It takes brains to rule brawn, whether in a biological system or in a motion control cell. Controllers are the brains of motion control. Back in the day, each axis of a system had its own servo amplifier. The motions of each axis were coordinated by a controller, which typically resided in a separate cabinet. No more. These days, the trend is toward pushing intelligence out into the system. Controllers are not just distributed, but integrated into the driver or into the robots themselves. Meanwhile, other products help integrate the separate control layers into a single unit.
“I think more and more intelligence is being put inside the drive or as near the drive as it can be,” says Sue Dorscheid, product line manager for Danaher Motion (Rockford, Illinois - www.danahermotion.com). “The drives in general are getting smarter in terms of the algorithms that are implemented inside them. There’s more processing power in the drive to control the motor, so you may as well use that power to do other things.”
Enter the new generation of controller products, which offers efficiencies of operation and savings in space. Factory floor space is always at a premium, and making room for control cabinets can be very expensive. Two decades ago, the control cabinets that housed the drives and controllers were the size of a washing machine. Advances in technology reduced them to the size of large microwaves, and then the size of a small breadbox. More than one breadbox, depending on your system architecture.
“You have to plan out a production cell in factory space,” says Dave Pap Rocki, chief technology officer at Adept Technology Inc. (Livermore, California - www.adept.com). “A key aspect is just making allowances for the control cabinets and the peripheral equipment that is required to operate these production cells.”
With the old controllers, a separate cable had to run from each controller to each axis, which made physical routing an issue. “The amount and size of the wiring became very intensive going out from the cabinets to the individual mechanisms,” says Pap Rocki. “As the bundle size gets bigger, physically routing it becomes more and more of an adventure.”
It’s not just a matter of size. Electronic noise is also an issue. Circumventing this can require running two separate cabling races, one to segregate the noisy cables. This adds to cost, takes up space, and still you run the risk of compromising performance.
Integration provides an answer.
Routes to Integration
Danaher’s approach to the issue of controller connections is to integrate the controller into the drive. Instead of being a separate box, the controller is repackaged onto a printed circuit board that can be installed in the servo amplifier itself. “Drives that have control capability have been around for many years,” says Dorscheid, “but usually they’re for a single axis, they’re only for the one drive.” The Danaher product can be daisy chained to other drives with Cat5 Ethernet cable to control up to 15 additional drives besides the one in which the control card resides.
The approach saves floor space, but it also offers other advantages as a result of its proximity to the drive, such as the ability to get real-time information like fault data. “You can change tuning parameters on the fly, you can change current limits, monitor average current and other things like that in real time while your process is running,” she says. This data can be displayed on the human/machine interface or, with the addition of the digital networks that are becoming standard features on today’s factory floors, sent to the central office.
The connectivity provided by the daisy-chain connection simplifies set up, Dorscheid notes. “Say you want to set up drives or change some of the tuning parameters as part of the machine commissioning process. You can connect just to that drive with the control in it and actually have access to the others. You can run an oscilloscope on them and do your tuning without having to move your cable around.”
Integrating the Layers
Today’s factory machines run through a complex ballet of motions and operations -- machining, motion control, and robotics. The traditional approach has been to have a separate programmable logic controller (PLC) run each operation. Now, the focus is on integrating the control of those layers. “There’s been a trend in the industry over the last five years to introduce control platforms that are more sophisticated and more capable than what would be considered normal for a standard PLC,” says John Browett, product marketing manager at Mitsubishi Electric Automation Inc. (Vernon Hills, Illinois - www.meau.com).
Different control layers means separate cabinets full of electronics that cost money and take up floor space. In addition, there’s the issue of making the system work as a whole. “In the past, maybe you’d have Company A’s PLC, Company B’s motion control, and Company C’s robotics,” says Browett. “The challenge was how do we make all this work together, what kind of structure do we need to build, what kind of software do we need to write to make them all talk to each other? There was a lot of time and effort spent on the engineering to glue all this together.”
Mitsubishi offers a platform that can not only perform functions like ladder programming and sequences of operations, it can also control up to 96 axes and offers flexibility to run applications alongside industrial I/O. The product integrates computer numeric controlled (CNC) operations so that it can run machine tools and motion control and, eventually, robot control, as well.
“The functions are enabled by what CPU you have installed on the platform,” says Browett. “You basically have the backplane and you add the function that you need, and each function is enabled by a different CPU. For example, if you need PLC capability then you add a PLC CPU, if you need motion control you add a motion control CPU, if you need CNC, then you add a CNC CPU. You build up the system you need to get what you want at the end.”
Adept’s solution to the space issue was to take the controllers from the cabinet and integrate them into the bases of their SCARA robots. “We embed a high-level controller that’s actually doing the logic sequencing along with a kinematic solution,” Pap Rocki says. “When you can embed the control closer and closer to the mechanism, you save substantial money in the cost of the floor space.”
The Adept design uses a high-speed, time-synchronized bus to communicate between the different modules and reduce wiring. Like Danaher, the company found other advantages in locating the intelligence out in the system and having storage near the mechanism itself.
“There is a calibration procedure that you go through at the factory in which you characterize each specific module or robot,” says Pap Rocki. “We found we could store the physical parameters in the controllers and that would be localized to the module itself or embedded so if a customer needed to swap out a module, [the replacement] would also contain the calibration constant.” A subset of the control language embedded in the unit also allows user to program simple applications.
One challenge to integrating controllers is thermal management. “When you look at how fast you can make a mechanism move, for a certain cycle time, your limiting factor quite often is how hard you can drive it before the motor starts to melt,” Pap Rocki says. “You get into extensive system balancing -- how much weight do I put out there? How much weight do I have to move? How efficient are my motors?”
There are many methods of thermal management, including fans, water cooling, thermo-electric coolers and even piezoelectric coolers. Ultimately, the path the Adept team took was the simplest. The robot casting was made of metal. By ensuring clear thermal conduction paths to the exterior of the casting, the team was able to easily dissipate heat.
Balancing for Performance
These designs don’t mean that standalone controllers are a thing of the past. If there is any one truism in engineering, it is that there is no universal solution. Always, there are tradeoffs. Systems with very high axes counts, for instance, or those with stringent process-control requirements or high functionality -- tactile or force sensing, say -- might be best suited to a more traditional design, at least for the time being.
“Currently, we can take a subset of our control system and add it into the base. When we need to add conveyor tracking or integrated vision, we go back to a secondary box,” says Pap Rocki. “This is kind of the next goal, can we get all that processing into the base of the robot? When does it make sense, now that some of that processing is getting to a level where it generates more heat than we can dissipate in that environment?”
Ultimately, it goes back to balancing. “Even if you could put your entire control capability down in the motor, you wouldn’t want to run all your wires there. Part of the logistics and layout of a cell is for the flow of diagnostics and hookup and so forth. The question to ask is where does it make sense to take the wires?”
Applied judiciously, integrating controllers is an approach that offers benefits in a number of applications. As a result, it’s a process that will be ongoing. “I do think that as we go on, more and more intelligence is going to migrate down to the drive,” says Dorscheid. Clearly, it’s a trend that’s here to stay.