Development of analog/hybrid terminals for teaching system dynamics

Transcription

1 Development of analog/hybrid terminals for teaching system dynamics by DONALD C. }\iartn North Carolina State University Raleigh, North Carolina NTRODUCTON A recent study completed by the School of Engineering at North Carolina State University brought to light a very serious weakness in our program to employ computers in the engineering curricula, i.e., the inherent limitation on student/computer interaction with our batch, multiprogrammed digital system. The primary digital system available to students and faculty is the BM SYSTEM 360/Model 75 located at the Triangle Universities Computation Center in the Research Triangle area. This facility is shared with Duke University and the University of North Carolina at Chapel Hill. n addition, the Engineering School operates a small educational hybrid facility consisting of an BM 1130 interfaced to an EA TR48 analog computer. We use some conversational mode terminals on the digital system but it has been our experience that they are of limited value in the classroom and, of course, only accommodate on the order of two students per hour. t is our feeling that terminals based on an analog or hybrid computer would materially improve student/ computer interaction, especially aiding the comprehension of those dynamic systems described by ordinary differential equations. This paper has resulted from our attempts to outline and define the requirements for an analog computer terminal system which would effectively improve our teaching in the broad area of system dynamics. The need for some reasonably priced dynamic. classroom device becomes apparent when we consider the ineffectiveness of traditional lecture methods in such courses as the ntroduction to Mechanics at North Carolina State University. This is an engineering common core course in mechanical system dynamics which has an enrollment of about 400 students per semester. This is a far cry from early Greek and Roman times when a few students gathered around a teacher, who made little or no attempt to teach facts but instead attempted to stimulate the students' imagination and enthusiasm. As pointed out by Professor Alan Rogers at a recent Joint SC/ ACEUG Meeting at NCSU, this intimate student-professor relationship simply cannot be achieved in today's large classes unless the instructor learns to make effective use of the modern tools of communication, i~e., movies, television, and computers. n effect, we turn students off by our failure to recognize the potential of these tools, especially the classroom use of computers. The material which follows points out the need for interactive terminals, describes the capabilities of prototype models already constructed, and then outlines the classroom system which is presently being installed for use in the fall of THE NEED FOR NTERACTVE TERMNALS Classroom demonstrations The computer has long held great promise both as a means for improving the content of scientific and. technical courses and as an aid for improving methods of teaching. While some of this promise has been realized in isolated cases, very little has been accomplished in either basic science courses or engineering science courses in universities where large numbers of students are involved. t is certainly true that significant improvement in course content can be achieved by using the computer to solve more realistic and meaningful problems. For example, the effects of changing parameters in the problem formulation can be studied. With centralized digital or analog computing facilities, this can be accomplished only in a very limited way, e.g., problems can be programmed by the instructor and used as a demonstration for the class. Some such demonstrations are desirable but it is impossible to get the student intimately involved, and at best they serve only as a supplement to a text book. At North Carolina State 241

2 242 Fall Joint Computer Conference, 1970 Figure 1 University we have developed a demonstration monitor for studies of dynamic systems which seems to be quite effective. We recently modified our analog computers so that the primary output display device is a video monitor which can be used on-line with the computer. Demonstrations have been conducted for other disciplines, for example, a Sturm Liunville quantum mechanics demonstration for a professor in the Chemistry Department. This demonstration very graphically illustrates the concept of eigenfunctions and eigenvalues for boundary value problems of this type. A picture of the classroom television display is shown in Figure 1. The instructor's control panel makes several parameters, function switches and output display channels available for control of the pre-patched demonstration problem. Switches are also available to control the operation of the analog computer. The direction we are proceeding in the area of demonstration problems is to supply the instructor with a pre-patched problem, indicate how to set potentiometers, and how to start and end the computation. Since the display is retained on a storage screen oscilloscope with video output, he can plot multiple solutions which will be stored for up to an hour, or he can push a button to erase the screen at any time. Some demonstrations have been put on the small video tape recorder, shown in Figure 1, but we find the loss of interactive capability drastically decreases the effectiveness of the demonstration. Student programming n addition to classroom demonstrations, the student can be assigned problems and required to do the programming. While we believe this to be an excellent approach for advanced engineering courses, such as design, systems analysis, etc., it has proven less than satisfactory for the first and second year science and engineering courses. Even when the students have had a basic programming course, valuable classroom time must be spent in techniques for programming, numerical methods, and discussions of debugging. The students tend to become involved in the mechanics of programming at the sacrifice of a serious study of the problem and its interpretation. With inexperienced students, even when turnaround time on the computer is excellent, the elapsed time between problem assignment and a satisfactory solution is usually much too long. We have tried this student programming approach for the past four years in a sophomore chemical engineering course. While it has been of some value, we are now convinced that it is not a satisfactory method of improving teaching. The teaching of basic computer programming does, however, have a great deal of merit in that it forces students to logically organize their thoughts and problem solving techniques. Also it helps to provide an understanding of the way computers can be applied to meaningful engineering problems. Thus, we intend to continue the teaching of basic digital computer programming in one of the higher level languages at the freshman or sophomore level, and then make. use of stored library programs for appropriate class assignments. n addition, we will continue to assign one or more term problems in which the student writes his own program, especially in courses roughly categorized as engineering design courses. Digital computer terminals t is appropriate at this point to emphasize why we feel time-shared or conversational mode terminals are not the answer to our current problem. t has been our experience that the conversational te~minal is an outstanding device for teaching programming, basic capabilities of computers, and solving student problems when the volume of data is limited. However, if the relatively slow typing capability of students is considered, we have found that a class of 20 or 30 students can obtain much faster turnaround time on the batch system. To be sure, the student at the terminal has his instantaneous response, but the sixth student in the queue is still waiting two hours to run his program and use his few tenths of a second CPU time. One can certainly argue that this is an unfair judgment since the solution is to simply buy more terminals for about

3 Development of Analog/Hybrid Terminals 243 $2500 each. Unfortunately, these terminals are like cameras and their use involves a continuing expense often greater than the initial cost. Connect time and communication costs for a sufficient number of terminals have discouraged such terminal use on the campus at the present time. The experience of the North Carolina Computer Orientation Project, which essentially provided a free terminal for one year at have-not schools, has been similar in that it proved very difficult for these schools to utilize the terminal in any science course other than a course in programming. Present limitations on classroom use of computers t is reasonable to ask why, in a university that has had good computing facilities for some time, computer use in the classroom is so limited. We feel there are several reasons why the majority of instructors do not use this means of communication to improve instruction techniques. The first of these reasons must be classed as faculty apathy. There is no other explanation for the fact that less than 25 percent of our engineering faculty use the digital computer and less than 5 percent avail themselves of our analog facility. Admittedly, it is extremely difficult for a physicist, chemist or engineer who is not proficient in computing to program demonstration problems for his classes. Because such demonstrations, while a step in the right direction, do not really make use of the interactive capability of the computer to excite the students' imagination, there is often little motivation for the professor to learn the mechanics of programming. Fortunate indeed is the occasional instructor who has a graduate student assistant competent to set up and document such demonstrations. The second reason, closely coupled to the first, is that computer output equipment for student use in the classroom is either not available or just too expensive for large classes. Digital graphics terminals, for instance, sell for between 15 and 70 thousand dollars, depending on terminal capability, and an analog computer of any sophistication at all will cost 5 to 10 thousand d,ollars with the associated readout equipment. n our basic introductory mechanics course, with ten sections of forty students, a minimum of twen ty such terminals would be required if we assume two students per terminal as a realistic ratio. Even if such analog or digital computer terminals were available, we would then be faced with the problem of teaching the students (and faculty) a considerable amount of detailed programming at the expense of other material in the curriculum. We feel that analogi hybrid computer terminals designed to accomplish a specific, well-defined task, will provide an economical interactive student display terminal for many engineering courses. Such a terminal is described in this paper. CLASSROOM TERMNAL SYSTEM We have recently received support from the National Science Foundation to study the effect of student interaction with the computer in courses which emphasize the dynamic modeling of physical systems. t is a well known fact that interaction with a computer improves productivity in a design and programming sense. The question to which we are seeking the answer is: Will computer interaction also improve the educational process effectively without leaving the student with the impression that we are using a "magic black box"? To accomplish this goal, we are installing sixteen analog/hybrid terminals in a classroom to serve thirtytwo students. The classroom in which these terminals will be placed is about 150 feet from the School's analog and digital computer facility. At this point, we should place some limits on terminal capability and function. f we accept the premise that the student need not learn actual patching to use the analog computer terminal and eliminate the traditional concept of submitting a hard copy of his computer output as an assignment, the desirable features of a terminal might be as follows: 1. Parameters: The student must have the capability of varying at least five different parameters in a specific problem. Three significant digits should be sufficient precision for these parameters and their value should be either continuously displayed or displayed on command. 2. Functions: The student should have access to function switches to perform such operations as changing from positive to negative feedback to demonstrate stability, adding forcing functions to illustrate superposition, adding higher harmonics to construct a Fourier approximation of a function, introducing transportation delay to a control system, etc. Three to five of these function switches should be sufficient. 3. Problems: The student should be able to select several different problems, say four, at any of the individual terminals. Depending on the size of the analog computer, the student could use the terminal to continue a study of a problem used in a prior class, compare linear and nonlinear models of a process, etc.

4 244 Fall Joint Computer Conference, Response Time: The response time for each of twenty terminals should be about one to three seconds, i.e., the maximum wait time to obtain one complete plot of his output would be something like one second plus operate time for the computer. Computer operate time has been selected as 20 milliseconds for our equipment although a new computer could operate at higher speeds if desired. 5. Display: The display device for each terminal must be a storage screen oscilloscope or refreshed display for x-y plots. Zero and scale factors must be provided so that positive and negative values and phase plane plots can be plotted. Scaling control must be presented in a manner which is easy for the student to use and understand. 6. Output selector: The student should be able to select from four to five output channels for display on the oscilloscope. 7. nstructor display: The instructor should have a terminal' with a large screen display which the entire class can observe. His control console should have all the features of the student terminals and should also have the capability for displaying anyone of the student problem solutions when desired. He should also have master r--' mode controls to activate all display ~-.--'- i to!,~~ ~~? ~~; _T!t_,mK_s-' ;t J=> '.. - ;?".--- ::hl ~ i~, -,"~.-:~ , ----,ril... ~--' ~ ~ U' ~ t \~ L ~ ) '.;,_ b ~~t ~~ ~ ~~~~.),~~~ ;;,~,!, CON'RO. ~.;-UNG A) 1.:~NG 9>!HL VGl'Pi GOMP'?j'ER FOR CON'ROl. PlD SCALNG O} '!'EHMlNALS Figure 2-Flow sheet for proposed analog/hybrid terminal system terminals. t would be advantageous for the instructor to have a small screen display to monitor student progress without presenting the solutions to the class on the large screen. Given a terminal system with these features, we have then defined the primary objective as being a study of the use of this classroom in some specific courses. We are initiating this evaluation with one course in Engineering Mechanics, ntroduction to Mechanics (EM 200) and two courses in Chemical Engineering, Process Analysis and Control (CHE 425), and ntroduction to System Analysis (CHE 225). Thus, we start our evaluation with one sophomore engineering core course with three to four hundred students per semester and two courses with thirty to forty students per semester at the sophomore and senior level. n addition to these two courses, we are attempting to schedule as many demonstrations of the system as possible for other departments in the hope of stimulating their imaginative use of the terminals. A flow sheet for the classroom terminal system is given in Figure 2. The system includes a small digital mini computer which is to be used for control, storage, and scaling of terminal information. The system consists of the following components: a. Sixteen student terminals b. One instructor terminal c. Digital mini computer with /O device for programming d. Control interface to analog computer The inclusion of a small digital computer in this terminal system opens up some very interesting future possibilities such as using the terminals for digital simulation as well as analog. As will be seen in the next section, the digital computer provides scaliilg so that parameters can be entered in original problem units. t also acts as temporary storage for each terminal as it awaits service and controls the terminal system. The system is designed to operate in the following manner. The instructor informs the computer operator that he wishes to use problem CHE 2 during a certain class or laboratory period. The operator places the proper problem board on the TR-48 and then sets the servo potentiometers which are not controlled by the student terminals and static checks the problem with the hybrid computer. He also sets up the program CHE 2 on the mini computer just prior to class. From this point on the system is under the control of the instructor and students.

5 Development of Analog/Hybrid Terminals 245 FUNCTONAL DESCRPTON OF TERMNAL General The basic terminal configuration is shown in Figures 3 and 4. All display functions are located on the upper display panel and control or data input is provided on the inclined control panel. +\yisi ioisiul D!'~:rAl. Dl;':r'LA:t fhoiol ~ a Control The controi functions available to the student include power on-off, store, non-store, erase, and trace intensity. Th~se controls are located on either side of the oscilloscope display as shown in Figure 4. ndicator lights are also provided in this area for terminal identification, error and terminal ready status. The erase function is used quite often by the student and thus is also available on the keyboard. n addition to the basic operating controls, the student can request either a single solution or repeated solutions on his display unit. For the single solution mode, he displays one solution as soon as the sequencer reaches his terminal address. n the repeat solution mode, his oscilloscope is unblanked and displays a solution each time the sequencer reaches his address. DBDEB Display Scaling Figure t RESERVED P'OR fwure ALPHAl4ERC KlTOCAiW EXf'ANSlVt. ~ J EJ [!]!!) El ffi@lill 0 CD0[1 0 kj@]~ GB~ EJ Figure 4-Terminal display and control panels ~ B E ~ The worst case response time for either mode would be on the order of one second, even if all other terminals had solution requests pending. Output display and scaling The primary output display device is a Tektronix type 601 storage screen oscilloscope. This oscilloscope is mounted directly behind the vertical front display panel as shown in Figures 3 and 4. These oscilloscopes are modified to provide for s~aling as shown below. The operating level circuit is modified to provide a switched store and non-store operation for the user. Output scaling is automatically selected when the student depresses anyone of the four push buttons on the left hand side of the display panel. The picture which indicates the normalized scaling is printed on the face of the lighted push button switch. The left hand switch scales the output signal to display positive X and Y values. The second switch. displays positive X and allows for both positive and negative values of the Y variable. Phase plane plots can be displayed by selecting the right hand switch in this set. We have been using this type of output scaling for oscilloscopes in our laboratory for over a year with very satisfactory feedback from student users. The student must have the option of selecting from several X and Y analog output lines. This option is provided at the terminal by depressing either or [!] then the appropriate number on the keyboard, and then the 1 ENTER key. For example, if

6 246 Fall Joint Computer Conference, 1970 the student is instructed to plot Xl versus Y4, he would actuat~ the following keys on the terminal keyboard as shown below: This system allows the student to enter parameter values in the actual problem units, e.g., if the input temperature for a heat exchanger simulation is 150 F, he sets this value rather than some normalized fraction. f an error is made before ENTER is depressed, The digital software sets up the linkage between the requested output line and the control unit which switches variable outputs for specific problems to the two analog output lines leading to the classroom. All analog outputs are on the two X and Y lines, but each terminal Z line is energized only at the appropriate time, i.e., in answer toa solution request for that terminal. Parameter entry There are many ways in which parameters can be set fron an analog or hybrid computer terminal. n the first terminals we constructed, parameters were simply multiplexed potentiometers connected to the analog computer with long shielded cable. Thumbwheel switches can be effectively used to set digital to analog coefficient units or servo potentiometers.at the analog computer. Since this hybrid terminal system includes a small digital computer, parameters will be entered with a fifteen digit keyboard as indicated in Figure 4. The parameter function switch, rp AR, keyboard, and ENTER keys are used in the following sequence. Suppose the student wishes to set parameter number four at a value of 132. He would depress switches in the following sequence: par E EJ EJ o ENTER or PAR lenter f the parameter number three were less than unity, say 0.05, he would enter PAR EJElOEJ01ENTER or PAR le 0 EJ E]ENTER the register can be cleared with the CLR key. The use of actual rather than normalized parameters requires additional registers in the terminal but is essential for beginning students. We must remember that they are studying the dynamic system, not analog computer scaling. t is also in keeping with the concept of using the terminal as input and output for the hybrid computer. f the parameters represent frequency, temperature, pressure, etc., they should be entered as they appear in the problem statement if at all possible. Since "parameter values are scaled by a program in the digital computer, scientific notation can also be used to permit both E and F format data entry from the hybrid terminal. The digital software interprets the input data to separate the mantissa and exponent portions of the number entered in E format. For example, the student might enter fenter and the digital computer would convert this number to Reading parameter values One significant advantage of the thumbwheel switch parameter entry as opposed to the keyboard is the ability to remember a specific parameter value at any time. f the student forgets the value, he needs to be able to display it at the terminal on request. This capability is provided by a six character plus sign display module located in the upper left corner of the terminal as shown in Figures 3 and 4. This display unit automatically shows the" student that his parameter value was correctly entered at the keyboard and accepted by the digital computer. The format of the output display is controlled by the digital computer software. n addition to displaying the correct value of the parameter when entered at the keyboard, the

7 Development of Analog/Hybrid Terminals 247 value of any parameter previously set can be indicated using the DSPLAY key. Suppose the student wishes to retrieve but not change the value of parameter three. He would press ~, thenmon the keyboard, and then the DSPLAY key. The digital computer software then causes the proper value to be displayed and light keyboard button number ] to identify the requested parameter number. A separate digital display module could be used to indicate the requested parameter number, but lighting the keyboard number has some advantage when displaying the status of function switches as noted later. Analog/digital readout One of the advantages which immediately becomes apparent when the terminal includes digital readout and a small digital computer is the capability of returning numbers which are the result of an iterative calculation. A' first order boundary value problem where the unknown initial condition or time constant is sought would be one illustration. Another example which we have been using in a basic process instrumentation course is to demonstrate the operation of a voltage to frequency converter or time interval analog to digital converter. Any digital or analog number can thus be returned to a terminal by selecting one of the output lines for display. The student presses 0, followed by the appropriate number on the keyboard, and then the DSPLAY J function key. This display feature is also extremely valuable in presenting calculated results from the hybrid computer through the trunk lines as indicated in the overall system diagram, Figure 1. One variable parameter The availability of a control which can vary one parameter through some range of values is an important feature of the terminal. Thumbwheel switches and keyboard entry of parameters are fine but tend to be somewhat slow when the student is interested in observing the effect of a range of parameter variations on a simulation or fitting a model to experimental data points. A potentiometer on the terminal, as we have employed in the past, avoids this problem but involves transmission of analog signals over long lines. As a compromise, either a two or four position switch is used to increment the value of any selected.parameter at a rate determined by the digital software. Thus, the student increases or decreases a particular parameter value at either a fast or slow rate with this switch. The sequence of operations would be as follows: suppose the student wishes to vary parameter four through a range of values to observe its effect on the solution. He might choose a starting value of zero for this parameter which, for example, might represent the damping coefficient in a linear, second order system. He presses PAR, followed by the numbers 0 and 0 on the keyboard and then selects the ENTER key. He then selects the REPEAT SOLUTON mode to observe the solution each time the sequencer reaches his terminal number. f the student has selected the STORE mode, he can then plot a family of curves as he increases the damping factor from zero to unity by pushing the parameter slewing switch in the increase direction. A four position switch allows a choice of either fast or slow incremental changes in the parameter value. Another obvious application for this function is in curve fitting of experimental data with one or more parameters. Function switches Control of the electronic function switches is provided at each terminal. To set function switch one in theion position, the student presses FUN,followed by the number [] and 'o"k on the keyboard and then the ENTER key. f he wishes to know the present state of any function switch, he presses 'FUN, the switch number, and then the

8 248 Fall Joint Computer Conference, 1970 DSPLAY key. The terminal response is to light the function switch number and either the o"k or OFF keys on the keyborad to indicate the present state of that particular switch. These function switches can be used in any way desired by the instructor, e.g., adding successive terms of a power or Fourier series to demonstrate the validity of these approximations, adding various controller modes in a process control simulation, etc. The instructor's terminal The instructor's terminal is designated as terminal number zero. This termi~l uses as its primary output device a Tektronix Type 4501 storage oscilloscope instead of the Type 601. Since this scanning oscilloscope has video output, the instructor can display his solution on the closed circuit television monitor for the class at any time. n addition, the instructor has the capability of un blanking any or all student terminals to let them have a "copy" of his solution to compare with their own. He can also unblank his terminal and pick selected student solutions for display to the rest of the class. SOFTWARE Basic operating system The basic software to serve the analog terminals is written in assembly language for the PDP-8 control computer. This is a 4K machine with hardware multiply and divide, although this feature is not essential for terminal operation. The basic cycle time for the system is controlled by the analog clock which alternately places the analog computer in the initial condition and operate modes every twenty milliseconds. The first ten milliseconds of each initial condition period is to ensure adequate time for problem and function selection by the relay multiplexer. The second ten milliseconds is the normal initial condition time to charge the integrating capacitors as shown below. ::.t~rx NORMfJ MlX NOPA!,L, 3TCH~l'} AN.t.LOG 301 T CHNG ANALOG PROSLEM SOLUT ON TME...1 C TME TT.fE g C TME ~ AN"; LOG NrrAL...1 g ANALOG OPERATE ANALOG NTAL ~ CONDrrON MODE MODE CONDrrrON MODE 8 Pl R ~ ms --~1Mo1.r ms ~... ~, ms A terminal user can activate an action key at any time, e.g., t ENTER, DSPLAY, SNGLE SOLUTON, or REPEAT SOLUTON. This request for action, along with the necessary data and address is stored in a 32 bit shift register in each terminal. As each terminal is interrogated in sequence by the PDP-8 the action bit is tested. f the user wants service, his data is transferred to a specific core area. For instance, suppose he wishes to set parameter number 1 at a value of He activates the following keys: ENTER. The ~ []] rn 0 m [!] ENTER( key is the action code in this instance. When the sequencer reaches his terminal, this data is transferred to storage in the proper memory locations in the PDP-8. A similar action is taken to set function switches, and select problems or output channels. The basic operating system software controls all of these action operations. When a solution is requested, the parameters, functions, and outputs, along with an unblanking signal, is sent back to the terminal during the next analog computer operate cycle. The basic system software also converts the floating point parameter values supplied by the student to integer values used by the digital coefficient units or digital to analog converters. This feature of floating point data entry requires that the instructor provide the scaling information for each problem as described next in the application software section. A pplication software The application software is written in a special interactive language developed for the PDP-8. This language makes use of a cassette tape recorder in our system, but could be used from the teletype if necessary. The information required by the terminal operating system to convert floating point to integer parameters is their maximum and minimum values. When the instructor is setting up the terminal problem, the computer software solicits responses similar to the following: DENTFY YOUR PROBLEM NUMBER The instructor would then type, CHE4

9 Development of Analog/Hybrid Terminals 249 The computer responds with PROVDE MAXMUM AND MNMUM VALUES FOR THE PARAMETERS f the instructor wishes to give the student control over parameters one and two, he types PAR 1, MAX 50, MN 25 PAR 2, MAX 0.5, MN 0 END PAR LST A similar conversational procedure is used to identify function switches, problems, and analog computer output channels. n our system, this scaling and switching data is stored on the magnetic tape cassette. A paper tape unit could also be used if desired. When the instructor wishes to use the terminal system at a later date, he places his cassette in the tape deck and the proper problem board on the analog computer. From this point on, the problem or problems can be controlled from the individual terminals. COSTS The cost of a system such as described in this paper is naturally dependent on the number of terminals involved. Since our system was developed jointly with Electronic Associates, it is difficult to evaluate the actual development and design costs. The individual terminals, including a type 601 Tektronix storage oscilloscope should be on the order of $3500 to $4000 each. Mini computers such as used in this system would range from $6000 to $10,000 and cassette tape systems are available for about $3000. The major question mark in the estimation of system cost is the hybrid control interface to couple the analog and digital computers. f a special interface could be developed for about $10,000, the cost of a ten terminal system would be on the order of $60,000. This system could be coupled to any analog computer and, of course, provides basic hybrid capability as well as terminal operation. f a hybrid computer were already available, the terminals could be added for about $3500 to $4000 each plus wiring costs. CONCLUSON The key to student and instructor use of these terminals is the development of appropriate handout materials. Several of these handouts have been written in programmed instruction form and have resulted in very favorable feedback from students who used early models of the terminals. Although the complete classroom system will be used for the first time in the fall of 1970, we have been very gratified with student acceptance of the few terminals now in use. Laboratory reports now consist of answering specific questions concerning the dynamic system under study rather than computer diagrams and output. Also, the student can really proceed at his own pace, and return at any time to repeat a laboratory exercise simply by giving the computer operator the problem number. We are excited about the potential of this classroom terminal system and believe that we will see significant improvement in the students' understanding of dynamic systems as the system is used in additional curricula.

10