Motion Control Resources
Smart Drives Take Control
by Kristin Lewotsky, Contributing Editor
Motion Control & Motor Association Posted 01/22/2008
A new generation of drives with embedded digital signal processors reduces cost, simplifies maintenance, and even performs control operations.
Intelligence. It’s the buzzword these days in motion control as sophisticated electronics elevate traditional components like motors and encoders into something more. Nowhere is that trend more apparent than in the new generation of intelligent drives, which can store data, reduce wiring and even replace motion control cards in some applications.
An intelligent drive includes a digital signal processor (DSP), often specialized for motion control, as well as peripherals and both volatile and non-volatile memory. Technosoft Inc. (Bevaix, Switzerland), for example, uses RAM to store user variables and for data logging and digital oscilloscope operations. The company’s drives use flash memory to store firmware, while user-developed programs are held in EEPROM. “This way, you can change the functionality of the drive without affecting the programs, and you can change the user programs without affecting functionality of the drive,” says John Chandler, Vice President of Sales, North America.
“If you pick the right DSP, you get all the peripherals that interface to encoders and I/O analog and digital communications, plus the drivers for the power stage and the power transistor,” says George Procter, Vice President of Copley Controls Corp. (Canton, Massachusetts). “A lot of the circuitry that used to be in a driver is now in firmware in the chip.” In fact, firmware is at the heart of the design process. “It’s really an order of magnitude in the scope of the development of software and firmware versus hardware now.”
The Benefits of Intelligence
Because of their design, digital drives can take input from a range of feedback devices, including incremental encoders, resolvers, sine/cosine analog and absolute encoders. “Among the things we found to be very nice with our intelligent drives is not only the ability to take that feedback in but also how we can interpret the data much more effectively,” says John McLaughlin, Manager, North America for Elmo Motion Control (Westford, Massachusetts).
One of the biggest benefits of the intelligent drive approach is the elimination of wiring. A centralized architecture requires each motor to be wired into the central motion controller, which means lots of cabling and a cabling cabinet to house it all. All that cabling is expensive, and from an engineering standpoint, designing a wiring harness can be a job in and of itself. Cabling also represents the most common point of failure in motion control systems.
By minimizing wiring, intelligent drives not only reduce cost but lower or eliminate the need for cabling cabinets, which minimizes the footprint and complexity of the overall system. According to Proctor, the distributed architecture enabled by intelligent drives can save 35% to 55% over a centralized system.
Though it’s not a requirement, networking digital drives maximizes their usefulness, bringing them one of the classic benefits of intelligent components -- simplified maintenance. “In the old days with analog drives you relied on a limited number of diagnostics,” says Proctor. “For example, you might have an output pin assigned to a fault line. Great, now you know there's a fault on the drive. When you have [intelligent drives and] a network connection, you can find out exactly what that fault was, you can get the error log from the drive and find out what occurred and when. It’s a whole different ballgame when you can talk to something with intelligence.”
Because the drives have onboard memory, they can store encoder parameters so that when the encoder goes down, it can be swapped out and the operating parameters can be downloaded from the drive.
Better yet, the drives themselves can be swapped out and their operating parameters reloaded. Gone are the days of spending hours tuning the pots in a system, then facing a repeat of the whole process when the drive has to be replaced. The parameters can be stored in the network and downloaded to the new drive when it is installed, for plug and play operation.
Intelligent drives can perform control tasks. Pushed to their logical extent, they can, in some cases, even eliminate the need for a motion control card. This brings us to one of the current big questions in motion control -- distributed architecture with intelligent drives or centralized architecture with motion controllers? Both have their merits.
Appreciating the subtleties of the two architectures requires understanding the hierarchy of tasks in a motion control system. In a traditional architecture with a so-called dumb drive, the PC is fitted with a motion control card that performs path planning and closes the servo loops. In a distributed architecture, intelligent drives have the processing power to close their own loops, while the PC can take on path planning as a software task.
Consider a drive that is following position, velocity and time commands (PVTs). The PC generates new points in space for all axes at specific intervals, say every 10 ms. An intelligent drive interpolates that PVT data to 250-µs increments to create the appropriate trajectory commands, and closes all the servo loops. In such a system, the motion control card is superfluous.
“What's happening is that the tasks that were performed by the motion control card have gone in two directions,” says Proctor. “Closing the servo loops has gone down to the drive and the path planning has gone to the CPU. That is because two things have happened: the CPUs have become immensely powerful and the drives are more intelligent.”
The distributed approach yields a scalable architecture. In general, motion controllers are modular, meaning that they are designed to handle a set number of axes, typically four. An integrator or machine builder designing a five-axis system would have no choice but to buy two controllers, overspending and overspecifying the system. The intelligent drives approach allows designers to buy just what they need.
The Pros and Cons
Of course, as in all things motion control, there are tradeoffs. Distributed architectures are good at single-axis motion, or even series axis motion (axis A goes, then axis B, then axis C). When it comes to more sophisticated problems, say the classic example of drawing a circle, the results are less effective.
Manufacturers are working on solutions. Technosoft, for example, is developing software to essentially group drives on a CANopen network and designate one of the drives to act as a multi-axis motion-control host to the other drives. “You will be able to group drives on a network and tell one drive to draw a circle. It will take responsibility and act as a host to the rest of the drives for that [operation],” says Chandler.
In the case of certain sophisticated applications, such as motion of tightly coupled robotic arms, however, there may be nothing for it but the motion control card. “When you’re doing some fairly intricate motion with articulated arms, all those servo loops are interdependent,” says Proctor. “As you move one axis, the mechanical load changes on another axis. It's beneficial in those applications to have the position and velocity loop actually closed in a central location and not distributed down to the drive. It’s a fairly unusual situation that normally doesn't happen in motion control applications,” he adds, “but in that case, central control is a good idea.”
One challenge from the user point of view is complexity. Today’s digital drives offer users a wide array of software tools, including control panels, programming wizards and digital storage scope functions. “Users are given full access to every parameter and variable within each control loop today,” says Chandler. “They can also, for example, dynamically switch from one control loop structure to another, selecting different reference and feedback signals for each loop.”
Like the word processors that replaced typewriters, however, intelligent drives require the user to master a more complex tool before being able to take full advantage. “Tuning pots and setting jumpers? Those simple days are now gone,” he says. “Optimizing programming within the drive requires expertise. Once it’s effectively deployed, though, you end up with a very maintainable, serviceable system.”
“You just have to grit your teeth and get on with it because life’s a lot better,” Proctor says matter-of-factly. “In the old days with analog drives, you’d have drift and things would change with temperature. They might have been easy to set up but they had limited diagnostic capabilities and you had little flexibility. You couldn’t change servo loop gains on the fly, for example, in an analog system.”
Today’s computer-savvy graduates also bring a new perspective to the process. They expect to deal with software, they expect to deal with firmware. It’s a changing demographic that’s primed to take advantage of the new drives and their intelligence.
“When you have this intelligence, it opens up an element of customization that was difficult before,” says Proctor. “Instead of potentiometers and circuits, you can customize firmware quickly without incurring cost.”
He’s not alone in seeing distributed control and intelligent drives as the wave of the future. “It's one of the core raging debates within motion control,” says Chandler. “I continue to think distributed will win slowly but it's not been a landslide because it’s difficult to replace the functionality that centralized control brings in some applications.”
And in motion control, as we all know, it comes down to the application. “I’m a realist,” says McLaughlin. “The centralized motion control approach still has a strong presence in the marketplace today. I’m sure manufacturers feel the pressure of intelligent drives and networking, though. Cost and performance are the same, if not better.”
Sounds pretty smart to me.