Essentials of Mechatronics, Thesis of Mechatronics

Download Essentials of Mechatronics and more Thesis Mechatronics in PDF only on Docsity!

Essentials of

Mechatronics

Essentials of

Mechatronics

John Billingsley

University of Southern Queensland Queensland, Australia

A John Wiley & Sons, Inc., Publication

v

Contents

Preface ix

ix

Preface

There are many defi nitions of mechatronics, but most involve the concept of blending mechanisms, electronics, sensors, and control strategies into a design, knitted together with software. With an abundant wealth of topics to choose from, authors of mechatronics textbooks are tempted to bundle them all into a massive compendium. This book seeks to throw out all but the essentials; although perhaps in hanging onto the baby, some bathwater will still remain. There are a hundred ways of achieving all except the simplest of mecha- tronic design tasks. At every step, choice and compromise will be involved. Should a precision motor be used, or will a simple sensor and a sprinkle of feedback allow something cheaper and easier to do the trick? What does the end user ask for, really want, actually need—or eventually buy? Specialists can handle the fi ne detail, the composition of the molded plastic, the choice of components for the electronic interface, machining drawings, embedded computer, or software development platform. At the top of the pyramid, however, there must be a mechatronic designer capable of making the design tradeoffs that will transform a client’s demands or a bright idea into a successful commercial product. In some ways, mechatronics is as much a philosophy as a science. At its heart is a way of looking at tasks that will, if necessary, achieve their objective by ducking aside into an alternative technology. The mechatronic engineer knows where to look for the side roads and has a shrewd idea of the merits of the diversion.

xi

Acknowledgments

This book is the result of so many influences that there is a danger of this becoming the longest section. Perhaps I should start with the engineers of the autopilot industry who introduced me to the practical aspects of control system design. Laury Ambrose and Mike Skinner left me in no doubt as to their opinions of the quality of the servo loop designed with my new graduate academic skills. Later, John Coales fi lled me with enthusiasm to research abstruse control methods such as fast-model predictive control. My team of Cambridge researchers, including David Hedgeland, John Moughton, Matthew Dixon, and Roger Kinns, led the charge to embed processor boards in the most unlikely applications. In Portsmouth, life became even more exciting. Mechatronics and robotics abounded with the help of Harjit Singh, Fazel Naghdy, David Harrison, David Sanders, David Robinson, and many others. Arthur Collie lent the wisdom of years in industry to a passion for walking robots. Tim Dadd, now my son- in-law, joined me in meeting the problems of running a company that designed software for embedding in mass-produced appliances. Australia has been fun. Sam Cubero, Jason Stone, Matt Petty, Stuart McCarthy, Brad Schultz, and others all pushed robotics forward, while Mark Phythian has taken up the cudgels of running Micromouse and Bilby contests. Mark Dunn has thrown himself into vision research, with more practical applications than you can shake a stick at. The achievements and energy of my children Berry-Anne, Richard, and William have all helped to keep up my enthusiasm, while my wife Rosalind’s play-writing successes have sometimes diverted my time to thespian activities.

2 INTRODUCTION

The method is now commonly found in the laser printer. A spinning mirror scans a laser beam across the photosensitive fi lm, building up the image by rapid switching of the beam. Letter shapes are held in computer memory, and the entire mechanical design is simplified. I consider this tradeoff between mechanics, electronics, and computing power to be the guiding principle of mechatronics. The research team were soon knitting similar computers into a variety of real-time applications, including an “acoustic telescope” to build the signals from 14 microphones into an image of the source. Hydrofoils were simulated, violins were analyzed for their “Stradivarius-like qualities,” and music was synthesized. A display for a color television, novel in those days, depended on a minimum of electronics and a wealth of software. But computing power soon came in increasingly small packages. Texas Instruments had produced a single chip that could function as a pocket cal- culator. By the time I had moved from Cambridge to Portsmouth, Intel and Motorola were head-to-head with competing microprocessors. In Britain, the Microprocessor Awareness Project (MAP) triggered a deluge of applications—but only a small proportion of them deserve truly to be considered as mechatronics. Industrial fi rms were offered 2000 pounds’-worth of consultancy to con- sider how microprocessors could be added to their products. Some sharp operators made a killing, providing virtually identical reports to a diversity of clients. Others “brokered” projects to earnest academics. Printing machines sprouted boxes with twinkling LEDs (light-emitting diodes), wiring and relays patched on top of the “standard model.” In many cases it made the machines virtually unusable and impossible to maintain. Gradually, however, the concept percolated through that the computing aspect could be made fundamental to the operation of a machine. The mechanical precision and complexity could be traded off against electronics and computing power, just as in the case of the typesetter. One MAP project concerned the design of a clock for a domestic cooker. Not very romantic, perhaps, but the client’s choice of the primordial chip as used in the earliest pocket calculators made it a conundrum with attitude. It took several years and many generations of the product to persuade the company to adopt something simpler to program. The manufacturers of the original chip kept halving their price. The chips were supplied, mask-programmed, in batches of 10,000. That concentrated the mind wonderfully on making sure that the code was correct. But once we had weaned the company off the TMS1000, there was room in the chip’s memory not only for the job at hand but also for the next version we had in mind. One focus of our research was the Craftsman Robot. An energy regulator is the switching element behind the knob that allows the power of a cooking ring to be varied. During its manufacture, several adjustments have to be made. We used a Unimation Puma 560 robot to pick each regulator from a

A PERSONAL VIEW 3

tray and offer it to a test rig. Instead of acting as a simple “mover,” however, the Puma was equipped with a screwdriver to adjust the regulator when it was still held in its gripper. Of course, we could not resist taking the robot apart and analyzing its software and drive circuitry. Other industrial projects included marine autopilots and a flux-gate compass. But another interest would soon seize my attention. In 1979, planning started for holding the Euromicro conference in London. Lionel Thompson, the chairman, wanted an added showpiece, and his mind was on “The Amazing Micromouse Maze Contest” that had just been announced by IEEE Spectrum. I put my hand up to organize the contest. I then started to follow the news from the United States. Blows were nearly exchanged when the “dumb wall followers” sprinted through the maze from the entrance at one corner to the exit at the other, much faster than their brainier rivals. How could the rules be massaged to give brains the edge? Donald Michie, a guru of technical conundrums, was all for making the objectives more abstract, perhaps adding a cat to the fray. The solution lay in the opposite direction, to give the mouse builders more specific information that could be designed into the logic of their machines. Our maze was speci- fied as 16 × 16 squares, with the target at the center, not on the edge. In that way, paths could circle the center to form “moats” that no mere wall-follower could cross. A preliminary run was held in Portsmouth in July, with results that literally gave me nightmares. Of the six mice that competed, only one could make any attempt to follow a passageway, let alone fi nd the center. Japanese observers were there in force, cameras snapping away, and I was amazed that everyone seemed to enjoy the show. At the conference in September, 15 mice competed. A sleek machine from Lausanne should perhaps have won—but it expected more precision of the maze than the carpenters had provided and became lodged on a join in the boards of the base. The winner was a clanking contraption, cobbled together around a brilliant maze-solving algorithm that has remained relevant to this day. The contest went from strength to strength, held in Paris, Tampere, Madrid, and Copenhagen, but for these fi rst few years something struck me as strange. Not one of the winners was trained as an engineer. Great machines came from mathematicians, computer maintenance staff, and programmers for manufacturing industry, but engineers were notable by their absence. In 1985 I was invited to Tsukuba, to see what the Japanese had made of the contest. There were 200 contestants, but the champion, Idani, was not an engineer in the formal sense. Later that year we took the contest back to the United States—the Japanese funded the trip to put some life back into an old adversary. A future champion was unearthed in MIT—but he was not then an academic; he was part of the laboratory staff.

A PERSONAL VIEW 5

of thousands of dollars. A demand was swiftly seen for an interface between the GPS system and the actual steering of the tractor. The steering submodule that was a small part of the vision guidance system was just what was wanted. This time the price was set at several times the price of the entire original vision system, and sales were very good. With a new commercial partner, we will soon combine vision with a low- cost precision GPS technique that we have developed. The project will be rolling again. Another project with journalist appeal was Robocow—a nimble mobile robot for training horses for cutting contests. In some ways, as technology advances the task of exploiting it becomes harder. The traditional approach to embedding some computing power was to take a microprocessor chip, add some supporting memory and interfaces, and then write the software “from the ground up.” The concept of an “operat- ing system” would be as alien as adding antilock braking to a rollerskate. But when Webcams can be bought with drivers to interface them via DirectShow to Windows-based applications, how far up the evolutionary tree do you have to go to fi nd your computing power? The price of a fully equipped PC card is today little more than that of an evaluation board for a Motorola HC12. Are we locked into complicated but popular technology “because it’s there”? That is certainly the line we have been taking with a deluge of agri- cultural application opportunities. The data capture is quick and dirty, and we can concentrate on innovating ways to analyze it. A project that appears strange—but actually makes good sense—is based on the ability to discriminate between animal species. When a sheep approaches a watering place, it is recognized and allowed to pass through a gate. When a feral pig comes the same way, it is also recognized and allowed to pass through an adjacent gateway, to another water source. The difference is that the sheep will be allowed to go on its way after drinking, while the pig is confi ned until the farmer comes to pay it some serious attention. The economics of damage by feral pigs and the trade in feral pork are convincing reasons for funding the project. The dynamic behavior of small marsupials is another area of interest. There is a breeding program for an endangered species of sminthopsis. The problem is that if the lady is not “in the mood,” the animals are apt to kill each other. By tracking the movement of separated partners in adjoining cages, we hope to detect in real time when true love can take its course. Texture analysis is usually a lengthy business, requiring substantial com- puting effort for correlations. Two applications require a speedy solution. The fi rst is for the grading of oranges, where the extent of “goose bumps” on the surface is an indicator of quality. The second is for the game of football. A speedy analysis of the status of the grass cover must be made, at least to avoid a lawsuit when an overvalued player slips on a bare patch and falls on his fundament. But is this really mechatronics?

6 INTRODUCTION

So, what of the next generation of mechatronic engineers? How do we give them skill and ability with the essentials, without deluging them with the entire contents of the textbooks of at least three diverse disciplines? The Micromouse experience suggests that hands-on experimentation is an essen- tial ingredient. While learning, software must be “crafted” by the student, rather than being ladled into the project as a bought-in commodity. The student must be prepared to deal with hydraulics or electromechanics, treat- ing them as two sides of the same coin. After the “bare essentials” whistle-stop tour of mechatronics, some experi- ments are presented that could whet the appetites of students to study the more detailed material that follows. “Seat of the pants” engineering will cer- tainly get you started, but will go only so far. Mechatronics is special. It is no more a mere mixture of electronics, mechanics, and computing than a Chateau Latour (or Grange Hermitage ) vintage wine is a mixture of yeast and grape juice.

1.2 WHAT IS AND IS NOT MECHATRONICS?

Long ago, Caryl Capek wrote a book, Rossum’s Universal Robots. It was as little about robotics as Animal Farm was about agriculture, but the term had been coined. Science fiction writers grew fat on the theme, and the idea of mechanical slave workers was lodged in the mind of the public. When Devol designed a mechanical manipulator for Engelberger’s fi rm, Unimation, it was endowed with the term “a robot arm.” As a research topic, robotics ceased to be about tin men and turned to the articulation of mechani- cal joints to move a gripper or workpiece to a precise set of coordinates. The new “three laws of robotics” concerned the Denavit–Hartenberg transforma- tion matrices, discrete-time control algorithms, and precision sensors. Robotics is just a narrow subset of mechatronics. It is true that it has all the ingredients of sensing, actuation, and a quantity of computer-assisted strategy in between, but with every day the list of mechatronic products increases. In videorecorders, DVD players, jet airliners, fuel injection motor engines, advanced sewing machines, and Mars rovers, not to mention all the gadgetry that surrounds a computer, the jigsaw pieces of mechatronics are slotted together. In something as simple as a thermostat, sensing and actuation of the heater are linked. But the element of computation is missing. It is not mechatronic. In automatic sliding doors, however, the criterion is not as cut and dried. A few simple logic circuits are enough to link the passive infrared sensor to the door motor, but the designer might have found that the alternative of embedding a microprocessor was in fact simpler to design and cheaper to construct. Before 1960, autopilots were capable of automatic landing. Their compu- tational processes were based on magnetic amplifi ers , circuits using the satu-

8 INTRODUCTION

A hairdryer marketed some years ago featured a “bonnet,” coupled by a hose to the hot-air unit. A plastic knob could be rotated to give continuously variable temperature control. So, how would you go about designing it? When the question is put to university classes, it always brings answers featuring potentiometers, thyristor power controllers, and often a microcomputer. The product was actually much simpler. The airflow was divided into two paths after the fan. In one path was a heating element, regulated by a simple thermostat just “downstream,” while the other simply blew cold air. The ornate knob moved a shutter that closed off one or other flow, or allowed a variable mixture of the two. Good design can often demand an awareness of how to avoid excessive technology.

9

The Bare Essentials

2.1 ACTUATORS

A mechatronic system must “do” something, even if it is just to move a pointer or change a display. The industrial robot must have motors with which to move an end effector, perhaps a gripper, while another system’s output might concern heaters. The mechatronic engineer should not be in too much of a hurry to run to the catalog to choose an electric motor. To the electrical engineer, motors are a fascinating playground around which to debate the merits and challenges of axial flux, windage losses, rotor resistance, or commutation. The mecha- tronic engineer is by no means certain that the solution does not instead lie with something hydraulic or pneumatic. This section attempts to put a selection of the vast range of actuators into some sort of perspective.

2.1.1 Choosing a Technology

The fi rst question to ask is: “What must the output do?” At the bottom end of the list, in terms of power, is the task of displaying a value on an indicator. Many automobile instrument panels have now been taken over by liquid crystal displays, probably putting them outside the grasp of mechatronics, but they are just the tip of the iceberg.

Essentials of Mechatronics , by John Billingsley Copyright © 2006 John Wiley & Sons, Inc.

ACTUATORS 11

Think of it in terms of a compass needle being pulled into line by a pair of coils, arranged north–south and east–west (see Fig. 2.1). Current can be passed through these coils in either direction, so we might start with both the NS and EW coils being driven in the “positive” direction, resulting in the needle pointing northeast. Now if we reverse the drive the NS coil, the needle will move to point southeast. Reverse the EW coil, and it will rotate to point southwest. Make the drive to the NS winding positive again, and the needle moves on to point northwest. Finally reverse the EW drive to be positive and the needle completes the circle to point northeast once again. You can see an animation of this at www.essmech.com/2/1/3.htm In practice, the magnet of a stepper motor has a large number of poles, and the windings are helped by a similar large number of salient polepieces (Fig. 2.2) in the soft iron on which they are wound. As a result, the switching sequence must be repeated 50 times for a “200-step” motor to make one complete revolution.

N

S

W E Figure 2.1 Stepper schematic—NSEW.

N

N

N

S

S

S

Figure 2.2 Stepper schematic—polepieces.

12 THE BARE ESSENTIALS

Simple software can command the motor to move to a desired position, so the stepper motor has great appeal for the amateur robotics builder. But it has a great number of shortcomings. There is a limit to the torque it can resist before it “clunks” out of the desired position and rotates to a different stable location. If a transient of excessive torque causes it to “drop out of step”, then, without a separate position transducer, the slip goes unnoticed by the proces- sor and the error remains uncorrected. What is more, this dropout torque decreases markedly with speed. An attempt to accelerate the motor too rapidly can be disastrous and the software is made more complex by the need to ramp the speedup gently. Of course, there are other ways than the use of a permanent magnet for producing a magnetic field. More powerful DC motors, such as automobile starter motors, use current in a field winding to generate the stator’s magnetic field. Similar motors are not restricted to using direct current. By connecting the stator and rotor windings in series, the torque will be in the same sense whether positive or negative voltage is applied across it. The motor can be driven by either an AC or DC voltage. This is the universal motor (Fig. 2.3), to be found in vacuum cleaners and a host of other domestic gadgets.

Field Armature

Figure 2.3 Universal motor.

2.1.4 AC Motors

Another family of motors depend on alternating current for their fundamen- tal mode of operation. They use rotating fields. If the stator has two sets of windings at right angles and if a sine-wave current flows in one winding and a cosine-wave current flows in the other, then the result is a magnetic field that rotates at the supply frequency. This is illustrated at www.essmech.com/2/1/4.htm.