Motion control has always been challenging to control systems because it is the hardest of hard real time applications.  But our notion of time has changed.  The processors that manage electronically commutated motors operate at 50mHz with incredible code efficiency, requiring nonosecond precision oscilloscopes to measure events.  Considering that we used to be thrilled at the prospect of controlling things in the microsecond world, I’d say that’s game-changing (another over-used catch phrase).

So corresponding changes in the communications realms shouldn’t be a surprise.   Though I think that we hadn’t even invented Ethernet 30 years ago.  But ignoring that detail, we’ve seen Ethernet technology bloom from a mere megabit per second to Gigabit Ethernet.  Bandwidth is not a problem.  And with universal adoption of the technology by business and telephone communications systems, the cost of the physical layer and connectors have dropped to incredibly low levels.  As you would expect.

In fact, Ethernet is so cost effective, it’s starting to take over the industrial control landscape.  To the point where Ethernet connectors are available for dust tight and wash down environments.  That’s pretty extreme for a consumer grade communications platform.  But that’s wasn’t my primary point.

The dilemma for applying a communications protocol to motion control is it’s ability to move data faster than the synchronous control of the motor.  So industrial networks, even at a few megahertz, can’t keep up.  Particularly when there are two axes of motion which must coordinate their relative motion at the update rate of their position feedback sensors.  And in the case of linear motors with extremely high resolution tape scale encoders, you can end up in situations where the feedback is running 20 mHz.  So good luck coordinating 2 or 3 linear motor axes.

But, of course, we do this kind of thing every day.  With micro controllers.  But not with networks.  But what if you could?  It’s coming soon.  IEEE-1588 is a time synchronous version of Ethernet that permits precise coordination of data movement over Ethernet.  And it’s compatible with existing Ethernet networks.  Sounds like a pretty good deal.  And that makes it a workable solution for coordinated axes of motion control.  It’s been tested at a technical university in Zurich Switzerland.  And like good solutions, may have many other applications.  Look for it in new motor control products coming soon.

The ACOPSOSmulti cooling system is available in standard, feed-through and water-cooled versions.

By Leslie Langnau, Managing Editor
Design World

In mechatronic projects, the focus is often on the mechanical and electrical aspects of a system as engineers concentrate on throughput, speeds, accuracy, and so on.  How these system goals affect the desired control selection may not be addressed until too late to make changes. Mechanical engineers do their part, then electrical engineers do their part, then, the controls engineers must make it all work.

In addition to finding ways to improve the communication and interaction of the various engineering disciplines, there are other design aspects that affect controls to keep in mind. Robert Muehlfellner, Director Automation Technology, B&R Industrial Automation Corp., offers a few.

— Don’t overlook the fact that mechanical components have weight and inertia, which influence speed and throughput.  These properties can heavily influence the control choice.

— I/O, motion, and communication can be managed by one control or multiple controls. The choice will affect connectivity and communication decisions.  Increasingly controls vendors offer a “one-stop-shopping” approach to their products, so if you can choose one control for all these needs, you can save you or your customer time and headaches.

— Consider controls that are modular and with features that can be scaled up or down. Consider controls available in family style configurations too.

— Work with development software that combines or works with mechanical, electrical, and control design tools. A development package that also interfaces with E-cad packages, for example, helps move control system hardware configuration data into electrical packages for faster creation of electrical drawings.

— Test the desired control with a machine model. Mechanical engineers can develop machine models in packages like Simulink, which lets controls engineers run models of their control design in conjunction with the application. Such programs allow virtual development and verification of expected machine throughput prior to testing the controls application on a real prototype machine.

— Consider power needs. It may be desirable to configure independent supply voltages for some drive systems, so as to, for example, eliminate the need for isolation transformers.

— Consider the need for diagnostic features in the control system. Quick and accurate diagnosis of machine problems improves machine uptime. Choose controls that automatically collect data or run diagnostic routines. A seamless integration of subsystems like drive controls combined with smart features can also allow for predictive rather then preventive maintenance.

— Choose controls and components that conserve energy without compromising function. For example, the efficiency of hydraulic systems can be improved with variable speed control of the pump to maintain constant system pressure instead of the conventional way of running the pump at constant speed and bleeding excess pressure through the pressure relief valve.

— To save time and costs for engineering changes throughout the lifecycle of the machine, choose equipment that will be available for many years to come.

— And keep in mind that while maintaining your current technology, plan to leverage new technology while reusing the intellectual investment. Focus on making your next generation application better instead of dealing with migration issues.

We all share the concern that unemployment is up and many areas of the economy are slow.  But let’s be sure that when the government says its going to spend our money, that the decisions are based on sound strategy.  Maybe government spending money that it doesn’t currently have isn’t such a great idea.

Is the technology being evaluated on the cost per kilowatt hour?  Or are other metrics being used for evaluation?

In the solar energy field, what is the opportunity to generate electricity on the flat roof of buildings?  If we look at the number of photovoltaic panels that can be put on a roof, its a big number.  But if the panels cost $550. each and they are 11.625 square feet, it is going to cost $47./square foot to pay for the panels not including installation labor and materials.  That’s a lot of money.

If its a high quality PV panel at 210 Watts of output, it produces 18 Watts per square foot.  With the number of square feet of roof available, we could generate enough electricity to (fill in the blank with whatever you dream about that you think will save the world).  This is where the politically savvy get you to agree with them, even though they may not know what they are talking about.  They link the program they are selling to your hopes and dreams to get you to buy into the deal.

If we could put solar trackers to follow the sun, we would be able to increase the total output of the solar cells by 30% or more.  (That data comes from the National Renewable Energy Laboratory, and they should know).  Tracking panels cast shadows, so you can’t use 100% of the square footage on the roof for panels.  So some may argue that solar trackers will destroy our hopes of energy independence by decreasing the number of panels that can be installed on rooftops.  (Watts per square foot).

But 30% increase in available power means we can either produce more power, or produce the same power as before with less panels.  Less panels means less dollar$ per kilowatt.  Less cost means quicker return on investment.

Ask how much it costs.  Make sure you get an answer.  When does the system pay for itself?  Make sure you get an answer. If the thing is going to be financed with tax dollars, who owns it?  If we paid for it with tax dollars, shouldn’t we get the energy for free?  Which calculation makes sense?

ProBatch+ runs under Microsoft Windows and uses the powerful MS Access database to manage large quantities of raw material, recipe and production data, easily and securely. Universal record-keeping of production data provides all batch record traceability data for the user’s SOP and Quality Control protocols.

ProBatch+ is ideal for manufacturers in the cosmetic, chemical and food industries as it covers a wide variety of products and at the same time can react quickly to changing market conditions and procedures.

Key advantages of Sartorius ProBatch+ Software:
Material management: ProBatch+ can manage a large number of materials pertaining to a production line. With only a few operator steps, raw material quantities used and recipe quantities produced can be called up. Automatic stock monitoring is included. Material tracking with three freely selectable batch parameters and one data field can be optionally activated.
Report management: At the end of every recipe run, ProBatch+ generates batch and production reports. Material management makes reports available on consumption, production quantities and stock levels. All reports can easily be individually adapted to the user’s SOP.
Connection using established standards: Communication between ProBatch+ and controllers takes place using OPC in accordance with ANSI/ISO S88.01 Batch Control Standards. In order to connect to visualization systems, dynamic data is supplied via DDE and OPC. For data transfer of production data to an ERP System, the SQL and ODBC database interfaces can be used. This consistent use of established standards simplifies connection and reduces commissioning times and costs.
Script function: The script language integrated in ProBatch+ can be used for easy and flexible conversion of the user’s specific process requirements. The script functions make it possible to implement rapid adaptations to process changes without the need to carry out expensive programming in the PLC.

www.sartorius-mechatronics.com

Critical to any mechatronics system is the control. One of the newest controllers is the programmable automation controller. Here are tips on selecting one for your specific application.

By Kelly Downey,
Electrical Engineer
Opto 22

Industrial applications continue to increase in complexity, requiring controls that can integrate multiple systems that incorporate discrete, motion control, and process tasks and that can gather, process, and transmit real time data to company databases. Programmable automation controllers (PACs) can be one choice for managing this complexity because they combine the capabilities of several traditional control and monitoring systems. Typically, they have features found in programmable logic controllers (PLCs), distributed control systems (DCSs),remote terminal units (RTUs), and personal computers (PCs).
Even so, control manufacturers offer PACs with varying capabilities. Thus, there are several considerations to keep in mind with your selection.

Opto 22’s PAC Project software suite is an example of PAC software. Integrated with SNAP PAC controllers, the suite includes control programming and HMI development software, plus optional OPC server and database connectivity software. I/O points and variables are stored in a single tagname database. Flowchart-based PAC Control development software from Opto 22 is fully integrated with SNAP PAC programmable automation controllers.

— For a new installation, most PACs feature monitoring, control, and data acquisition capabilities plus Ethernet connections. They should also include full analog and digital control, motion control, the ability to interface with serial devices, and remote monitoring. You may not have an immediate need for these features, but needs have a way of expanding.
— Look for PACs with features that can reduce network hardware expenditures, such as extra built-in network interfaces.
— For an existing installation, a PAC can consolidate separate systems and link them with company computers to exchange control, production, and monitoring data as needed. It is important to verify that new PACs are compatible with all legacy systems, including all networks and protocols.
— Choose the PAC to suit the size of your system. For example, a PAC that mounts on the I/O rack may be more suited to cell control and RTU-type installations. Extensive systems may need more powerful, standalone PACs.
— For efficiency, choose a PAC-based system that uses distributed intelligence, not just a distributed architecture. Distributed intelligence offloads many control functions to remote processors co-located with distributed I/O. Distributed intelligence shortens wiring runs, reduces network traffic, maintains critical control should communications fail, and frees the central controller for supervisory tasks.
— Choose a PAC that has all the networking and communication options you need—and anticipate needing—built in. Networks may include Ethernet (either wired or wireless), and serial. Other communication options include OPC, Modbus® or Modbus/TCP, Profibus®, Allen-Bradley DF1, and other control standards. For communication with computer networks, you may also need protocols such as TCP/IP, SMTP for email, SNMP for network management, and FTP for file transfer.
— PACs tend to be hardened for industrial environments. However, if your application involves extreme temperatures, vibration, dampness, dust, electrical noise, or other exceptional conditions, provide necessary enclosures and protection just as you would for a traditional control system.
— Signal requirements for inputs and outputs vary widely, including temperature, rate, RMS, pH/ORP, load cell, in addition to voltage and current. The PAC should communicate with all signal types natively rather than requiring signal conditioners. Where cabinet space is limited, high-density I/O is a good choice. Software-configurable I/O—for example, an input module configurable as any of several thermocouple types—offers flexibility and reduces the number of spares you need to have on hand.
— Many PAC-based systems have advanced control capabilities built in. They can satisfy requirements for high-speed digital control, motion control, PID loop control, and mathematically complex logic, without expensive add-ons.
— Because a PAC is similar to a PC, its integrated software includes such programming features as subroutines, string handling, complex conditions, and floating-point math. In addition, a PAC can often be programmed in C or other standard programming languages. While it may seem easier to program with tools you are familiar with (such as ladder logic), you may find that software designed for a PAC is more efficient for expanding needs. Plus, it will likely save development time and effort.
— For data acquisition applications, a PAC should have substantial memory for acquiring and storing data; and the ability to share data directly with corporate databases over an Ethernet network. If control networks and computer networks need to be separated, consider how you’ll accomplish this. One way is to segment networks using independent Ethernet network interfaces on the PAC itself.
— Check out the human-machine interface (HMI) options. Often, the software will use a single tagname database. Once you define variables and I/O in the control software, you can immediately use them in the HMI software. A PAC should also communicate with third-party HMIs using OPC. Other options, such as a touch screen terminal, may be available as well.
— Finally, plan for the future. When your needs change, additional distributed I/O should be handled by the same PAC—as should process, discrete, and motion control.
All of these types of control should be programmable in the same software as part of the same system, and most changes should require no middleware or add-ons.

This PAC from Opto 22 includes two independent Ethernet network interfaces plus four serial ports configurable for RS-232 or RS-485/422.
Opto 22
http://www.opto22.com

According to the ARC Advisory Group, generally credited with coining the term PAC, among a PAC’s defining characteristics are three elements directly related to software:

• Tightly integrated controller hardware and software. The software used with a programmable automation controller is designed specifically for the PAC.

• A single development platform, using common tagging and a single database for development tasks across a range of disciplines.

• Programmability using software tools capable of designing control programs to support a process that “flows” across several machines or units.

Let’s take a closer look at these characteristics.

Tightly integrated hardware and software
When hardware and software are designed together, not only does it often lower software costs, systems are usually easy and fast to build. The integration eliminates the need for drivers, which eliminates the task of debugging.

If problems occur, you have just one company to call for product support or one website to visit for information. Documentation is often more complete, as well.

A single development platform
The software built for a PAC is not just integrated with the hardware it runs on. It is also internally integrated: it provides an integrated development environment (IDE) for programming as well as a suite of related programs for HMI development and other purposes.

The IDE is a single software program that handles everything related to control programming, such as editing, compiling, and debugging. Plus, a software suite made of two or more applications offers a similar look and feel in all of them.

More importantly, the software applications in the suite work together behind the scenes in ways that significantly reduce development time.

Common tagging means that names and definitions you set up in one of the suite’s applications are also used in the others. For example, if you define a string variable in the control development software, that same definition will be used in the human-machine interface (HMI) development software. If you name a digital I/O point in the control software, that name will automatically appear when you’re configuring OPC data communications.

Because all these common tags are kept in a single database used by all the applications in the software suite, you don’t have to reenter tags or maintain and reconcile lists. Development tasks can be finished more quickly and easily.

Supporting a process that flows
The same hardware can be used in multiple domains, including logic, motion, drives, and process control. Thus, it follows that the software must be capable of programming all control and monitoring tasks of multiple domains.

The software must let the developer mix and incorporate tasks as needed into control programs, so these programs can “flow” as the requirements of the application dictate.

The number of PACs needed will depend on application requirements, however, each PAC can be used in any domain or in multiple domains. Because the application requires processes that flow into each other over space and time, the PAC software accommodates that flow and integrates these multiple domains into one system.

This table compares features of a PAC with common features of PCs, PLCs, DCSs, and RTUs to help you choose all the features you anticipate needing, both now and in the future.

Networks for motion are still a in category by themselves. Sercos has gone through major evolution to its current level, Sercos III, to continue to hold its position as the top performing motion network. Other implementations of Ethernet, EtherCat, Powerlink and others bring the high bandwidth available in Ethernet and add features to the network to insure its performance for motion control applications.

But the network technology momentum continues. There is a specification from IEEE, 1588 Protocol, which adds hard real time to Ethernet by clocking to make sure that messages get where they are supposed to be WHEN they are supposed to be there. This feature creates a level of determinism that has long been a stumbling block to more broad acceptance of Ethernet in the industrial community, possibly eliminating any serious impediment to using Ethernet for motion.

The controversy is usually around the question of “What is Real Time?” How fast is fast? Well, its usually whatever is fast enough for your specific application. But that doesn’t really help control system manufacturers when developing solutions for a broad audience. So its nice to find that the technology migration is starting to resolve some of the basic issues with respect to motion control, with something more broad than a vendor specific solution.

Even the Open DeviceNet Vendors Association seems to be exploring the potential of IEEE 1588 as their Common Industrial Protocol as a platform for bringing the legacy networks of manufacturers together as an overall solution. This is a very significant effort, one that has been difficult to achieve, that many users need help with. Operating a manufacturing or process plant is hard enough without having 3 or 4 different networks to maintain, and worse still, exchange information across different platforms.

My guess is the cost pressure of inexpensive Ethernet components will continue to push manufacturers toward finding similar solutions. But is sure would bring everyone along more quickly if the competition among control system providers were balanced with an option that everyone can find acceptable. IEEE 1588 is certainly a possibility worth considering.

Motion Networks are a breed apart, Sercos, Firewire, USB and many proprietary solutions are offered in today’s marketplace. Claims are made about, not surprisingly, speed. And determinism. That’s the tricky part. Guarantee the message got there when it was supposed to.

And its kind of the same problem for PLC’s doing motion control. Yes, its true that most applications do not require tight coordination between two or more axes. So its perfectly reasonable to put the motion in the PLC and let it referee when independent motion axes start. It leads to the assumption that everything is fine no matter how many axes and if any are coordinated, synchronized or registered to a moving target.

Also, not really. Position register information about where the axis is at any given time is a data request that must be sent and received before it can be acted on. That means at least two cycles of the system must be executed before the value is updated and read by the program. Very confusing problem to diagnose based on the behavior of the machine, because at low speeds everything may work fine. But when you crank it up, the process doesn’t work.

So it is with networks. Speed and determinism bandwidths must be fast enough to guarantee that the position information for any axis is where it is supposed to be when it is supposed to be there.

So the network communications wars have gone the same way. Ethernet was a big deal at 1 Megabit. Did we think we would ever need more bandwidth? Now we have 100 Megabit as the common platform with 1Gig on the horizon, inexpensive connectors and wire, wireless if you like, and things are great. Except the “legacy” networks haven’t exceeded their useful life expectancy and don’t convert to Ethernet easily.

So what will the future bring? More network solutions. More attempts at building the perfect network. Perfect for what? Every application requirement has to be considered. For the motion world, where the most demanding performance is required, a different breed of network. Give us something with clock based synchronism to make all that bandwidth useful.

To be continued

By Andy Urda, Director Channel & Industry Marketing

Yaskawa Electric

In 1953, Yaskawa Electric began its active role in advancing technology in the field of motion control when the company introduced its servo motor line, the Minertia® Motor (named for minimum inertia). These servomotors made rotation exact to the proportion of conduction. Due to their very low inertia, they handle extremely fast starts and stops. Originally, they were applied to electrical actuators for the control of mechanical arms. Today, they are also found in many industrial automation applications. Yaskawa’s signature phrase of the late 1960’s and early 1970’s was “Mochintrol,” which was created from the combination of motor, machine, and control. Mochintrol became a registered trademark for the company in 1971.

In 1969, though, Yaskawa engineers took a completely different look at machine control. They were looking for a way to describe the trend of organically joining mechanisms with electronics. The result was the term “Mechatronics.” Yaskawa applied for a registered trademark for this term on August 26, 1970. It was registered on January 22, 1973.

However, the engineers and managers felt too far ahead of the time and did not spread the term. It wasn’t until the debut in 1987 of the magazine “Mechatronics” issued by Gicho (The Technical Investigations Society) that their concern changed. As the term gained widespread use, rather than contribute to confusion that might ensue if they persisted in maintaining exclusive rights to the term, Yaskawa chose to relinquish its trademark rights.

The Evolution of “Mechatronics”
Industrial networks grew out of the merging of mechanisms and electronics. Mechatrolink was inspired by Mechatronics. This high-speed motion control interface between machine controllers, servos, and variable frequency drives also moves data among third party controllers, sensors, I/O, and test and measurement devices.

Industrial networks grew out of the merging of mechanisms and electronics. Mechatrolink was inspired by Mechatronics. This high-speed motion control interface between machine controllers, servos, and variable frequency drives also moves data among third party controllers, sensors, I/O, and test and measurement devices.

The word Mechatronics has now reached global usage. It is the defacto term used to describe the blending of mechanical and electrical apparatus, as well as the inclusion and integration of computers, control systems and software with that apparatus. A Google search yields pages of links of companies with Mechatronics as part of their name. Such a search also brings up a number of colleges with courses focused on the subject. One educational and research site is that of the Mechatronics Research Unit, (MRO), which is currently working on a project known as the Elektroadhesive Robot Microgripper. This gripper can handle components less than 1 mm in diameter.

The term Mechatronics, though, continues to evolve. From describing a list of engineering disciplines that are brought to bear on a design, the term now includes how these engineering disciplines affect the manufacture, transport, and consumption of products. What was created to describe the synergy of electromechanical components on the factory floor has now moved to describe the synergy of engineering into everyday life.

Perhaps the best example of this is the modern automobile and its increasing number of control systems. According to a 2000 report from The Competitive and Sustainable Growth (GROWTH) Programme, a European research group, “The demand for ever-increasing safety and comfort, from ABS brakes to electric seats, has increased the number of electric motors in a typical car to 40. Future cars may contain upwards of 100 electric motors.”

Managing these electric motors are semiconductor chips with embedded control functions. The number of semiconductor chips used in cars is up an average of 8%. A typical car has over two hundred dollars worth of control chips, whereas the average cell phone has less than ten dollars1. A common application is car safety, such as vehicle stability control. A system of sensors and components detect when a car begins to lose stability. They send data to the control chips, which then work to prevent the car from rolling over. Other safety-oriented projects in development include sensors that alert a driver when it is safe to change lanes. Mechatronic technology also enables drive-by-wire functions, replacing traditional mechanical and hydraulic control systems with electromechanical actuators and human-machine interfaces, such as pedal and steering-feel emulators. In many cases, the steering column, intermediate shafts, pumps, hoses, fluids, belts and brake boosters and master cylinders can be eliminated from the vehicle. However, the challenge for automotive engineers is to use robust systems that are safe and reliable. Some question the wisdom of moving away from “time tested” mechanical and hydraulic systems to mechatronic systems that rely on chips and software that might fail. Anyone that has had their steering linkage or brakes go out in a moving car can appreciate these concerns. Therefore most car makers are looking at secondary systems such as parking brakes to prove this control concept before making wholesale changes to traditional components2.

Mechatronics and robotics

The robotics field continues to be a hotbed of development for mechatronic systems. Demonstrating their usefulness in non-industrial fields is the RoboBar, from Motoman. Handling bartender duties, this two-armed robot mixes and pours your favorite cocktail.

The robotics field has been the best showcase for mechatronics capabilities and benefits. Although it is not specifically mentioned on most mechatronics illustrations, some consider the robot to be the poster child for the Mechatronics Revolution in modern manufacturing. Robots have replaced humans in many dangerous and repetitive tasks. Along with other applied mechatronic approaches, they have allowed companies to stay in America and compete with inexpensive labor from other areas of the world.

Today mechatronics systems are mostly in factory automation and process automation fields. The advent of industrial computers, dedicated controllers and supervisor PCs have given operators “real-time” control and feedback. Variable speed drives on power conveyors move products along at high speeds. Smart sensors recognize patterns, and vision systems determine if a product is good or defective. Feedback is instantly transmitted to the control system to signal a servo-based actuator to remove the defective part. All production data are available to management regardless of where in the world these data were generated.

Industrial networks grew out of the merging of mechanisms and electronics. The idea of data moving through devices, machines, process and whole factories was initially developed by General Motors and Boeing in the early 1980s and resulted in the Machine Automation Protocol (MAP). The dream was to have a global standard that would interconnect all machines and processes, allowing “real-time” monitoring and control on the factory floor. MAP was not the answer, and some would say that a true open standard is yet to be achieved. However, several communication systems go a long way toward enabling effortless transmission of data, including Ethernet.

Some networks have grown from a focus on one finite task to encompass more functions. One example is Mechatrolink, inspired by Mechatronics itself. It is designed as a high-speed motion control interface between machine controllers, servos, and variable frequency drives. It has moved up in capabilities to work with third party controllers, sensors, I/O, and test and measurement devices.

The Future of “Mechatronics” Several development projects have the potential to spread the application of Mechatronics, with some key to unlocking the next generations of discoveries. One emerging area is Nanotechnology, also known as Microelectromechanical Systems (MEMS), “Nanoelectromechanical” Systems (NEMS), or “Nanomechatronics Systems” (NMS).

No longer bolted to the ground, the Honda ASIMO robot is humanoid in shape and can run. ASIMO is short for “Advanced Step in Innovative Mobility.”

In the medical field, these minuscule mechatronics wonders are being researched and tested to do what conventional medicine cannot do today. For example, science is investigating micro sized MEMs that can be injected into a person’s body to deliver treatment for an illness. Other engineers are investigating their use in fields from appliances to wind powered devices. The robotics field continues to be a hotbed of development for mechatronic systems. The use of such systems enables robots that are more human-like in appearance and capabilities. It first started with industrial robots tasked with doing non-industrial jobs, such as a bartender. Motoman’s RoboBar is an example. This two-armed robot mixes and pours your favorite cocktail.

Now, however, robots exist that are not bolted to the ground. Honda has ASIMO, short for “Advanced Step in Innovative Mobility.” ASIMO is unique in that it is humanoid in shape and can run. In the future, more robots will become part of our daily lives. As the current “baby boom” population ages, the need for assisted living and other care giving holds great promise for the future of consumer robotics.

Forecasting has its risks, but if trends stay true, mechatronics will play a big part in how we will do our work, enjoy our play, and heal our illness.

Yaskawa Electric – www.yaskawa.com

* www.honda.com
* www.motoman.com

Footnotes

* http://semico.com/mediacov/mar05_SemicoMediaCoverage_03-17-05_AZRepublic.pdf
* http://www.worldwidewords.org/turnsofphrase/tp-dri1.htm
* World Wide Words © Michael Quinion, 1996–2007.