronics-oriented design tools emerging today may give us a little help.

For More Information

Resource material on mechatronics and motion-control design is available online. These are a few places to look.

+ NI LabVIEW-SolidWorks mechatronics toolkit, alpha version download and Webcasts: http://www.ni.com/mechatronics/.

+ Mechanical linear actuator-sizing tool by Danaher Motion: http://www.linearmotioneering.com/.

+ Multivendor motor-sizing tool by Copperhill Technologies: http://www.copperhilltech.com/.

+ A Comprehensible Guide to Servo Motor-Sizing by Wilfried Voss: http://www.lulu.com/ content/1135964.

Brian MacCleery is the senior product manager for control and embedded design at National Instruments in Austin, Texas. Nipun Mathur is product manager for National Instruments' motion and control hardware products.