Skip to main content

Discrete output C(z) with feedback H(s)

Short answer

C(z) = G₂(z) G₁R(z) / [1 + G₁G₂H(z)], where G₁R(z)=Z[G₁(s)R(s)] and G₁G₂H(z)=Z[G₁(s)G₂(s)H(s)]. No closed-loop transfer function C(z)/R(z) exists because the input passes through G₁ before the sampler.

Find the discrete time output C(z) of the following closed loop system. [Figure: R(s) → summing junction (+) → E(s) → G₁(s) → sampler → G₂(s) → C(s) → sampler → C(z); C(s) is fed back through H(s) as B(s) to the negative input of the summing junction]

There is no sampler on the error signal. The only sampler is after G1(s)G_1(s), so let its output be V∗(s)V^*(s), the sampled version of V(s)=G1(s)E(s)V(s) = G_1(s)E(s). The output sampler only reads C(z)C(z); it is not inside the loop.

       E(s)  +-----+ V   _/ V*  +-----+  C(s)   _/
R(s)-->(S)-->| G1  |---o/ o--->| G2  |---+---o/ o--> C(z)
       ^ -   +-----+     T     +-----+   |    T
       |            +-----+              |
       +------------|  H  |<-------------+
            B(s)    +-----+

Step 1: Write the signal equations

C(s)=G2(s)V∗(s)E(s)=R(s)−H(s)C(s)=R(s)−G2(s)H(s)V∗(s)V(s)=G1(s)E(s)=G1(s)R(s)−G1(s)G2(s)H(s)V∗(s)\begin{aligned} C(s) &= G_2(s)V^*(s) \\ E(s) &= R(s) - H(s)C(s) = R(s) - G_2(s)H(s)V^*(s) \\ V(s) &= G_1(s)E(s) = G_1(s)R(s) - G_1(s)G_2(s)H(s)V^*(s) \end{aligned}

Step 2: Take starred (sampled) transforms

Sampling both sides, a starred quantity can be taken outside: [G1G2HV∗]∗=(G1G2H)∗V∗[G_1G_2HV^*]^* = (G_1G_2H)^*V^*. This gives

V∗(s)=(G1R)∗(s)−(G1G2H)∗(s)V∗(s)V∗(s)=(G1R)∗(s)1+(G1G2H)∗(s)\begin{aligned} V^*(s) &= (G_1R)^*(s) - (G_1G_2H)^*(s)V^*(s) \\ V^*(s) &= \frac{(G_1R)^*(s)}{1 + (G_1G_2H)^*(s)} \end{aligned}

Step 3: Output

C∗(s)=G2∗(s)V∗(s)C^*(s) = G_2^*(s)V^*(s)

In z-notation:

C(z)=G2(z) G1R(z)1+G1G2H(z)C(z) = \frac{G_2(z)\,G_1R(z)}{1 + G_1G_2H(z)}

where

  • G2(z)=Z[G2(s)]G_2(z) = \mathcal{Z}[G_2(s)]
  • G1R(z)=Z[G1(s)R(s)]G_1R(z) = \mathcal{Z}[G_1(s)R(s)]
  • G1G2H(z)=Z[G1(s)G2(s)H(s)]G_1G_2H(z) = \mathcal{Z}[G_1(s)G_2(s)H(s)]

Note: The input R(s)R(s) reaches the first sampler only after passing through G1(s)G_1(s), so it cannot be separated from G1G_1. You can find the output C(z)C(z) for a given input, but no closed-loop pulse transfer function C(z)/R(z)C(z)/R(z) exists for this system.

✓ Answer checked by Claude Opus 5.5 (high effort)

All Digital Control System chapter 3 questions →