A design engineer faces many demands when designing a motion control system satisfying the motion and interface requirements, finding the right tools for development staying within budget, and providing a straightforward path for future enhancements. The good news is that motion control manufacturers offer a huge number of options to satisfy these requirements. The bad news is that choosing among them can be difficult, if not a bit overwhelming.
Using the latest control schemes in a design can save a great deal of time and money, but it's important to analyze the application to determine whether a stand-alone, bus-based, or PC-based controller is the best way to go. While some of the newer technologies may offer advantages over traditional approaches, it's important to look at the advantages and disadvantages of each type.
This is particularly true in light of continuing advancements in communications, which are making it possible for users to support their machinery as never before. These advancements will determine the future of programmable motion control in terms of how it interacts not only with the machinery, but also with the entire enterprise.
Though there are many communication options available, we feel that Ethernet technology is the most cost effective way to go. The following presents a case for the use of this technology to implement a variety of motion control architectures.
Stand-alone motion control Systems
The first motion controllers were stand-alone proprietary hardware devices that used a custom hardware core for the control of the motion systems. The motion controller was an individual component mounted in an industrial enclosure that communicated to the machine control element via an Rs-232 link. Stand-alone controllers still offer a number of advantages:
- There's usually room for the many discrete connections needed for motion control.
- They are easier to troubleshoot.
- It's easier for the manufacturer to predict the operating environment and test for these situations, which is why stand-alone systems are often perceived as being more reliable.
The stand-alone system, however, has a weak spot: the communication interface. High-performance controllers are limited in performance by the ability to command and transfer data at high speeds. The need is for high-speed communication that places the controller virtually on the bus.
With a bus-based, board-level control scheme, the motion controller has a standard hardware interface and is installed in an expansion slot of the computer. This approach offers three advantages:
- The motion controller communicates via the parallel bus of the computer, rather than using RS232 serial communication. This allows the motion controller and the computer to exchange information at a level and rate not available in a stand-alone controller.
- The packaging of the product can be eliminated and the motion controller can use the power supply of the computer, saving dollars.
- The developer can access all of the hardware and software available to the PC industry, such as video and data acquisition, which are more difficult to integrate with a stand-alone solution.
However, the bus is not the perfect solution either. In a bus-based system, communication connections are forced outside of the computer, which mandates that all signals be transferred outside of the computer. For example, eight axes of control require that 200 discrete wires be connected from the controller to the machine. The standard PC card has a 3-inch edge connection to make the connections from the controller to the machine. This makes it extremely difficult to ensure the reliability of connections. And future system expansion is made difficult by requiring that yet another computer slot be used for connections.
"From a development standpoint, one of the most difficult aspects of designing motion products has been in the prediction of which bus structures will be desirable to our customers," explains Bob Cook, engineering manager at Parker Compumotor. Cook continues, "We were continually chasing the next important bus and ended with a myriad of bus structures from STD to VMF. For Compumotor, the solution was to standardize on the ISA bus as its only bus-based control. This was a liberating decision because it allowed us a hardware focus that we formerly didn't have. However, we knew there were market trends that couldn't be ignored, such as the emergence of the touch-screen computer...."
Industrial computers for motion and machine control
Compumotor introduced its first bus-based controller in 1986. At that time, customers needed to either use a very expensive industrial PC or deploy a "white box" (a standard office PC) in the industrial environment. The development of passive backplane computers improved the industrial PC applications, but limited the depth to which the PC could penetrate. Compumotor was already concerned with the PC card format for motion controllers before the introduction of the first small touch-screen computers. As demand grew for the touch-screen computer it became evident that we needed a strategy to interface the motion controller with this important new hardware.
The challenge was package size. Spoiled by the large size of the industrial PCs, most motion companies built controllers for full size cards only, and even though motion controllers are highly integrated, the high performance controllers on the market are all full size PC cards. By building a full size PC card, the customer is limited to using only the largest models of touch-screen computer; but it was our impression that users were most excited about the 6-inch touch-screen computer, and needed a strategy to work with this product. We believe the solution to this problem is Ethernet.
As product architecture schemes were developed, the communication from the motion controller to the machine control element became more and more important. "Our controller sales are equally divided between stand-alone systems that use a PLC for machine control and those that use a PC for machine control," says Cook.
The machine control and motion control industries were previously totally independent. Design engineers would develop specific expertise in each of the products and would use simple means, such as RS-232 ASCII drivers or discrete I/O points, to bring them together. We feel this is likely to change in the future for several reasons.
First, we believe that the continued acceptance and development of fieldbus will replace RS-232 and discrete connections with a faster and more efficient alternative. The fieldbus will allow those who wish to continue using the PLC as their computing engine to improve their communication to the motion controller.
Second, use of the soft PLC is accelerating. Although fieldbus is an important option for the industry in our opinion it's not a good fit with the new breed of soft PLC's. Using fieldbus with a soft PLC requires a PC card for the fieldbus hardware that is native to the PLC system. We feel that the need for the card often offsets many of the cost gains derived from the fieldbus system. What's needed is a standard hardware platform with the performance of the fieldbus system and the might of the computer industry to bring its cost down. Here again, we say the answer is Ethernet.
What is Ethernet?
Ethernet is the most widely accepted local area network in the industry. Currently it operates at maximum speeds of either 10 million and 100 million bits per second (Mbps), with l0 Mbps being the more widely used of the two technologies.
Ethernet is a true industry standard, governed by the IEEE 802.3 specifications relating to hardware design and communication protocol. This open standard ensures that no matter which vendor's equipment is used, inter-operability is guaranteed. Couple Ethernet with the TCP/IP protocol and virtually any TCP/IP capable device, or software, can communicate with another such device.
The most common physical medium for transferring Ethernet signals at this time is 10Base-T: 10M bits per second carried over a twisted pair cable terminated on each end with RJ-45 telephone style jacks. The other more increasingly encountered medium is 10Base-F or, 10M bits per second carried over a fiber optic cable for increased noise immunity. Both of these media are used in the Fast Ethernet (100-Mbps) as well.
Ethernet is also vendor neutral and, therefore, inexpensive to implement. Most computers can be hardware configured for Ethernet capability for around $100.00. Laptop computers use PC Cards (formerly known as PCMCIA). Desktop computers use Ethernet cards in any standard bus structure (PCI, ISA, VMF, etc).
Ethernet speed and determinism
As Ethernet gains acceptance in industrial networking and motion control, two issues will arise. The first will pertain to which high-speed serial communication protocol is the fastest. The second will focus on Ethernet's nondeterministic nature and question whether it's suitable for information exchange requiring a known transmission time.
Before drawing any conclusions, it is important to understand the difference between theoretical and actual data transmission speed, and which factors contribute to collisions on an Ethernet-based network.
First, even though a data transmission medium claims to have a 10M bits per second 10Base-T Ethernet) or 12M bits per second (Universal Serial Bus) or 100M bits per second (100Base-T Ethernet) maximum data transmission rate, the actual data transmission rate achieved from a device using one of these media may not come close to these values. Playing a much larger role in the speed of data transmission are microprocessor computing power and the ability of the device to accept and process the information being passed to it over the high-speed bus. Command language complexity is another factor that affects the overall performance of data transmission, and application intensity will impact the data transmission and execution speed of a motion controller.
For example, consider these two systems:
- A controller using a 20M bit per second bus and a microprocessor capable of 8 MIPS (million instructions per second),
- A controller using a 100M bit per second bus and a microprocessor capable of 8 MIPS.
Looking solely at this comparison, it would seem as though system two is the better choice. It uses a higher speed bus with the same processor.
But before making a choice, take a look at the command language complexity and the application complexity of the two systems. Let's say that the first system is using a relatively simplistic motion control language and executing a very basic pick-and-place application, while the second system is using a more complex programming language and performing a data-intensive cam profiling application. Now the decision as to which system is better has become more difficult. Although system two uses a higher speed bus, the processor may need more time to handle all the information being sent to it. In fact, the 8 MIPS processor may not even be able to handle all the information being sent over the 20 Mbps bus, which negates the higher speed bus of system two.
Does all of this sound confusing? It should, because it takes much more knowledge than just the maximum data transmission rates of a certain bus to determine if it will solve an application. The lesson here is that you really need to investigate the application and controller parameters of the system before making a selection.
The second issue, regarding Ethernet's nondeterministic nature, needs some investigation as well. First of all, what is it that causes Ethernet to be nondeterministic? The answer is network traffic. Network traffic, or data from other devices on the same network, can cause collisions of data. A collision occurs when two stations on a network attempt to transmit at exactly the same time. When this happens, Ethernet uses a back-off algorithm that randomly assigns a time delay to each station before it can retransmit its data. This ensures that collisions are handled quickly and fairly, usually within microseconds of collision detection.
For applications that do not require timing-critical data exchange, this isn't likely to a problem and probably will never be noticed. However, where data exchange timing is critical, how can Ethernet collision problems be avoided? The solution is simple: by using a closed network. A closed network is one in which only the controller and host computer are connected. Data is exchanged between two devices, which eliminates the possibility of network traffic causing collisions. This is now a system in which data transmission times are repeatable, and the user receives all the benefits of integrating Ethernet.
If an application absolutely requires linking multiple controllers via Ethernet, a closed network consisting of only the controllers and computers communicating will eliminate most of the network traffic from things like random e-mails and web browser sessions. This will ensure a data transmission time that is rapid and reliable.
The controller of the future
The controller of the future will have the ability to solve complex motion tasks, communicate with a wide variety of machine control elements, and allow simple and low-cost connections to the rest of the machine.
Through the use of Ethernet, a stand-alone motion controller fits within machine control schemes using PLCs or PCs with one efficient industrial package.
Motion control can be a complicated task. However, emerging communication technologies have come together with the motion industry to greatly simplify problems facing machine builders.
About the author
Bill Green is product planning manager at Parker Hannifin, Compumotor Division. He holds a bachelor's degree in mechanical engineering from University of Wisconsin, and has more than ten years experience in the motion control industry.