Proceedings of the 2005/2006Multi-Disciplinary Engineering Design Conference
Kate Gleason College of EngineeringRochester Institute of Technology
Rochester, New York 14623
Project Number: 06508
CONTROL SYSTEMS LAB INTERFACE IMPROVEMENT
Neil Burkell / Electrical Engineering Michael Abbott / Electrical Engineering
ABSTRACT
This work focused on the improvement of the current laboratory setup used in the Control Systems Design course taught in the Electrical Engineering Department at the Rochester Institute of Technology. A new digital interface was designed to interface Simulink within MATLAB to the servo DC motor used in the lab. Two different software toolboxes were used and compared to generate code from a Simulink block diagram for a data acquisition card to act as the digital controller. The results from testing show that the digital control interface was successfully designed and built with both toolboxes.
INTRODUCTION
The current Control Systems Lab equipment was purchased from Feedback Instruments LTD in London, England. The equipment that was purchased contained an analog control board part number 33-110, digital control board part number 33-002, and a servo DC motor board part number 33-100. The analog control board is used in the current lab setup for hardware experiments with the motor board. The digital control board is no longer operational. The board was found from past work by Ruben Mathew, a former electrical engineering student at RIT, to be outdated and poorly designed. Current digital control lab experiments are taught purely through the use of simulation in MATLAB and Simulink.
The goal of this project was to develop a digital control interface between the existing servo DC motor board and Simulink. The designed interface was to establish control of the servo DC motor through the use of a Simulink block diagram control algorithm. The digital control interface needed to acquire real
time data from the servo DC motor board’s sensors and be able to output real time outputs to the motor board after being processed by the control algorithm drawn in Simulink. The interface also needed to allow students to capture the acquired data for processing after an experiment was complete.
The lab interface will be used in the future course offerings within the Electrical Engineering Department and therefore all funding for this project has been provided from the RIT Electrical Engineering Department. The interface was tested with the Control Systems Design course in mind. This interface could be used in other controls classes such as Fuzzy Logic, Modern Control Theory, and Digital Control Systems as well as other classes dealing with signal processing and robotics with only minor changes in the design.
NOMENCLATURE
NC -- No ConnectionKm – Motor ConstantKtach – Tachometer Voltage Constantτe – Electrical Time Constantτm – Mechanical Time ConstantPCB – Printed Circuit BoardOS – OvershootTr – Rise TimeVss – Steady State VoltageTp – Peak timeRTW – Real Time WindowsPCI – Peripheral Component InterfaceDAQ – Data Acquisition
PROJECT REQUIREMENTS
Requirements were developed for this project with Dr. Mathew and Dr. Phillips, advisors from the Electrical
Proceedings of the Multi-Disciplinary Engineering Design Conference Page 2
Engineering Department. The first requirement developed was to use the existing equipment from the current Control Systems Lab including the Feedback 33-100 servo DC motor, the Feedback power supply, and MATLAB/Simulink on the lab computers. Functionality requirements were then developed. The interface needed to utilize Simulink with a control algorithm block diagram on one end and the servo DC motor board on the other. With this requirement students do not need to learn a programming language to develop digital control algorithms. When a control algorithm was used in a hardware experiment with the motor board, it was essential that the student be able to see the acquired sensor and output data going to and from the motor board. It was essential that the sampling time of the system be easily changeable in order to observe the effect on the system as sampling time varied. The final and most important functionality requirement is that the results from the digital controller coincide with the theoretical and analog board results already investigated in the lab course. It was also determined that the digital control interface be covered to protect the interface and students from damaging the system or themselves. Once the functionality requirements of the system were developed, specifications were determined for the digital controller. All analog inputs used in acquiring sensor data including velocity and position were determined to have a resolution of 16 bits, a voltage range of -10 V to 10 V, and a sampling rate that can be varied from 300 ms to 1 ms. All analog outputs used to drive the power amplifier for the DC motor board were determined to have a resolution of 16 bits, a voltage range of -10 V to 10 V, and a sampling rate that can be varied from 300 ms to 1 ms. The servo DC motor board also contains digital outputs in gray code that give the position of the motor shaft. Therefore, it was determined that the digital controller have a minimum of 6 digital inputs. In order to make the design robust for other applications other than the servo DC motor board, it was required that the digital controller have additional inputs and outputs. The additional analog inputs and analog outputs also had to have a 16-bit resolution with a voltage range of -10 V to 10 V and achieve a sample rate of at least 1 ms.
DESIGN CONCEPTS/FEASIBILITY
As a first deliverable to the Electrical Engineering Department, many different design concepts were developed. These concepts were then analyzed for their feasibility, cost, and ease of use in the laboratory setting. There were four main design concepts that were analyzed. These included using a DSP board from Analog Devices, a Universal Serial Bus Data Acquisition Board from National Instruments, a PCI data acquisition card inside a lab computer that is
supported by MATLAB’s Real Time Windows Target, and using MATLAB’s xPC Target with a target computer which has data acquisition capabilities and runs with an xPC kernel, external from any operating system. These concepts were developed, compared, and presented to the Electrical Engineering Department.
The final decision after presenting the information was to develop two engineering models using Measurement Computing PCI-DAS1602/16 DAQ cards with both the Real Time Windows Target toolbox and xPC Target toolbox. This card was supported by both toolboxes, met all specifications for the project, and had the best overall cost analysis. By developing and comparing both concepts, the Electrical Engineering Department could make a decision on which route would best meet there needs for the controls laboratory.
PRELIMINARY DESIGN
The preliminary design was accomplished by combining knowledge of the major components of both systems being developed to be able to synthesize a working system design. This involved understanding completely the current controls lab equipment, the PCI-DAS1602/16 card, the Real Time Windows Target Toolbox, and the xPC Target Toolbox.
The current controls lab equipment includes the analog control board and the 33-100 servo DC motor board. The motor board contains a 34 pin ribbon cable header on the board. In order to control the motor each pin on the motor board was investigated in order to verify which inputs or outputs from the PCI card were necessary for control. The documentation from the manufacturer was lacking information about the 34 pin ribbon cable header. Therefore, an electrical schematic of the motor board was created by tracing the motor board. After completing the schematic, it was found that there were some pins that were impossible to trace on the motor board. An electrical schematic of the analog board was created to help identify these pins. From this information, the necessary inputs and outputs to control the motor were found.
A photo of the chosen Measurement Computing PCI-DAS1602/16 PCI card can be seen in Figure 1.
Paper Number nnnnn
Proceedings of the KGCOE Multi-Disciplinary Engineering Design Conference Page 3
The PCI-DAS1602/16 offers 16 single ended analog inputs with 16-bit resolution, a range of -10 V to +10 V, and a 200 kS/sec maximum sampling rate. The card has 2 analog outputs with 16-bit resolution, a range of -10 V to +10 V, and a maximum output rate of 100 kS/sec. The card has 24 digital inputs or outputs that can be configured through software to be either in groups of 8. The card has two 50 pin ribbon cables that plug into the card carrying the different inputs and outputs. Measurement Computing had very specific information describing the different pins on the two 50 pin ribbon cables.
Once the servo DC motor board and PCI card were understood, an interface was needed to connect the necessary inputs and outputs from the PCI card to the motor board. This was accomplished by using two Measurement Computing CIO-MINI50 breakout boxes. Using 22 gauge wire connected to the breakout boxes and crimped sockets that fit the pins on the 34 pin header on the servo DC motor board, an interface was created. This allowed for preliminary testing of the system in order to make sure that the correct inputs and outputs were being used from the servo DC motor board to the PCI card.
The Real Time Windows Target toolbox is operated within the Windows environment. Real-Time Windows Target is a means of interfacing MATLAB/Simulink with the Measurement Computing PCI card to implement control algorithms. The Real Time Windows toolbox allows a given block diagram input by the user in Simulink to be used in generating C code to run the PCI-DAS1602/16 card. The target toolbox offers blocks for various analog and digital inputs or outputs which can be placed in a Simulink Diagram. When these are implemented with control algorithms in Simulink, sensor data from the servo DC motor board can be acquired and processed with the algorithm and then output to the motor board to control the motor. Figure 2 contains a block diagram of the Real Time Windows Target design.
Figure 2: Block Diagram of Real Time Windows Target Design
The xPC Real Time Target uses a standalone processor or computer to execute generated code from Real Time Workshop. In the xPC Target design, the target will be a desktop computer with the PCI-DAS1602/16 card installed. Using the xPC Real Time Target requires that the desktop computer be booted with the xPC kernel created in MATLAB. Similar to the Real Time Windows Target, there are blocks for analog and digital inputs and outputs that can be implemented with a control algorithm into a Simulink block diagram. This block diagram can then be processed on the host computer running MATLAB to generate a C executable program that is downloaded to a target computer through serial or Ethernet cables. For the design serial RS-232 cables were utilized. Once the executable program has been downloaded, it can be triggered to run from the host computer. Once triggered, the target computer runs independently from the host computer acquiring inputs and using the control algorithm built into the executable to create new outputs to control the motor. Once the executable has finished running, the acquired inputs of interest are sent back to the host computer and are available for use in MATLAB. A block diagram of the xPC Target design can be seen in Figure 3.
Figure 3: Block Diagram of xPC Target Design
Preliminary testing was done with both designs in order to verify that the control signals being sent to and read from the servo DC motor board were correct and resembled results from using the Feedback analog control board.
Once preliminary testing verified that the test setup for the digital controller had results within 5% of the theoretical results, a failure analysis was done on the entire system. The use of either of the digital controller setups will be for student use in the
Proceedings of the Multi-Disciplinary Engineering Design Conference Page 4
laboratory. The system should be designed with safety for students and the equipment in mind. The interface board between the servo DC motor board and the PCI-DAS1602/16 card was the only system item that had the freedom to design in safety precautions. Therefore, the servo DC motor board and the PCI-DAS1602/16 card were analyzed for potential problems.
An application engineer from Measurement Computing, Matthias Watkins, was contacted on January 11, 2006, to see what kind of protection the PCI-DAS1602/16 card needed. His response stated that the analog inputs are rated for a maximum of ±15 V. Watkins explained that the front end components of the PCI card could be damaged if input voltage to the card exceeded ±15 V. Watkins also stated that the analog outputs were already designed with short circuit protection. Watkins stated that the analog outputs positive and negative terminals could be shorted together with a clamped output current of 25 mA and could be run in this mode forever. Therefore, the analog outputs are already protected. The analog inputs are the only potential hazard.
The servo DC motor board does have a ±15 V sources through the 34 pin connector. The input shaft position, output shaft position, and tachogenerator signals from the board being read by analog inputs have a ±10 V range. There will be no connection between the +15 V source or -15 V source on the interface board to the PCI card. Therefore, the only failure mode where the PCI card can have ±15 V input to it is if there is a short on the interface card between the ±15 V and an analog input. Watkins stated that the card is rated up to ±15 V so even if this occurs the card should not be damaged. The only failure mode where the card could get damaged is if there is a similar short on the interface card and the power supply, which provides power to the servo DC motor board, malfunctions outputting a voltage higher then ±15 V. There is a very low probability that both these events will occur at the same time.
The servo DC motor board’s power amplifier inputs are rated up to ±10 V, which is the maximum output range of the analog outputs. This is the only connection to the servo DC motor board that receives an input. Therefore, the motor board should be safe from any input from the PCI-DAS1602/16 card.
From the failure analysis, it was found that the interface board did not need any protective circuitry. Laboratory codes were examined to see what needed to be done for our design. Since the motor board, analog board, and PCI card are already UL certified by their manufacturer, these items already meet laboratory code. The interface board should be
covered to protect the users in the laboratory environment.
The interface board was designed as a printed circuit board. It was designed to contain holes in each corner to be able to screw on a Plexiglas cover over it to protect the board and students from potential electric shock. The board was designed with the same part number 1-103308-7 34 pin header that is used on the servo DC motor board and two part number 1-103308-0 50 pin headers. The final dimensions of the board are 3.8 inches by 2.5 inches. The connections between pins can be seen in Table 1. A two layer PCB was needed to accomplish the connections. This concluded the preliminary design for both xPC Target and Real Time Windows Target.
Table 1: Interface Board Connection Table
Paper Number nnnnn
Proceedings of the KGCOE Multi-Disciplinary Engineering Design Conference Page 5
SYSTEM MODELING
The 33-100 Servo DC Motor had to be modeled in order to be able to gather simulation results to compare with the hardware results from the digital control interface. The 33-100 Servo DC Motor was modeled as a permanent magnet DC motor two-pole transfer function as shown in (1).
The values for Km, Ktach and τm had to be determined in order to complete the model for the motor. The tachometer constant, Ktach, is defined as the ratio of tachometer voltage to shaft velocity. The motor constant, Km, is defined as the ratio of shaft velocity to input voltage and τm is defined as the mechanical time constant. The system was placed in open loop configuration as shown in Figure 4.
Figure 4: Open Loop Hardware Configuration
Using the existing Feedback Instruments analog board, the system was wired so that the gain value of Ka was one. Once this was accomplished, tachometer voltage and shaft velocity of the motor were both measured for varying inputs from -5 V to +5 V. The tachometer voltage was plotted against shaft velocity and the slope of the best fit line to the data was found to be Ktach with a value of 0.0252 V-sec/rad. Similarly, the shaft velocity was plotted against input voltage. The slope of the best fit line of the data was found to be Km with a value of 35.54 rad/(V-sec).
In order to determine a value for τm, a step input of 3 V was sent to the motor. The tachometer voltage was then measured using the oscilloscope and the time constant, 0.632 of steady state, of the step response could then be determined as the value of τm. τm was found to be 0.44 sec. The electrical time constant, τe, was given by our advisor, Dr. Mathew, to be 0.05.
ENGINEERING MODEL
When the Real Time Windows Target system and xPC Target systems were built a couple of issues were found. The interface printed circuit board had one via that was not drilled through to connect the top layer to
the bottom layer. Also there was a short found between the 5 V supply from the motor board to the positive tachogenerator pin. These errors were fixed by cutting the shorted runs on the board and soldering jumper wires to complete the correct connections listed in Table 1.
The 50 pin connectors were not delivered at the time of assembling both engineering models so generic pin headers were soldered onto the PCB to serve as the connection between the interface board and the two 50 pin ribbon cables from the PCI-DAS1602/16. The 1-103308-7 34 pin header was soldered on in order to connect the interface board to the 33-100 Servo DC Motor. This made both systems operational and ready for testing.
EXPERIMENTAL SETUP AND PROCEDURE
There were two different setups for the experimental testing dependent on the Real Time Windows Target and xPC Target. One PCI-DAS1602/16 card was installed in one BRITE 3.2 GHz Pentium IV lab computer for use with the Real Time Windows Target. Another PCI card was installed in a Dell Optiplex 1 GHz Pentium III computer for use with the xPC Target as the Target PC. The two computers were connected with a RS-232 cable in order to allow communication between the Host and Target PC. The desired experiments were run on both setups with the 33-100 Servo DC motor powered by the Feedback power supply.
In addition, the current analog board setup was used with the servo DC motor in order to provide comparable control data with the digital interface designs. The analog board connects directly to the servo DC motor via a 34 pin ribbon cable. Results were captured from the analog board using the PCI-DAS1602/16 PCI card on the BRITE lab computer. Data was acquired by connecting the tachogenerator output of the analog board to a test point on the interface board. The test point corresponded to an analog input on the PCI-DAS1602/16 which allowed data to be acquired into MATLAB.
In order to test the functionality of the interface, several different speed control algorithms were used in order to control the servo DC motor. The transfer functions of the four controllers can be seen in (2)-(5).
Proceedings of the Multi-Disciplinary Engineering Design Conference Page 6
Using the model of the servo DC motor as shown in (1), the system was simulated using a simple integrator control algorithm as shown in (2). Next the PI controller of (3) was designed using root locus design. The PI controller was also simulated using the model of (1). A one-pole controller as shown in (4) was used as well as a two-pole, one zero controller as shown in (5). Each controller’s simulation results were then compared with results from the xPC Target and Real Time Windows Target experimental data. When performing hardware and simulation testing a bias of 0.5 V was used to eliminate the dead zone of the DC motor and a step of 1 V was input after the system had come to steady state. Each controller was built on the analog board and experimentally tested. The 1 V step response were recorded for each configuration and the transient data was compared in order to evaluate the performance of the two designs against the simulation and analog board results.
The other test used was a position test. Using a feedback position controller the motor was rotated from the 270 degree position to the input position of 90 degrees. This was done on the analog board, xPC Target setup, and Real Time Windows Target setup.
The transfer function shown in (5) was generated from implementing the final lab of the current Control Systems class. The purpose of this lab is to investigate the effects of changing sampling time. Continuous transfer functions are converted to the digital domain using Zero-Order Hold and then simulated with varying sampling times. The results of the simulation were compared to the response of the motor to varying sampling time using the xPC and Real Time Windows targets.
EXPERIMENTAL RESULTS
Using the controller in (2), the 1 V step responses were plotted as shown in Figure 5. The step responses for simulation, Real Time Windows Target, xPC Target and the analog control board were shifted in order to view each response separately. The transient data was compared to theoretical values for each result and can be found in Table 2.
Table 2: Comparison of Transient Results for Integrator Controller
Next, the controller of (3) was used in order to control the 33-100 Servo DC Motor. The 1 V step responses of the system were plotted as shown in Figure 6. The transient data of each step response was measured and compared against the simulation step response. Theoretical calculations are not valid for this closed loop system because the complex poles are not dominant in this case. The tabular comparison can be found in Table 3.
Figure 6: PI Controller Step Response Comparison
Table 3: Comparison of Transient Results for PI Controller
The one-pole model of (4) was then used as the controller of the system. The 1 V step responses of the system were plotted as shown in Figure 7. The transient characteristics of each step response were compared to theoretical values. In addition to
Paper Number nnnnn
10 11 12 13 14 15 16 17 18 19 20-1.5
-1
-0.5
0
0.5
1
1.5
2
time [sec]
Tachometer Voltage [V]
Analog Control Board Result
Simulation Control Result
Digital Control MATLAB RTW Result
Digital Control MATLAB xPC Target Result
10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15-1.5
-1
-0.5
0
0.5
1
1.5
2
time [sec]
Tachometer Voltage [V]
Analog Control Board Result
Simulation Control Result
Digital Control MATLAB RTW Result
Digital Control MATLAB xPC Target Result
Proceedings of the KGCOE Multi-Disciplinary Engineering Design Conference Page 7
transient data, the steady state voltage was recorded as the single pole controller does not guarantee zero steady state error which is the case with (2) and (3) controllers. The results from the single pole controller experiments can be found in Table 4.
Table 4: Comparison of Transient Results for One-Pole Controller
The controller of (5) was used in order to show the effect of sampling time on the system. The system was run with sampling times of 0.2, 0.1, 0.05, 0.0375 and 0.025 seconds. The step responses were plotted for varying sampling times for each Real Time Workshop Target. The responses compared to the simulation results in Simulink can be seen in Table 5. All the results were very similar with the exception of the rise times. With the sampling times used it is difficult to find the rise time depending on how the sampling time period occurs relative to the time the system reaches the steady state voltage for the first time. If they do not coincide exactly then an interpolation is needed to estimate the rise time from the acquire data.
Table 5: Comparison Transient Results of Varying Sampling Times
Finally, the position data of the input and output shafts of the servo DC motor was acquired by the PCI-DAS1602/16 card using the analog board, the xPC Target and the Real Time Windows Target. The input shaft was preset to 90 degrees and then the position error of the input and output shaft was sent to the motor as an input until the output shaft was at the same position as the input shaft. The plot of the position data was common to all three experiments and can be found in Figure 8.
Figure 8: Experimental Position Data of Input and Output Shaft
From the results it is apparent that both setups can be used for the type of controllers implemented for the Controls Systems Lab. A final test was developed to compare the performance of the Real Time Windows Toolbox and xPC Target Toolbox. A fuzzy logic incremental PI controller was designed and implemented by both toolboxes to control the servo DC motor. The designed controller uses the error of the system and the derivative of error of the system each with different membership functions as inputs and verbal rules that were written to calculate how much the controller output should increase or decrease. This type of controller needs to use more processing power. A summary of the results of testing this controller with different sampling frequencies can be seen in Table 6. The Real Time Windows Toolbox allows the processing time being used by the experiment to be found. It is recommended by MathWorks to keep the percentage used for the experiment below 20 %. If the percentage goes above 20 % the results may not be accurate or the experiment will shut down automatically. The xPC Target Toolbox allows information about how much time it takes for an input to be sampled and an output to be calculated and sent to the motor otherwise known as task execution time.
Proceedings of the Multi-Disciplinary Engineering Design Conference Page 8
Table 6: Performance Comparison of Fuzzy Logic Controller with RTW Toolbox and xPC Toolbox
DISCUSSION
The results from the experimental testing of the two interfaces were very comparable. The performance of the xPC and Real Time Windows Targets met or outperformed that of the analog board in comparison with the transient data. The use of the PCI-DAS1602/16 data acquisition card allows the student to import data from the experiment directly into MATLAB as opposed to gathering transient data from screenshots on the oscilloscope. This allows data analysis to be much more efficient and user friendly. Both targets provided a good design solution to the problem of developing a digital interface to control the 33-100 Servo DC Motor. There was not a noticeable difference in performance of the two targets for the application of the undergraduate Control Systems Design laboratory.
The test comparing the performance of the Real Time Windows Toolbox to the xPC Toolbox showed that the xPC Toolbox is more robust. The xPC Toolbox setup is only limited by the processing speed of the target computer system. This can be seen in Table 6 where the maximum task execution time is approximately the same for all the different sampling frequencies. The Real Time Windows Toolbox gets overloaded with the higher sampling frequencies. The Electrical Engineering Department has to make a decision based upon the results gathered in this work. The optimal solution may lie in purchasing the PCI-DAS1602/16 cards from Measurement Computing for each lab station in the Control Systems lab. Each adjacent lab PC could then be connected with an RS-232 cable. This setup would allow for both targets to be used in any application and alleviate the need for additional standalone computers to be present for the xPC Target.
RECOMMENDATIONS FOR FUTURE WORK
Future work on this design provides some potentially interesting projects. The interface board can be redesigned in order to control other systems. This could be used in other courses taught at the Rochester Institute of Technology.
ACKNOWLEDGMENTS
The Control Systems Interface Improvement team would like to thank their advisor/coordinator Dr. D. Phillips and advisor Dr. A. Mathew for their invaluable input. The team would also like to thank Ken Snyder and Jim Stefano of the EE Department for their assistance in setting up and acquiring equipment for use on the project.
![[PPT]PowerPoint Presentationedge.rit.edu/edge/OldEDGE/public/Archives/P06201/PDR.ppt · Web viewTitle PowerPoint Presentation Last modified by brl4482 Created Date 1/1/1601 12:00:00](https://static.documents.pub/doc/80x56/5ab71b827f8b9ab7638e764e/pptpowerpoint-viewtitle-powerpoint-presentation-last-modified-by-brl4482-created.jpg)
