Skip to content

Latest commit

 

History

History
52 lines (46 loc) · 3.13 KB

README.md

File metadata and controls

52 lines (46 loc) · 3.13 KB
Raw

PID-FPGA

PID controller on an FPGA with custom RS232 addressing protocol.

alt text

To capture the motor's precise behavior, data sampling speed is synchronized with the TX module. But unfortunately, most of the serial port terminal programs, including Terminal v1.9b by Br@y++ and Termite, failed to keep up with this speed. HTerm was the only one I found that operated with no problems.

Communication Protocol

RS232:

  • Data Bits: 8
  • Stop Bits: 1
  • Parity: None
  • Baud Rate 115200

The instructions are 64-bit. The first and the last 8 bits are must be AA and BB, respectively, to ensure package integrity.
|AA|8bit address|4 bit useless|36Bit data|BB| The Numbers are fixed point 36 bit signed |1bit|26bit|9bit|

  • Accessing to address zero sets the global reset to one.
  • Address one is KP
  • Address two is KD
  • Address three is KI
  • Accessing to other addresses sets the global reset to zero.
    Examples:
    AA00xxxxxxxxxxBB // reset = 1
    AA010000000500BB // KP = 2.5
    AA020000000500BB // KD = 2.5
    AA030000000500BB // KI = 2.5
    AA0D000000FFFFBB // reset = 0

Hardware

  • Altera DE2 FPGA: EP2C35F672C6
  • DC Motor: (unknown)
  • Encoder: ARC S 050
  • Motor Driver: Custom
    alt text

Verification of The Design

  • Quartus project is verified by testbenches.
  • Overall project verified by oscilloscope measurements of the frequency of the encoder signals. The motor speed is 14.15708 RPS (7078.54/500). Which is quite close to the desired speed: 14 RPS.
    alt text
  • Motor's response to disturbances under PID control:
    alt text
    Sudden up and down jumps that created the white halo effect are not measurement errors. A misaligned encoder and motor shaft created them. If you zoom in on the plot on MATLAB, you can see that it is, in fact, a sinusoidal signal.

Possible Design Improvements and Current Flaws

  • Rotatory Encoder creates 500 pulses for every rotation. But, the calculations are done by dividing the pulse count by 512 in order to avoid a separate division circuit.
  • There is a bug in the multiplication circuit. The numbers are ~0.8% off.
  • In my hardware setup, the DC motor and the Encoder are connected via a custom imperfect shaft that causes a sinusoidal noise on the speed values.
  • After the final debugging, I forgot to change the source of the desired speed from switches to RS232. So, you have to enter the desired speed from switches.
  • Current KP, KD, and KI values are quite low. It takes a couple of seconds for the motor to reach its final value.
  • There is no integral windup protection.
  • I am a sophomore student with no Control background, so this is the best I can do in a week.