Skip to content
Tutorials Center
My learning

Topics

Lesson 7 of 37 ยท Modelling and Transfer Functions

Transfer Functions, Poles and Zeros

Define transfer functions, find poles, zeros and DC gain, and see how zeros in the right half-plane cause undershoot, all computed with NumPy and SciPy.

A transfer function is the Laplace-domain description of a linear, time-invariant system: the ratio of the transform of the output to the transform of the input, assuming zero initial conditions.

Text
G(s) = Y(s) / U(s)

If you know G(s) you know how the system reacts to any input: Y(s) = G(s) U(s). This is the object the rest of classical control works with, and it is why block diagrams multiply.

From a differential equation to G(s)

Take the mass-spring-damper from Modelling Physical Systems, m x'' + c x' + k x = F. Transform with zero initial conditions: (m s^2 + c s + k) X(s) = F(s). So

Text
G(s) = X(s) / F(s) = 1 / (m s^2 + c s + k)

In general, G(s) = N(s) / D(s), a ratio of polynomials. The degree of D is the system's order. The system is proper if deg N <= deg D and strictly proper if deg N < deg D (real physical systems are usually strictly proper, because a pure differentiator would amplify noise without limit).

Poles, zeros and gain

Factor the polynomials:

Text
G(s) = K (s - z1)(s - z2)... / ((s - p1)(s - p2)...)
  • Zeros z_i are roots of N(s): values of s where the transfer function is zero.
  • Poles p_i are roots of D(s): where G becomes infinite. They determine the natural modes e^(p t) of the system.
  • Characteristic equation: D(s) = 0.
  • DC gain: G(0), the steady-state output per unit constant input (when the system is stable).

Poles and zeros

Poles Roots of the denominator. Set the modes.
Zeros Roots of the numerator. Shape the response.
Order Number of poles.
DC gain G(0), the long-run ratio.

Poles decide stability; zeros decide shape.

1 / 2

Computing poles and zeros

scipy.signal.tf2zpk returns zeros, poles and gain. Or use np.roots on the coefficient lists. Here is a mildly interesting example, G(s) = (s + 2) / (s^2 + 3 s + 2), whose denominator factors as (s+1)(s+2):

Python Poles, zeros and cancellation

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

A zero at -2 exactly cancels the pole at -2, so the second-order system acts first-order. Cancelling a pole against a zero in the left half-plane is mathematically fine, though it hides a mode that can still be excited by initial conditions. Never cancel an unstable pole against a controller zero: the unstable mode remains inside the system even though the transfer function hides it.

The effect of zeros

Poles are more famous, but zeros shape the response. A left-half-plane zero adds a "lead" and typically increases overshoot. A right-half-plane zero (a non-minimum-phase system) is stranger: the output initially moves in the wrong direction. Backing a long vehicle into a parking space, or a boat's rudder, behave this way, and so does the water level of a boiler drum when the feed rate changes.

Python Undershoot from a right-half-plane zero

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The step response first dips below zero (the minimum is about -0.21 near t = 0.5) before rising to its final value 1. Because the controller cannot undo physics, a non-minimum-phase plant limits how fast any feedback loop around it can be.

From a transfer function to a pole-zero map

A pole-zero map plots poles as crosses and zeros as circles in the complex s-plane. Even without a plot, you can read stability from the pole locations: all poles in the open left half-plane means every mode decays. You will use these maps heavily in Root Locus.

Impulse response and the transfer function

The inverse Laplace transform of G(s) is the system's impulse response g(t). For any input, y(t) is the convolution of g with u. So the transfer function and the impulse response carry the same information in two different forms. The transfer function is easier to combine; the impulse response is easier to interpret in time.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What are the poles of G(s) = 5 / ((s + 2)(s + 3))?

    Poles are roots of the denominator: s + 2 = 0 and s + 3 = 0, giving s = -2 and s = -3.
  2. 2. What is the DC gain of G(s) = 6 / (s^2 + 5 s + 3)?

    G(0) = 6 / 3 = 2.
  3. 3. A right-half-plane zero typically makes the step response

    Non-minimum-phase zeros cause initial undershoot. They do not by themselves make an open-loop system unstable, but they limit achievable feedback performance.
  4. 4. Which system is strictly proper?

    Strictly proper means the numerator degree is smaller than the denominator degree. Only 1/(s+4) qualifies.

Summary

  • A transfer function G(s) = Y(s)/U(s) describes an LTI system with zero initial conditions.
  • Poles (roots of the denominator) decide the modes and stability; zeros shape the response.
  • scipy.signal.tf2zpk and np.roots compute poles and zeros; G(0) is the DC gain.
  • Never cancel an unstable pole with a zero in a controller.
  • Right-half-plane zeros cause undershoot and limit achievable performance.

Next: Block Diagram Reduction