Skip to main content

Response sequence with digital PI controller

Short answer

With Kronecker delta input, c(0)=0. For k≥1, c(k)=0.9448(0.8836)^{k-1}cos[0.8867(k-1)-1.0483] − 0.0301(0.8096)^{k-1}. The response is a damped oscillation that decays to zero.

Obtain in a closed form the response sequence c(kT) of the system as shown in figure below when subjected to a Kronecker delta input r(k). Assume that the sampling period T is 1 second. Consider Kₚ=1 and KI=0.2 of PI controller. [Figure: unity-feedback loop: r(t) → summing junction → e(t) → sampler (T = 1) → e(kT) → digital PI controller GD*(s) → ZOH Gₕ(s) → 1/s(s+1) → C(t)]

Step 1: Pulse transfer functions

PI controller (KP=1K_P = 1, KI=0.2K_I = 0.2):

GD(z)=KP+KI1−z−1=1.2−z−11−z−1=1.2z−1z−1G_D(z) = K_P + \frac{K_I}{1-z^{-1}} = \frac{1.2 - z^{-1}}{1-z^{-1}} = \frac{1.2z - 1}{z-1}

ZOH + plant 1/[s(s+1)]1/[s(s+1)] with T=1T = 1:

G(z)=(1−z−1) Z[1s2(s+1)]=0.3679z+0.2642(z−1)(z−0.3679)G(z) = (1-z^{-1})\,\mathcal{Z}\left[\frac{1}{s^2(s+1)}\right] = \frac{0.3679z + 0.2642}{(z-1)(z-0.3679)}

Step 2: Closed-loop pulse transfer function

GDG=(1.2z−1)(0.3679z+0.2642)(z−1)2(z−0.3679)=0.4415z2−0.0508z−0.2642z3−2.3679z2+1.7358z−0.3679C(z)R(z)=0.4415z2−0.0508z−0.2642z3−1.9264z2+1.6850z−0.6321\begin{aligned} G_DG &= \frac{(1.2z-1)(0.3679z+0.2642)}{(z-1)^2(z-0.3679)} = \frac{0.4415z^2 - 0.0508z - 0.2642}{z^3 - 2.3679z^2 + 1.7358z - 0.3679} \\ \frac{C(z)}{R(z)} &= \frac{0.4415z^2 - 0.0508z - 0.2642}{z^3 - 1.9264z^2 + 1.6850z - 0.6321} \end{aligned}

Step 3: Kronecker delta input

For r(0)=1r(0) = 1 and r(k)=0r(k) = 0 for k≠0k \neq 0, we have R(z)=1R(z) = 1, so C(z)C(z) equals the closed-loop pulse transfer function.

The closed-loop poles are the roots of z3−1.9264z2+1.6850z−0.6321=0z^3 - 1.9264z^2 + 1.6850z - 0.6321 = 0:

p1=0.8096,p2,3=0.5584±j0.6848=0.8836 ∠±0.8867 rad (±50.80∘)p_1 = 0.8096,\qquad p_{2,3} = 0.5584 \pm j0.6848 = 0.8836\,\angle \pm 0.8867\ \text{rad}\ (\pm 50.80^\circ)

All the poles are inside the unit circle, so the system is stable.

Step 4: Inverse z-transform (residue method)

c(k)=∑Res[C(z)zk−1]=∑iAi pi k−1,Ai=N(pi)D′(pi)c(k) = \sum \text{Res}\left[C(z)z^{k-1}\right] = \sum_i A_i\,p_i^{\,k-1}, \qquad A_i = \frac{N(p_i)}{D'(p_i)}
  • A1=−0.0301A_1 = -0.0301 for p1=0.8096p_1 = 0.8096
  • A2,3=0.2358∓j0.4094=0.4724 ∠∓1.0483 radA_{2,3} = 0.2358 \mp j0.4094 = 0.4724\,\angle \mp 1.0483\ \text{rad}

The complex pair combines into one real cosine term:

c(k)=0.9448 (0.8836)k−1cos⁡[0.8867(k−1)−1.0483]−0.0301 (0.8096)k−1,k≥1c(k) = 0.9448\,(0.8836)^{k-1}\cos\big[0.8867(k-1) - 1.0483\big] - 0.0301\,(0.8096)^{k-1}, \quad k \ge 1

and c(0)=0c(0) = 0, because the numerator degree is less than the denominator degree.

Step 5: First few values

Check the closed form against long division or the difference equation c(k)=1.9264c(k−1)−1.6850c(k−2)+0.6321c(k−3)+0.4415r(k−1)−0.0508r(k−2)−0.2642r(k−3)c(k) = 1.9264c(k-1) - 1.6850c(k-2) + 0.6321c(k-3) + 0.4415r(k-1) - 0.0508r(k-2) - 0.2642r(k-3).

kk01234567
c(k)c(k)00.44150.79960.5324−0.0428−0.4739−0.5044−0.2002

For example, at k=1k = 1: 0.9448cos⁡(−1.0483)−0.0301=0.4716−0.0301=0.44150.9448\cos(-1.0483) - 0.0301 = 0.4716 - 0.0301 = 0.4415 (here 0.9448cos⁡1.0483=2 Re(A2)=0.47160.9448\cos 1.0483 = 2\,\text{Re}(A_2) = 0.4716). ✔

Answer: c(kT)=0.9448(0.8836)k−1cos⁡[0.8867(k−1)−1.0483]−0.0301(0.8096)k−1c(kT) = 0.9448(0.8836)^{k-1}\cos[0.8867(k-1) - 1.0483] - 0.0301(0.8096)^{k-1} for k=1,2,3,…k = 1, 2, 3, \dots (with c(0)=0c(0) = 0). This is a damped oscillation that decays to zero.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Digital Control System chapter 3 questions →