PC controls: An unfulfilled promise

Manufacturing Engineering, Jul 1998 by Graham, Bryan A

A user's take on how to make PC-based control successful and why it's not

As the personal computer's popularity soared in the '80s, so did the appeal of controllers based on commodity-type products. Since 1986, when the Intel x86 microprocessor opened the doors to PC-based control, users have embraced logic or motion controllers with the Intel x86 at the heart of the system. They hoped that in the '90s PCbased controls would replace proprietary controllers so that they could leverage commodity components, generic hardware, and generic software.

Although Lamb Technicon develops machine tools, not control systems, we search out and use the controls that offer the most value to our customers. In the early '70s we were involved in developing solidstate controls. In the late '70s and in the '80s we helped develop parallel computers attached to logic controllers for real-time diagnostics.

In the '90s, as the automotive industry became committed to PC-based controls, we, perhaps more than any other machine tool builder, began to use them in the metalcutting environment. We used Taylor Waltz and Flo-Pro software on dial transfer and on palletized machines with as many as 6000 I/O points.

We recently surveyed purchasers of our metalcutting equipment to find out what they wanted from the new controls. They told us simplicity. Industry consortia and trade journals define simplicity as lower system costs; improved reliability of machinery; reduced order life cycle through reusing logic and applications; agile and reconfigurable manufacturing systems; smoother integration of motion control with machine logic; and the combination of Human-Machine Interface (HMI functions with application logic.

These goals are just promises right now. Let's look at the reality today.

Lower Cost. PC-based control systems based on commodity components can deliver on the cost promise in the long term. I stress long term because you won't save money buying one today. To the high cost of industrialized PCs and flat-panel displays you must add the cost of training and standards development for each new control software and operating system configuration.

Machine Reliability. PC-based systems tend to degrade reliability. Their operating systems are subject to crashes, computer viruses, software bugs, and human error. We've seen users cause spectacular crashes by unintentionally rebooting the computer while machines were operating. PC-based equipment is not built to withstand the harsh metalcutting environment. Proprietary systems are certainly more bug-free and robust than PC-based systems in 1998.

Reduced Order Life Cycle.

As machine tool builders, we thought PC-based systems' common hardware and software would let us re-use applicationspecific software across multiple platforms. Controls development is typically a critical path for machine deliveries, and we expected order life cycles to drop dramatically.

Instead, we face a proliferation of systems whose application code cannot be interchanged. Though many of them adopted the IEC-1131 standard intended to commonize instructions in several programming environments, no certification exists for products claiming IEC-1131 conformance, nor is there a standard file format that allows users to port a machine builder's standard logic from one system to another.

Agility and Reconfigurability. OEMs and users struggle to make their manufacturing systems flexible enough for quick reconfiguration to support product or volume changes. They expect their PC-based controls to support this effort, since new software can be quickly added to a generic hardware platform. Instead, as with business software and operating systems, after two or three years in which microprocessor clock rates increase exponentially, the control software obsoletes older systems in speed, data bus size, memory, and peripheral capabilities. What's more, new logic packages have only limited ability to make insertions or completely reconfigure data table and device network I/O assignments. Hence a company's controls engineers may need to work hard and long to make a simple mechanical modification.

Integrating Motion Control. Unlike proprietary controls vendors, open controls vendors (and IEC1131 software suppliers) must simplify and provide standard open interfaces between control logic, generic interpolator software, and motion control interfaces such as SERCOS or analog systems. If interpolation software can be integrated, one architecture can handle varying levels of motion control in flexible transfer machine applications.

Today no commercially available product meets this challenge. Instead, control logic interfaces are complex and limited to proprietary motion coprocessor hardware or discrete interfaces to external drives.

Combining HMI Functions with Logic. Here's the PC's big success story. Because the same processor and monitor can be used, we see tighter coupling of the HMI function to the control hardware. Though control logic and diagnostic logic are often separate programs or routines, they all reside within the same executive software program or at least on the same operating system.