ee301-assignment 4

What is the effect of adding a zero to a control system?
Consider second order closed loop transfer function of the form given by
C(s)/R(s) =ωn2/ s2 +2ζωns+ωn2
ζ=damping factor(or damping ratio)
ωn =undamped natural frequency

Let a zero at s=-z be added to this transfer function .Then we have
C(s)/R(s)= [(s+z)/z]ωn2 / s2 +2ζωns+ωn2 →→→→(1)
Note that multiplication term in numerator of above expression has been adjusted so that steady state gain C/R(0) of system is unity .This gives steady state value of output Css=1 When input is unit step.Thus the system will track the step input with zero steady error .

From equation we have
C(s)/R(s)= ωn2/ s2 +2ζωns+ωn2 +s/z{ωn2/ s2 +2ζωns+ωn2 }
Let Cz(t) be the response of the system with a zero at s=-z.Then from the above equation we have
Cz(t)=c(t) + 1/z ( dc(t)/dt) (a)

The effect of added derivative term may be seen by examining fig (1) where a case for a typical value of ζ(less than one )is considered.We dee from this figure that the effect of zero is to contribute a pronounced early peak to the system’s response wherby peak overshoot may increase appreciably.From equation(a) and fig (1)it is seen that smaller the value of z ie closer is the value of zero to the origin,the more pronounced is the peaking phenomenon.On account of this fact the zeros of real axis near the origin are generally avoided in design.However ,in asluggish system artful introduction of a zero at the proper position can improve the transient response.

We further observe from equation(a)that as z increases ie zero moves farther into left half of s plane ,its effect become less pronounced.For sufficiently large values of z,the effect of zero on transient response may become negligible.
For closed loop transfer function of eq(1) peak percentage overshoot for a unit step input can be read from log –log graphs of fig (1) as function of z/ζωn for various values of ζ<=1

Reference: (1)Control Systems Engineering-I.J Nagarath,M.Gopal

(2)Control Systems Engineering-S.N Sivanandam

ee301 assignment 3

What do the poles and zeros contribute in the control system?
The open loop transfer function of a general linear feedback system can be written as
G(s)H(s)=k(s+x 1)(s+x2)…………/(s+p1)(s+p2)……..→→→→→→→→→→→→(1)
Where - x 1 ,-x 2, -x 3…………..are the zeros and -p1 ,-p2 ,-p3………..are the poles of open-loop transfer function.
The characterstic equation in terms of open loop transfer function is formed as
1+G(s)H(s)= 1 + k(s+x 1)(s+x2)…………/(s+p1)(s+p2)……..
= (s+p1)(s+p2 )…k(s+x1)(s+x2)…/(s+p1)(s+p2)….→→→→→→→(2)
Equation (2) can also be written as
1+G(s)H(s)= k(s+z 1)(s+z2)………/(s+p1)(s+p2)……..→→→→→→→→→(3)
The roots of charactersatic equation are obtained from the expression
1+G(s)H(s)=0
Hence ,for this condition equation (3)yields that
(s+z 1)(s+z2)……=0
These roots are (-z1 ,-z2 ,-z3 .........................) called as zeros of the characterstic equation .It canbe observed that the poles of open loop transfer function are also the poles of characterstic equation (2)
Criterion for stability:The linear feedback control system represented by characterstic equation (1+G(s)H(s))will be stable only if zeros - z1 ,-z2 ,-z3……………….lie in left half of s plane ie roots of the characterstic equation must contain negative real part.This indicates that no zero of charactrstic equation should lie in right half of the s plane for the system to be stable


Reference:
(1)Control Systems Engineering-I.J Nagarath,M.Gopal
(2)Control Systems Engineering-S.N Sivanandam

ee301 assignment 2

What are incremental encoders ?Are they useful to us in any way?
Increment encoders are optical encoders that are frequently used in control systems to convert linear or rotory displacement into digital code or pulse signals. Their output is a pulse for each increment of resolution but these make no distinction between increments.
An incremental encoder typically has four parts:a light source(LED),a rotory (or translatory) disc ,a stationary mask and a sensor (photodiode) .The disc has alternate opaque and transparent sectors which are etched by means of photographic process on to a plastic disc.(slots are cut out in case a metal disc is used).As disc rotates during half of the increment cycle the transparent sectors of rotating and stationary discs come in alignment permitting the light from LED to reach the sensor thereby generating an electric pulse .
The waveform of the sensor output of an encoder is generally triangular or sinusoidal depending upon the resolution required.Square wave signal compatible with digital logic are obtained from it by means of linear OPAM and comparator.The resolution of incremental encoders are given by

Basic resolution =360/N where N=number of sectors of disc:each sector is half transparent and half opaque.

Reference:
(1)Control Systems Engineering-I.J Nagarath,M.Gopal
(2)Control Systems Engineering-S.N Sivanandam

ee301 assignment 1

What is synchro?Is it relalated in any way to stepper motor?
Synchros are electromagnetic devices which are used to transmit angular displacement ,velocity e.t.c.It is commercially known as selsyn or autosyn.The types of synchros according to operations are given below:
1.synchro transmitter: This is an electromechanical unit in which the input ,the mechanical signal is given to the rotor is converted into an electrical signal in the stator.
2.Synchro receiver.:This is also known as synchro control transformer,in which the input electrical signal given to the stator is converted to the mechanical signal[angular displacement] in the rotor.
3.Synchro differential transmitter and differential receiver:The differential transmitter is an electro mechanical unit in which the rotor is driven mechanically to modify the received signal and to transmit an electrical signal which is the sum of difference of impressed signal and modified signal.

There are two distinctly different ways of using stepper motors in control systems.One is the open loop mode and other is the closed loop mode.
The stepper motor is a digital device whose output in shaft angular displacement is completely determined by the number of input pulses.Consequently,there is no need for a feedback device to determine the position of motor shaft and ,therefore,of the load connected to the motor shaft.This means that an open –loop step servo system can be designed to yield the the same accuracy as that of a closed loop analog system.figure(1) shows use of stepper motor in open loop mode. Use of steppermotor in closedloop mode requires synchros as error detector.Here the motor is used like conventional servomotor.A signal from the output is fed back and is used to operate a gate controlling the pulses from a pulse generator.This is shown in the figure below:

Reference:
(1)Control Systems Engineering-I.J Nagarath,M.Gopal
(2)Control Systems Engineering-S.N Sivanandam

T3-776 Robot

The Design and Development of a Control System for a T3-776
RobotJoseph W. Geisinger
Research Associate :Martin P. Aalund
Dr. Delbert Tesar
Carol Cockrell Chair in Engineering
Department of Mechanical Engineering
The University of Texas at AustinAustin, Texas 78713
Abstract - A robot control system was designed, developed andtested at The University of Texas at Austin for a six degree offreedom (T976) industrial robot. The new robotic controlsystem can be divided into two major subsections; the servoamplifiers and the system controller. The following articlediscusses the improvements in performance, -me of theimplementation difficulties, and the lessons learned during theimplementation of the new control system. The origiial systemconsisted of an inflexible system controller and SCR basedpower ampliiers. The new system controller has a 32 bit highspeedp " r that can be easily expanded, reprogrammed,and interfaced to other computers. The servo amplifiers use aPWM methodology to control an €I-bridge which utilizes stateof the art Insulated Gate Bi-polar Transistor (IGBT) powermodules, and is capable of both current and velocity control.

INTRODUCTION
A typical industrial robotic controller consists of twocontroller levels, a system level controller and servo loopcontrollers. The system level controller calculates the setpoints for each servo loop controller. Various types ofcontrol algorithms can be used to generate the set pointssuch as position, velocity or torque. The servo loopcontrollers perform low level control on each of the robot'sactuators. A 6 degree of freedom robot requires 6 servocontrollers.
The University of Texas at Austin Robotics ResearchGroup is engaged in research which demonstrates that anindustrial robot can be utilized in precision light machiningwithout jigs [l] and for other high value added processes.The robot used by the Robotics Research Group is aCincinnati Milacfon Inc. (CMI) T3-776 heavy dutyindustrial robot. The robot's factory controller consists of aCMI ACRAMATIC version 4 system controller and CMISilicon Controlled Rectifier (SCR) servo loop controllers 'Ws work was qqmcted in pnrt by the State dTexas un&r ATPGrant No.4679 and U. S. Depaament d Energy under Grant No. DE--86NE379966 which are manufactured by Siemens. It became apparentthat the existing system controller of the T3-776 was notcapable of executing complex control strategies.Most current industrial robot system controllers aredesigned for a particular machine. Each brand and, in somecases, model of robot has a different controller architecture.Thus control algorithms designed for one robot cannotnecessarily be ported to anothex. The Robotics ResearchGroup decided to replace the system controller with a highspeedgeneric controller which can be interfad to almostany robot with minimal changes to the system architecture.The new system controller, the Multi-Channel RoboticController (MCRC), by-passed the ACRAMATIC controllerby interfacing to the robot's sensors and outputting thecommand set points directly U, the Semen servo loopControllers. After the initial upgrade of the systemcontroller, it became apparent that the servo controllers werea major bottleneck in the robot's performance. Theremaining sections in this paper will discuss the new systemcontroller, the new servo controllers, and the robot'sperformance enhancements due to these new controllers.

SYSTEM CONTROLLER
The Multi-Channel Robotic Controller (MCRC) isdesigned to control a standard industrial robot insuring highquality coordinated control of all axes. It allows for futureexpansion to control multiple robots or other machines. Thearchitecture is designed around a VME bus with NationalSemiconductor 32532 CPUs equipped with Weitek 3164floating-point units (see Fig 1). Each processor unit is ratedwith a peak performance of 15 million instructions persecond (MIPS) and 10 million floating-point operations persecond (MFLOPS). This produces a powerful distributedprocessing system with high bandwidth interprocessorcommunication and YO capabilities. The use of a VRTXoperating system kernel results in a powerful, flexible andexpandable machine controller. The fmt application of thiscontroller is to replace the logic side of a CMI controller inorder to apply advanced control algorithms too demandingfor the original controller.

BACKGROUND
Work on the MCRC was initiated in association withthe Precision Light Machining project at The University ofTexas at Austin. The goal of this project is to use anindustrial robot, a CMI T3-776, to perform precision, lightduty machining without templates. The T3-776 robot is alarge, 6 degree-of-freedom industrial robot capable of lifting150 lbs with a repeatability of 0.01 inches. The Originalcontroller (a CMI ACRAMATIC version 4) employs amultibus card cage with an Intel 8086 CPU as bus master.Communication with peripheral devices is accomplishedwith a serial communication board and digital YO strips. Itbecame apparent that there were limitations with thiscontroller that would create problems when we tried toimplement more advanced control algorithms. A systemwith both higher computational performance and YObandwidth is required to produce tracking of sufficient quality.
It was for the above reasons as well as the inflexibilityof the ACRAMATIC control program, its low processorperformance and poor communications, that the logic side ofthe T3-776 was replaced. A set of minimum requirements tosuccessfully meet the goals of this project and futureprojects was developed. The requirements were thenmodified so that the controller would be expandable andflexible enough to be interfaced to different robots in thefuture with a minimum of effort.

MCRC Design and Features
The MCRC was required to be compatible with the T3-776's hardware. Furthermore, the controller was required toinitially be connected to the CMI T3-776, leaving the robot'sservo controllers intact. The controller can be quicklyswitched back to the CMI controller, and this has beenaccomplished on several occasions in under 5 minutes. Theprocedure requires swapping 7 plugs, and turning 2switches. The MCRC design allows for easy migration toother industrial robotic systems. This was accomplished byusing a filter box to scale the tachometers to acceptablevalues, and a resolvers to digital box to convert the resolversinto a digital representation of the angles. Interfacing toother robots would require modification of thesecomponents. For example, if an incremental encoder inputis required, an external counting circuit to make theencoders appear as absolute encoders can be used, and theresulting signal fed into the parallel VO board. Absoluteencoders can be interfaced directly. For robots withdifferent tachometer constants, a filter box with a differentscaling factor can be built. The tachometer signals can besampled at a resolution of 16 bits and a speed up to 4OOO Hzcontinuously using a swapped buffer mode of the A/Dboard. The output command to the amplifiers is also 16 bitsand has a settling time of 16 p-sec. The requirement forposition measurement specified, that all axes should be readto a minimum of sixteen bits. This requirement wascompletely satisfied. In fact, the base, shoulder, and elbowhave a resolution of better than 22-bits per axis revolution,and the wrist axes have a resolution of 17-bits per motor turn.
Several control routines were implemented on theMCRC for general path following. The routines used theCMI amplifier section in both its closed and open velocityloop modes, and the new servo controllers in current andvelocity modes. The paths were downloaded to thecontroller from files residing on a 386 compatibles'harddkk. One of the most beneficial features of the MCRCis its ability to record states during a process. The states arestored in memory on the MCRC in real time and thendownloaded to a 386 for analysis after the process. The datasets in this paper were collected by this procedure.The focus of this work was to design and implement asystem controller that can be interfaced to multiple robots[2]. The results of this work will be used to develop future system controllers capable of interfacing to a generalreconfigurable modular mechanical structure. A set ofminimum performance standards was developed, and thecontroller was designed to meet those criteria. Thecontroller was then assembled and interfaced to an existingindustrial robot, a Cincinnati Milacron T3-776. Thisinvolved, among other things, the design andimplementation of boards to convert resolver signals intodigital representations that can be utilized by the controller,the cabling between robot and controller, the writing of lowleveldevice drivers, and the building of anti-aliasing filters.Control code was implemented in C on the system controllerfor a general 6 DOF arm. The current MCRC can be used tocontrol robots that very from 1 to 6 DOFs. Future systemcontrollers will be able to scale in both hardware andsoftware to match the mechanical structure.

SERVO CONTROLLERS
The CMI servo amplifiers manufactured by Siemens are3-phase, four quadrant Silicon Controlled Rectifier (SCR)controllers. The CMI amplifiers are divided into two driveunits: one for the lower three axes and one for the three wristaxes. Each drive unit contains three axis controller boards, apower supply board, a pulse transfonner board, and an SCRamplifier board.
The axis controller boards perform the local control foreach motor. The controller utilizes a velocity Proportional-Integral (PI) control loop which has a low and high speedproportional gain. They can operate either in velocity oropen control mode. In velocity control mode, the input fromthe system controller is a velocity command, and in opencontrol mode, the input from the system controller is theSCR fving command. Additional features include speedcurrent limiting, overcurrent protection, and overspeeddetection. The fuing angle, a, for the SCRs ranges from 30"to 150' of a cycle for each of the 6 SCRs. When thecontroller has a zero velocity command, the SCRs arealtemately fired in opposite directions at a retarded firingangle. This method, called crossfire, enhances the responsetime of the system. When the system dynamically brakes themotor, the energy stored in the motor's windings is returnedback into the main transmission lines. The pulsetransformer board is used to electrically isolate the axiscontroller board from the high power of rhe SCR board. Tomake the SCR switching amplifier a 4-quadrant controller,the motor return is fed into the neutral of a 3-Phasetransformer wired in a wye configuration. The inputs to therectifiers are the three legs of the wye. The two output legsof the 3-Phase rectifier bridge are tied together by a centertapped inductor. This allows the switching amplifier toproduce bi-directional current and voltage necessary for fourquadsant operation.
Evaluation of the T3-776 after replacing the systemcontroller determined that the robot's performance wassubstantially degraded due to limitations in the CMIamplifiers. The SCR switching amplifiers created 180 Hzharmonics on their motor outputs. Since the motorsbandwidth is above this frequency, it resulted in unwantedharmonics superimposed on the desired motions. In addition,the servo controllers do not contain current feedback loopsthat are necessary for f d m q u e control. Also, the EactoryCMI servo controllers did not allow for single axis controlwhich would allow us to study the robot's pafameters for aparticular axis.
The design objectives of the new CMI T3-776 poweramplifiers were to overcome the limitations of the CMIpower amplifiers. In addition, it was desirable to keep thepower connections to the robot the same as the CMI T3-776factory power amplifiers. The removal of harmonicsintroduced from the controlled rectifier circuit allows forsmoother motion at low speeds. The current control loop isnecessary for proper fdtorque control at the joint level,and it also allows for quicker responses to velocity commandinputs. The single axis control aids in obtaining local axisparameters and testing new control schemes. By keeping theidentid power connections to tbe robot, system performanceenhancement of the new Controllers is easily compared to theCMI factory amplifiers.
The system features can be divided into two categories;safety and design. The system safety features include Dcbus overvoltage protection, capacitor bleeder resistors, safestart-up, and safe shut down. Safety features for individualaxes include current limits, overspeed limits, individual gatepower supply undervoltage protection, and transistor shortcircuit detection. The DC bus overvoltage protection circuitprotects the high power electronics that are attached to theDC bus. The capacitor bleeder resistors drain the large DCbus capacitors to insure tbe system is safe to work on aftershutdown. The safe startup insures that the user does notbring up the robot with the brakes disabled or drives enabled.The trip time far a current limit is set exponentially by aresistor-capacitor network. The velocity limit is aninstantaneous limit that trips when the velocity surpasses aset maximum. Undervoltage gate power supply protectioncircuits protect the power transistors from operating in thelinear region which will cause unnecessary heating of thetransistors. The short circuit protection circuit is built intotbe power transistor gate driver circuits.The servo loop controller implements PI controltechniques in the current and velocity feedback loops. Forfordtorque control, the velocity loop can be switched out ofthe control loop structure. The velocity and current analogsignals are available to the system controller independent ofwhich control loop structure is active. The switchingamplifier topology used for controlling the motors is achopper-wired H-Bridge configuration. The voltage output of the H-bridge is controlled by Pulse Width Modulation(PWM). The transistors used in the H-Bridge are InsulatedGate Bi-polar Transistors (IGBTs). The IGBTs currentlyswitch at 16 kHz; however, the controller is designed toswitch at frequencies up to 20 kHz. The amplifier's output israted at 160 volts 40 amps continuous for the lower threeaxes, and 160 25 amps continuous for the upper three axes.The servo control system also allows for individual axiscontrol. When a velocity or current limit has occurred theaxis number is returned to the system controller for errordetection purposes.
The total cost for the new servo controllers was just over%4,OOO, or about $700 per axis. The estimated value for thesystem is $5,900. The major difference in these two figuresis the cost of the donated cabinet in which the componentswere mounted and the cost of a 20 kVA transformer used forstepping down the utility line voltage from 480V to 120V.Currently, the transformer used by the CMI amplifiers is alsoused for the new controllers. To gain a perspective on thecost of this system in relation to a commercial system, thecost of an Inland amplifier designed with basically the samefeatures as the new system, but rated at 14OV, 1OAcontinuous current, will cost about $1200 without thetransformer, or about twice the cost of the new controller.The cost of the new controller does not incorporate theengineering cost that was needed to build the system andthus a one-to-one cost comparison is not feasible.

Component Design
The power amplifiers designed for the CMI T3-776 canbe broken down into the four sections; a power interface, asensor interface, a system controller interface, and closedloop controller. The sensor interface and closed loopcontrollers are both contained on the axis controller board.Fig. 2 shows the basic layout of one servo controller, and theinterfaces between each section of the controller.The power interface section contains the brake andcontrol relay logic, main power rectifier and conditioner,Overvoltage Protection (OVP), the switching amplifiers, andthe control power supplies. The main power rectifm andconditioner circuit implements a soft start strategy to protectthe three phase rectifiers from the high "in rush" of currentat the initial charging of the capacitor. The OVP circuitprotects the system from excess voltage that builds up f"energy returned to the capacitor bank due to dynamicbraking of the motors and the switching of the amplifiertransistors (see Fig 3). The switching amplifier uses twostate PWM to control the output voltage. The switchingamplifier is designed to function up to 20 kHz before theIGBT switching losses become significant. An amplifier'speak switching frequency is limited by either the transistorswitching losses or the iron losses in the motor. In the caseof the CMI robot, the factor limiting the switching frequencyis the losses in the iron of the motors. At the initial designstage, the factory was consulted to determine if the motorcould operate at frequencies up to 20 kHz. Kollmorgen Inc.,the manufacture of the T3-776 motors, believed the motorscould operate at these frequencies but tests have shown thatthere is considerable heating in the windings of axes 2 and 3for a frequency of 16 kHz. A switching frequency thatminimizes the heating in the windings of axes 2 and 3 hasyet to be determined due to insufficient motor information.The switching amplifiers use 20V floating gate powersupplies to drive the transistors. The 20V power supplies areelectrically isolated from each other and the rest of the system.
The MCRC communicates to the axis controller systemthrough the system interface board. All power, with theexception of the gate driver power, for the axis controllerboard goes through the system interface board. The systeminterface board also resets the axis controller boards when acurrent or velocity limit is hipped and when the system isinitialized. When a velocity or current limit occurs, thesystem interface board en& the axis number for the system controller, and promptly deactivatesthe amplifiers. = The block diagram of an actuatorcontroller board is shown in Fig. 4. The axiscontroller gains for the current and velocityloops are set by a capacitor and apotentiometer. The board also featuressignal filters, limit circuitry, and two statePWM controller for the H-bridge amplifier.The filters used on the feedback signals aresecond order Bessel filters. Bessel filterswere chosen because they minimize the~hasesh ift of the signals over the fquenciesaif interest. The designed cutaff Gipencyfor the current feedback loops is 10 LHZ andthe designed cut-off frequency for thevelocity filters is 500 Hz. The PWM driver circuitrycontains a comparator, a triangular wave generator, gatelock-out circuitry, undervoltage gate protection, IGBT gatedrivers, and IGBT short circuit detection.

System Analysis
Two items will be discussed in this section. The first isthe overall packaging of the components into a cabinet. Thesecond topic will concem the design of the start-upThe servo control system was packaged in a cabinetdonated by Motorola, Inc. The cabinet is 22 inches wide by22 inches deep by 44 inches tall. The entire system iscontained inside the cabinet except for a control panel,(situated on top of the cabinet), and a remote enable box.The system is designed such that the ambient air enters atthe bottom of the cabinet and circulates over the controllerelectronics before it passes over the switching amplifiers andtheir heat sinks, and exits the cabinet. This allows the coolair to pass over the temperature sensitive electroniccomponents before passing over the power amplifien whichcan operate at a higher temperture.
The inside of the cabinet can be divided into four layers.Layer one contains the main power mMiers andconditioners for the base and wrist axes, the 48OVAC to12OVAC step down transformer, main power up relays, andthree terminal strips. Layer hkro contains all of the controlpower supplies. Layer three contains some of the controlrelay logic, all of the brake relay logic and brake powercircuitry, the system interface board, and the axis controllerboards. The axis controller boards slip into a modifiedMulti-Bus card cage. Layer four contains all of theswitching amplifier circuitry.
The system interfaces to the robot through receptacleson the back panel. The main disconnect is also on the backpanel. The main disconnect breaks the main power lineswhen the system draws more than thirty amps from theprocedure.
Fig. 4. Block diagram of the actuaIcs controna board.utility line. The interface is designed such that the pin-outsof the receptacles are exactly the same as the pin out on theCMI factory controller.
The system start-up procedure is designed such thatwhen a detectable fault occurs anywhere in the system, thesystem is shut completely down. The pracedure is alsodesigned such that the system cannot be brought up out ofsequence for safety reasons. The pracedure involves fmtcharging up the main JX bus systems for the wrist and baseaxes. Then, after the system is charged and the componentsreach a steady-state temperature to account for any drift dueto temperature, the brakes may be released and amplifiersenabled.

System Testing and Performance Analysis
Testing was first peaformed on each of the componentsto insure proper operation. The system testing began withdebugging the wrist controllers until they worked properly.Then the design made to the wrist axis controllers weremade to the base axis controllers. After making theappropriate modifications to base WO controllers, theywere functionally debugged. Finally, the entire system wasbrought up. The new actuator controller system was thencompared to the CMI servo controller system to determine ifthe original design objectives had been met.
The order in which the system components were testedfollowed the order in which the system was assembled. Thecomponents were tested as thoroughly as possible beforebeing integrated with other components in the system. Thesystem testing was a more difficult task. Many problemswere found in the system at this point. Two of the largestcontributors to the system problems were ground loops andElectromagnetic Interference (EMI). The ground loopsproved to be an easier problem to solve than the EM1because they were measurable and repeatable. The EMI, onthe other hand, was very unpredictable and often affectedother components unexpectedly.
AX66: IO A M P ~ S I ' E P_. D 0.b i i ..io 4 rir U oh U o.ia ob'=(=4Fig. 5. Axis 6 full current step.
The following precautions taken to alleviate the effectsof the EM:
00Twist wires carrying current to minimizeinductance;
Run the DC bus bars as close together as possible;
Shield the outside of the cabinet with a Ferromagneticmaterial so Eh41 leakage meets federalstandards;
0 Minimize the amount the large current carryingcable;
and,Isolate the large relays from components sensitiveto EMI.
The steps to eliminate the problems associated with theground loops were:
Isolate the digital relay returns from thedigital electronics returns;
Separate the signal grounds as much aspossible to eliminate crosstalk;
Shield all analog signals at one endonly;
AXB4 5 O X O F R R L ~ S l "
Fig. 7. Velocity step profiles using both contrdlas.-1100=w
Fig. 6. Velocity step profilea using both contrdlasSeparate all digital and analog grounds,and bring them together at onecommon point, prefembly where theground is brought into the board or atthe power supply; and,Ground all components such as thecabinet, power supplies, etc.
The system performance analysis consisted of two parts:
a current step response test, and a velocity step response test
The current response test was only performed on the newsystem since the CMI Conmller does not contain a currentloop.
Fig. 5 depicts a typical current response for a fullcurrent step on axis 6. The rise and fall times of the currentwaveforms were approximately 1 ms. The current stepresponse also demonsrrates the current tailing when the backelectromotive force approaches the driving voltage across themotor tenninals.
The velocity step test was performed on each of therobot's axes using both controllers. Fig. 6 and 7 show typicalresponses for a step of 50% of full scale on axes 3 and 4.Fig. 8 shows the results of stopping from half of the
Am4 M H I X o F R R L ~ ~ ~
-4 ' 4 u
b 4 4 w . i r . n - U ril rir
Fig. 8. Axis 4 braking waveform far both controllersu&ilLlh#-
Table 2 Velocity step data far 50% of full velocitymaximum velocity for axis 4. As seen from this figure, thenew controller clearly reacts at a much higher rate than theCMI factory controller. The complete results of the velocitystep tests for all the axes are presented in Table 1. The newservo system had at least one and a half times faster rise andfall times than the CMI controller for all axes, and was abauttwo times faster for all axes except axis 2. The newcontroller reduced the delay time between changes in thecommand signal and the actuator controller response by afactor of three on all axes. The new system can reactconsiderably faster to extemal disturbances than the CMIcontroller. The tests also showed that the new controllereliminates the 180 Hz pulsations which are present when theCMI controller is used (as shown in Fig. 5-8).

CONCLUSION
Several major accomplishments have been achieved.The MCRC was used to successfully control the T3-776Robot in a variety of demonstrations, and using severaldifferent control algorithms including compliance in themachining demonstration. Additional safety features thatwere implemented in both the system and servo controllerswere successfully tested, and still allowed for a highsampling rate of between 2 and 5 milli-seconds, TheMCRC was tested with both the original SCR amplifiers andthe new PWM amplifiers. The interchange between the twoamps requires only 5 minutes. This demonstrates how, withlittle modification, the controller can be moved betweenrobots. The MCRC proved very useful in the debugging ofthe new servo controllers. Its flexibility allowed theindividual axes to be tested in a variety of ways, and thedata to be downloaded and stored.
One of the largest improvements of the new servocontrollers over the CMI factory controllers arose from thedesign of the switching amplifier. The switching amplifiertopology chosen for the new controllers was the chopperwired in an H-bridge configuration. The H-bridge iscontrolled by a two state PWM methodology which switchesat approximately 16 kHz. The power transistors used in theH-bridge are IGBTs which are present state of the art technology.
Many lessons were learned from this work. Particularlythe problems associated with EM1 and ground loops both atthe system and servo level. By considering possibleinterference sources in the initial design specifications agreat deal of debugging can be eliminated. Only so muchimprovement can be made to an existing robot before thelimitations in the robot, its motors, and its sensors preventany further improvements.

REFFERENCE
J. Wander, J. Hudgens, M. A . l d D. T m "Recision Routing By AnIndusirial Robot Using Ddlcdioo tlaqemation," Proceedings ofM. Aalund., D. Tesar, J. Wander, "Multi-Qllmnel Robotic Controller".Proceedings qf the I 9 9 0 ASME Intematonol Computers mEngineering Cmferme. Boecoa Massachm. 1990, pp. 607412.Ted~nical Manual far Chimuti Milecroll SCR Drive Conhuller,publication I 7-000-0490hiA. Copyri@ by Cincinnati MilauonM~ketingc ortlpany 1984, CMI Eleftrial System Division, Lebanon,Ohio.
M. Aalund, D. T m , "Design and Development of a M u l t i b lRoboric Controller," Internal Report, University of Texas at Austin,May, 1991.
I. Geisinger, D. Tesar. "Actuator Coatroller Design and Implementation,"
fn&mal Report, University of Texas at Anstin, septrmba, 1992.MUttufacUhg htedonol'92, bhch B-Aprill, 1992. p~ 101-1 14.

Servomechanism

Servomechanism, or servo is an automatic device that uses error-sensing feedback to correct the performance of a mechanism. The term correctly applies only to systems where the feedback or error-correction signals help control mechanical position or other parameters. For example, an automotive power window control is not a servomechanism, as there is no automatic feedback which controls position—the operator does this by observation. By contrast the car's cruise control uses closed loop cruise control uses closed loop feedback which classifies it as a servomechanism.
A servomechanism is unique from other control systems because it controls a parameter by commanding the time-based derivative of that parameter. For example a servomechanism controlling position must be capable of changing the velocity of the system because the time-based derivative (rate change) of position is velocity. A hydraulic actuator controlled by a spool valve and a position sensor is a good example because the velocity of the actuator is proportional to the error signal of the position sensor.
Servomechanism may or may not use a servomotor. For example a household furnace controlled by thermostat is a servomechanism, yet there is no motor being controlled directly by the servomechanism.
Servomechanisms were first used in military fire-control and marine navigation equipment. Today servomechanisms are used in automatic machine tools, satellite-tracking antennas, remote control airplanes, automatic navigation systems on boats and planes, and antiaircraft-gun control systems. Other examples are fly-by-wire systems in aircraft which use servos to actuate the aircraft's control surfaces, and radio-controlled models which use RC servos for the same purpose. Many autofocus cameras also use a servomechanism to accurately move the lens, and thus adjust the focus. A modern hard disk drive has a magnetic servo system with sub-micrometre positioning accuracy.




Tyicpal servos give a rotary (angular) output. Linear types are common as well, using a screw thread or a linear motor to give linear motion.
Another device commonly referred to as a servo is used in automobiles to amplify the steering or braking force applied by the driver. However, these devices are not true servos, but rather mechanical amplifiers. (See also Power steering or Vacuum servo.)
In industrial machines, servos are used to perform complex motion.

In many applications, servomechanisms allow high-powered devices to be controlled by signals from devices of much lower power. The operation of the high-powered device results from a signal (called the error, or difference, signal) generated from a comparison of the desired position of the high-powered device with its actual position. The ratio between the power of the control signal and that of the device controlled can be on the order of billions to one.
All servomechanisms have at least these basic components: a controlled device, a command device, an error detector, an error-signal amplifier, and a device to perform any necessary error corrections (the servomotor). In the controlled device, that which is being regulated is usually position. This device must, therefore, have some means of generating a signal (such as a voltage), called the feedback signal, that represents its current position. This signal is sent to an error-detecting device. The command device receives information, usually from outside the system, that represents the desired position of the controlled device. This information is converted to a form usable by the system (such as a voltage) and is fed to the same error detector as is the signal from the controlled device. The error detector compares the feedback signal (representing actual position) with the command signal (representing desired position). Any discrepancy results in an error signal that represents the correction necessary to bring the controlled device to its desired position. The error-correction signal is sent to an amplifier, and the amplified voltage is used to drive the servomotor, which repositions the controlled device.
A typical system using a servomechanism is the communications-satellite–tracking antenna of a satellite Earth station. The objective is to keep the antenna aimed directly at the communications satellite in order to receive and transmit the strongest possible signal. One method used to accomplish this is to compare the signals from the satellite as received by two or more closely positioned receiving elements on the antenna. Any difference in the strengths of the signals received by these elements results in a correction signal being sent to the antenna servomotor. This continuous feedback method allows a terrestrial antenna to be aimed at a satellite 37,007 km (23,000 miles) above the Earth to an accuracy measured in hundredths of a centimetre.