1. P Control
   1. No perfect tracking
      1. for type 0
   2. Convergence
      1. Dictated by Gain
      2. Noise Sensitive
2. I Control
   1. Requires 1 pole at Origin
      1. for Close Loop System
   2. Use with P control
   3. 2 Gains
   4. Reduces ss error
3. D Control
   1. Speeds up P.I. Control
      1. If transient control is too slow
   2. Step changes blow out
   3. Amplifies high frequency noise
   4. High overshoot from step input
4. Ziegler-Nichols
   1. Quarter Decay Ratio Method
      2. Gains
         1. P Control
            1. 1/RL
         2. PI Control
            1. 0.9/RL
            2. L/0.3
         3. PID Control
            1. 1.2/RL
            2. 2L
            3. 0.5L
   2. Ultimate Sensitivity
      1. Topic
         1. Critical stability for Step/Pulse input
         2. Measure Gain Ku
      2. Gains
         1. P Control
            1. 0.5Ku
         2. PI Control
            1. 0.45Ku
            2. Pu/1.2
         3. PID Control
            1. 0.6Ku
            2. 0.5Pu
            3. Pu/8
   3. Best for 1st and 2nd order systems
5. Derivative Filtering
   1. Acts as low-pass filter
   2. Apply derivative action to output only
      1. Bypassing step and high frequency inputs
   3. I action must be applied to input errors for effect
   4. P action can also be applied to output only
   5. I-PD
      1. Response to disturbances and noise same
      2. better response to step change
6. Set-point Weighting
   1. PI-D structure
   2. P action only applied to fraction of some set-point R
   3. Controls posistion of zero
   4. Feed forward effect
7. Nest Control
   1. Embedding loops within each other
   2. Slower loops set level for faster loop
   3. Most effective for fast inner loops
   4. Knowledge = Power-Trade Offs
      1. More Accurate Control
         1. less overshoot
      2. Do not require full PID on each loop
      3. Faster loops have higher gain
      4. Requires more computation
         1. effectively n controllers
      5. Requires more system measurement
      6. Not true state feedback
         1. making system higher order through integrators
8. Integrator Wind-up
   1. I keeps controller growing during saturated contrller
      1. Longer time for response to change in error sign
   2. Anti-wind-up
      1. For PI Control
      2. Shuts integrator off during control saturation
      3. Non-linear effect
9. Estimator Design
   1. Poles should be 2 - 6 x fater than plant dynamics
   2. requires information of plant sensor noise
      1. much noise requires slower poles for better measurement
      2. much noise and low fidelity model speed up dynamics for better transient response
   3. Trade off between trusting plant and system noise
10. Emulator Design
    1. High Accuracy
    2. High Cost
    3. Highly sensitive to system noise
    4. Requires accurate model
    5. Requires high sampling rate
       1. fs>30fn
       2. Direct Digital only requires fs>5fn
11. Feed Forward
    1. Nx
       1. Converts desired value of y into values of states
    2. Nu
       1. Allows for steady state control input
       2. Uss=Nu * rss
    3. For set point tracking
    4. Assume steady state
       1. Integral would reduce ss error
12. Model Predictive Control
    1. Needs all internal information
    2. Uses all available data
    3. Can have "forgetting function"
       1. If system has transient response
          1. warming up
       2. Receding horizon
    4. Requires the most information