exp(-tau_d*s) = 1 - tau_d*s
--> error is very small when tau_d*s is < 1/2
--> if significant freq components of M(s) are below 0.5tau_d, then no error is introduced

- paper is based on tuning/adjusting controller
- pid is a naturally derived algorithm satisfying the purpose
of control
- effective tuning methods exist, so this paper focuses on making
tuning procedures easier to use
- paper also shows that there is a benefit to rate action (derivative)
- paper shows that PID can be controlled by P and I knobs only
- rate time / reset time (D / I) have a ratio of 0.25 most of the time
- second order lag plus deadtime process
- motorolla model 55rc
- ZN ultimate period formulas use 
- ZN PID, ZN PI, mod ZN PID where latter is the best
- mod ZN PID obtained by modifying ZN PI to include rate time (D) that is
equal to 1/4 of the reset time (I)
- Gp(s) = K*exp(-tau_d*s)/(tau_1*s + 1)(tau_2*s + 1)
- tau_1/tau_2 varied from 0.1 to 1.0
- tau_d / tau_2 varied from 0.2 to 1.0
- proportioning rate time to reset time is applicable
regardless of being electronic penumatic (DNE) or digital controllers
- "... good responses are consistently obtained using the rate-reset time
ratio of one-fourth"

