1. Joe Clifton Computer Science and Software Engineering University of Wisconsin – Platteville

2. Course Focus  Software-Oriented  Hardware Exposure  Balance Theory & Hands-on  Exposure to a range of Platforms, processor/memory configurations Languages - assembler, high level Testing and debugging methods & tools Analysis, design, process expectations

3. “Overview” Topics  Real-time embedded system examples  Processors and systems  Development languages  Development environments  Platforms and platform standards  RTOSes, RT kernels, tasking shells

4. “Practical” Topics  Simulators,Emulators,Monitors,ICE,BDM  RAM, ROM, EPROM, EEPROM, flash  Timers, counters, interrupts, watch dog  I/O: memory-mapped, port, DMA  Device drivers, UARTs, PPIs  Interfacing and communications

5. “Theoretical” Topics  Reliability, fault tolerance, exception handling  Concurrent programming, tasks, threads  Synchronization and communication  Process and resource scheduling  Resource control: semaphores, monitors  Device inter-processor communications  Real-time UML and OOA&D

6. Student Background  CS/SE 2630 - Object-Oriented Programming and Data Structures II,  CS/SE 3430 - Object-Oriented Analysis and Design, and either  EE 3780 - Introduction to Microprocessors, or  CS 3230 - Computer Architecture and Operating Systems  Senior Year, so usually much more

7. Platforms  PIC 16F84, 16F877A Harvard arch: 1/8 KB Flash, 68/368 B SRAM RISC - 35 instructions, 14 bits each (BTFSC) Freeware assembler, simulator, C compiler  Infineon C515C – 8051 derivative – 32K Keil uVision, C compiler, RTX51 Tiny RTOS  RLC XSCALE – Windows CE Touch Screen, USB, 32MB Flash / 64MB DRAM C# and Visual Studio.NET

8. Control Theory  Added in response to Advisory Board  Limited time available to cover 1 – 2 days  PID Control an appropriate choice Can be comprehended without elaborate theory  8501-lab a natural way to incorporate  Simulated devices cheap way to go!

9. Proportional-Integral-Derivative  Control some quantity Drive to a desired value  error = desired value – current value  correction value applied to controlling device  Proportional – constant times error  Integral – constant times sum of errors  Derivative – constant times change error  Correction value = P + I + D

10. Theoretically Kc = Controller Gain T I = Integral or Reset Time T D = Derivative or Rate Time From Gopal, Control System: Principles and Design

11. Proportional  Constant times error  Constant too small Slow rise time  Constant too large Overshoot  Ringing or Oscillation Overshoot that continues to travel back and forth

12. Integral  Constant times the sum of the errors Each term should be multiplied by the time interval Drive at a constant rate, time factor is a constant  Smoothing out the error: “History”  Longer “below”, larger integration term grows Decreases on overshoot, so smaller overshoot on other side  Can eliminate steady state error  Can grow large, so often needs to be limited

13. Derivative  Constant times the change in error Should be divided by the time interval Drive at a constant rate, time factor is a constant  Instantaneous rate of change of error Slope of Error: “Future”  As actual moves to desired value from one side, P & I have same sign but D has opposite sign  Can decrease overshoot and settling time  Can use: Previous Actual – Current Actual

14. Timing Tolerance  Integral “Sampling rate should vary by no more than 20% over any 10-sample interval”  Derivative “Sampling interval to be consistent to within 1% of the total at all times-the closer the better” From Wescott, “PID Without a PhD”,

15. Algorithm  currentError = desiredValue – currentValue  pTerm = Kp * currentError  iSum = iSum + currentError  if iSum > iSumMax then  iSum = iSumMax  elsif iSum < iSumMin then  iSum = iSumMin  iTerm = Ki * iSum  dTerm = Kd * (currentError – previousError)  previousError = currentError  controlOutput = pTerm + iTerm + dTerm

16. Tuning the Loop  Make a Model and use Tuning Software Use Test equipment, gather data, analyze More time & money than we have allocated for this  A Manual Approach Set Kd, Ki = 0, Kp = 1 or less: will be slow or oscillate Set Kd about 100 times Kp Increase until osc, excessive noise or overshoot ○ Inc /Dec by factor 8/10, then 2/4 when closer Back Kd off by a factor of 2 or 4 Increase Kp until excessive overshoot or Osc Back Kp off by a factor of 2 or 4 Increase Ki to reach desired within required time

17. Project  Control a simulated analog device  Device is simulated using a PIC  Output of device Analog signal in the range of 0 to 5 volts  Control via two discrete inputs Increase / Decrease Voltage The longer a 1 is written, the faster the rate of inc/dec  Multi-tasking environment Tasks: ADC reading, PID loop, Keypad for Desired, and Display, including bit-banged serial for “checking”

18. References  Tim Wescott, “PID Without a PhD”, Embedded Systems Programming, October 2000  M. Gopal, Control System: Principles and Design, McGraw Hill, 2008