Fast but steady could win the race

Berkeley Process Control, Inc., Richmond, CA, makes electronic machine-control products. In that business, it's one among many. Unlike many, it's grown by 40% a year for the last eight years. Its success hinges on two things: a stable of patents covering high-speed servo loops, and steady concentration on making control systems that people can actually use.

"From the start," explains BPC President Paul Sagues, "our focus was on making machines work, not on building boards." That focus is reflected in myriad ways, from controllers that haven't changed form factor while growing from 8- to 64-bit RISC architectures over the years, to consistent, simplified wiring schemes.

Most notable: source-code compatibility that's been maintained for more than ten years. Users can take a program written for a 16-bit BPC controller, load it unmodified on a 64-bit unit, and it will work. "It's a crucial element in gaining our customer's confidence," says Sagues.

Consistency, plus performance: According to Wayne Ferrari, BPC's marketing manager, the company's Bam(R) Series 64 controller offers sophisticated multiaxis servo control, compact distributed I/O for logic control, network communications, and operator interface control in one package. Simple connections to the company's touch-screen interface, I/O rack, and Universal Servo Amplifier, and single-cable servomotor connections complete the hardware configuration.

Prime example. The ELS-II Registration and Machine Control System developed by Dri-Tech, Inc., Milwaukee, WI, with cooperation from BPC, exemplifies the performance advantage of dedicated machine controls. ELS-II synchronizes multiple servomotors driving large rotating cylinders on high-speed commercial printing equipment.

Optical encoders on each axis signal the Bam(R) controller over the distributed I/O. By monitoring the signals, the controller can dynamically adjust each cylinder, regardless of speed, achieving printing accuracies of n 0.005 inch, comparable to mechanically synchronized printers. The controller also monitors web tension and multiple zone temperatures throughout the machine, reducing scrap production.

Machine commissioning went smoothly, largely due to BPC's QuickStart(TM) utility. Quickstart interrogates control network nodes, establishing device type and voltage level. It also performs auto-tuning, measuring motor and load characteristics to optimize servo-loop performance.

Compare this approach with others vying for machine-control business. PLC- and CNC-based schemes each have their limitations says Sagues. PLCs handle logic functions well, but "they're running out of steam, they can't do high-speed multi-axis servo control well." Alternatively, CNCs do yeoman work at controlling motors, but it's difficult to perform logic operations with them.

What's common to PLCs and CNCs that's kept them in the machine-control fight? Well-defined, well-maintained operating systems that plant-floor personnel can understand and use. Which brings up bus-based, "open" systems, the third major machine-control alternative.

An open question. As Berkeley people see it, open systems have some glaring disadvantages. First, assembling boards from different vendors for various machine-control functions and getting them to work together involves writing extensive application-specific software-and won't guarantee success. "Your desktop computer can crash with no harm done," Ferrari contends, "but that's unacceptable in a production machine."

The second shortcoming is the software itself. Ad hoc software is notoriously difficult to modify as machines are modified or redesigned. Data from the semiconductor-manufacturing industry, for example, shows that control-software costs skyrocket in the years after a machine's commissioning, as developers try to keep pace with changing customer needs.

Because the software for open systems reflects the programming styles of its writers, it's often difficult for others to decipher or make changes without unforeseen results. Lastly, changes in hardware can quickly render open-systems software obsolete. "Automation customers trade complex gears, belts, and cams for complex software," Sagues contends, "it's unfair to ask them to re-write it every time a new whiz-bang product comes along."

In short, Berkeley feels its approach to machine automation offers more reliability and more options, more easily, than so-called open systems. And, if ease of use and expandability is the real definition of openness, then the company feels its newest product-called Machine Works(TM) -provides the most open system of all.

A new approach. "Machine Works profoundly changes the way automation professionals approach multiaxis machines," says Sagues unabashedly. Instead of working on an idealized model offsite, then testing completed programs on the actual machine, Machine Works runs on the installed Bam controller, guiding automation-program development interactively with users over a touch-screen interface.

The system uses the previously described QuickStart utility to create a database of the machine's unique features. It asks users to define the individual tasks the machine has to perform, then define the sequence in which those tasks need to run. Users can have the system diagnose each separate task, then sequences, and then the completed control program for logical errors and component interference.

Because it contains an extensive library of machine operations, many applications will require no user-written software at all. Extensions to available routines-for example, custom cam-emulating tables-can be added easily, requiring a few lines, not a few thousand lines, to implement.

Although Machine Works was developed for relatively simple applications which had resisted automation because of the cost of conventional software development, it has proven valuable in running larger multitasking, multiaxis machines as well. In beta-site testing, it allowed plant-floor personnel to commission a 4-axis machine in just two days, saving some $25,000 in projected software development costs.

Sagues won't speculate on the eventual success of Machine Works. But discussing its place in his company's history of successful, workable automation products, he grins, saying, "It's the biggest thing we've ever done."