Nyquist Stability Criterion

Review

Sorry, your browser does not support SVG.

  • One Fundamental Question: Given transfer functions \(G(s),\,H(s)\), how to determine if the system is closed-loop stable?
    • Time Domain: Routh-Hurwitz stability criterion, Root Locus
    • Frequency Domain: Nyquist Plot + Nyquist Stability Criterion

Contour and Plot

  • The open-loop transfer function \(G(s)H(s)\) is a mapping from complex \(s\)-plane to complex \(GH\)-plane.
  • A closed-path (contour) on \(s\)-plane will be mapped to \(w = G(s)H(s)\) to a closed path (plot) on \(GH\)-plane.

Sorry, your browser does not support SVG.

Zeros and Poles

Consider the following transfer function \[G(s)H(s) = \frac{s}{s^2+2s+2} = \frac{s}{(s+1+j)(s+1-j)}.\] Sorry, your browser does not support SVG.

Observation

  • If a clockwise contour does not encircle zeros nor poles, then the plot will not encircle the origin.
  • If a clockwise contour encircles a zero, then the plot will encircle the origin clockwise once.
  • If a clockwise contour encircles a pole, then the plot will encircle the origin counterclockwise once.
  • Cauchy's Argument Principle: If a clockwise contour encircles \(Z\) zeros and \(P\) poles, then the number of clockwise encirclements of the origin \(N\), is given by

\[N = Z - P\Rightarrow Z = N+P.\]

Counting Closed-Loop Poles

Sorry, your browser does not support SVG.

  • The closed-loop transfer function is given by

\[\frac{G(s)}{1+G(s)H(s)}\]

  • Define \(F(s) = 1+G(s)H(s)\), we notice that
    • The poles of \(F(s)\) is the poles of \(G(s)H(s)\)
    • The zeros of \(F(s)\) will be poles for the close loop transfer function.

Counting Closed-Loop Poles using Argument Principle

  • By Cauchy's Argument Principle, for a clockwise contour \(\Gamma\) on the \(s\)-plane:

\[Z = N+P\]

  • \(Z\) is the number of zeros of \(F\), i.e., number of closed-loop poles in the contour
  • \(P\) is the number of poles of \(F\), i.e., number of open-loop poles in the contour
  • \(N\) is the number of clockwise encirclements of the origin for the plot \(F(\Gamma)\)
  • \(N\) is also the number of clockwise encirclements of \(-1\) for the plot \(G\circ H(\Gamma)\).

Example

  • Consider the following transfer function:

\[ G(s)H(s) = \frac{s}{s^2+2s+2}.\]

  • It has two open-loop poles at \(-1\pm j\).
  • \[F(s) = \frac{s^2+3s+2}{s^2+2s+2}\] has two zeros (close-loop poles) at \(-1\) and \(-2\).
  • The contour encircles 2 poles. \(P = 2\).
  • The \(F\)-plot encircles origin counterclockwise once. \(N = -1\).
  • Therefore, \(Z = N+P = 1.\)

Sorry, your browser does not support SVG.

Sorry, your browser does not support SVG.

Nyquist Contour \(\Gamma_n\)

  • The stability of system is related to whether there exists any closed-loop poles (or zeros of \(F(s)\)) on the Right Half Plane (RHP).
  • We select a contour (Nyquist contour) consisting of
    • Segment 1: The imaginary axis from 0 to \(+j\infty\).
    • Segment 2: A semicircle of infinite radius that encloses the entire right half \(s\)-plane.
    • Segment 3: The imaginary axis from \(-j\infty\) to \(0\).
  • The Nyquist Contour is a ‘big’ semicircle that encloses the RHP. The direction of the encirclement is clockwise.

Nyquist Contour \(\Gamma_n\)

Sorry, your browser does not support SVG.

Nyquist Stability Criterion

  • By the Cauchy’s Principle of Argument: \[Z = N+P.\]
    • \(Z\): number of unstable closed-loop poles (zeros of \(F\))
    • \(P\): number of unstable open-loop poles (poles of \(F\))
    • \(N\): number of clockwise encirclements of \(-1\) on the \(GH\)-plane for the plot \(G\circ H(\Gamma)\).
  • The closed-loop system is stable, i.e. \(Z = 0\), when \(N = -P\).

Nyquist Stability Criterion

  • A feedback system is stable if and only if \(N=-P\), i.e. the number of the counterclockwise encirclements of \(–1\) point by the Nyquist plot in the \(GH\)-plane is equal to the number of the unstable poles of the open-loop transfer function.

Implication of Nyquist Stability Criterion

  • If the open-loop system is stable(\(P=0\)), the closed-loop system is stable if and only if the Nyquist plot does not encircle \(–1\) point
  • If the open-loop system has \(P\) unstable poles, the closed-loop system is stable if and only if the Nyquist plot encircles \(–1\) point \(P\) times counterclockwise.
  • If the Nyquist plot passes through \(-1\), then the system has a closed-loop pole on the imaginary axis (critically stable).

Procedure for Determining Stability using Nyquist Stability Criterion

  • Draw the Nyquist Plot
  • Determine the clockwise encirclement \(N\).
  • From the open-loop transfer function, find the number of unstable open-loop poles (\(P\))
  • Stable if \(N = -P\).

How to Sketch the Nyquist Plot

Nyquist Contour consists of 3 segments:

  • Segment 1: The imaginary axis from 0 to \(+j\infty\).
  • Segment 2: A semicircle of infinite radius that encloses the entire right half \(s\)-plane.
  • Segment 3: The imaginary axis from \(-j\infty\) to \(0\).

Sorry, your browser does not support SVG.

Segment 1

  • We use Bode plot to help us sketch the first segment.
  • We need to find 4 types of point:
    • \(\omega = 0\);
    • Real intersection: Phase = \(180N^\circ\);
    • Imaginary intersection: Phase = \(180N+90^\circ\);
    • \(\omega = +\infty\);
  • We can also deduce the trend of the plot around those points:
    • If the phase is decreasing, the plot goes clockwise
    • If the phase is increasing, the plot goes counterclockwise
  • Plot those points on the \(GH\)-plane and draw a smooth line to connect them.

Segment 2

  • Consider the following open-loop transfer function:

\[G(s)H(s) = \begin{cases}\frac{s}{s^2+2s+2}&\text{strictly proper}\\\frac{s-1}{s+1}&\text{proper}\\s&\text{non-proper}\end{cases}\]

  • For strictly proper function, the order of the denominator is greater than the order of the numerator:

\[\lim_{s\rightarrow\infty}\frac{s}{s^2+2s+2} = \lim_{s\rightarrow\infty}\frac{s}{s^2} = 0\]

Segment 2

  • For proper function, the order of the denominator is no less than the order of the numerator:

\[\lim_{s\rightarrow\infty}\frac{s-1}{s+1} = \lim_{s\rightarrow\infty}\frac{s}{s} = 1\]

  • Segment 2 is the origin point for strictly proper function. It is a constant for proper function.
  • Non-proper transfer functions are not physically realizable.

Segment 3

  • From the property of Laplace transform:

\[G(s) = \overline{G(-s)}.\]

  • Therefore, segment 3 is the mirror reflection of segment 1 around the real axis.

Example 1

Consider the system with open-loop transfer function: \[G(s)H(s) = \frac{1}{(s+1)(0.1s+1)}.\] Determine the stability of the closed-loop system using the Nyquist stability criterion.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Sketch Nyquist Plot

  • Segment 1:
    • When \(\omega \rightarrow 0\), \(G(j\omega)H(j\omega)\rightarrow 1\).
    • There is no real intersection for \(0 < \omega < \infty\).
    • There is a imaginary intersection when \(\omega \approx 3\). The intersection is around \(0.3\angle -90^\circ\). (\(0.3\approx -10dB\))
    • More precisely, the intersection is at \(0.287\angle -90^\circ\) and the corresponding frequency is \(\omega = \sqrt{10}\).
    • When \(\omega \rightarrow \infty\), \(G(j\omega)H(j\omega)\rightarrow 0\).
    • The phase is always decreasing, therefore the plot goes clockwise.
  • Segment 2: Since the system is strictly proper, Segment 2 is the origin.
  • Segment 3: Mirror reflection of segment 1.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Finding \(N\) and \(P\)

  • The Nyquist plot does not encircle \(-1\). Therefore \(N = 0\).
  • The open-loop poles are \(-1\), \(-10\). Therefore \(P = 0\).
  • \(Z = N + P = 0\). The closed-loop system is stable.

Example 2

Consider a feedback system with open-loop transfer function \[G(s)H(s) = \frac{K(s-1)}{s^2+s+4},\,K>0\] Determine the range of \(K\) such that the feedback system is stable.

Nyquist Plot

Assume \(K=1\) first.

Sorry, your browser does not support SVG.

Nyquist Plot

  • Segment 1:
    • When \(\omega \rightarrow 0\), \(G(j\omega)H(j\omega)\rightarrow -0.25\).
    • There is a imaginary intersection when \(1<\omega <2\). The intersection is between \(0.1j\) and \(j\).
    • There is a real intersection when \(\omega \approx 2\). The intersection is around \(1\).
    • When \(\omega \rightarrow \infty\), \(G(j\omega)H(j\omega)\rightarrow 0\).
    • The phase is always decreasing. Therefore the plot goes clockwise.
  • Segment 2: Since the system is strictly proper, Segment 2 is the origin.
  • Segment 3: Mirror reflection of segment 1.

Nyquist Plot

Sorry, your browser does not support SVG.

Determine \(N\) and \(P\)

  • For the open-loop system, the poles are at \(-0.5\pm 1.94j\). Therefore, \(P = 0\)
  • If \(K < 4\), then \(-0.25K > -1\), the Nyquist plot does not encircle \(-1\). Therefore \(N = 0\) and the system is closed-loop stable.
  • If \(K > 4\), then \(-0.25K < -1\), the Nyquist plot encircle \(-1\) clockwise once. Therefore \(N = 1\) and \(Z = 1\). There is one unstable pole for the closed-loop system.

Example 3

Consider the two loops feedback system:

Sorry, your browser does not support SVG.

Determine the range of gain \(K\) for stability of the system using Nyquist stability criterion.

Sketch Nyquist Plot

  • We first compute the closed-loop transfer function of the inner loop.

\[G_2(s) = \frac{1/(s^3+s^2)}{1+1/(s^3+s^2)} = \frac{1}{s^3+s^2+1}.\]

  • The open-loop transfer function is

$$ G(s) = \frac{K(s+0.5)}{s^3+s^2+1}.$$

Sorry, your browser does not support SVG.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Sketch Nyquist Plot

  • Segment 1:
    • When \(\omega \rightarrow 0\), \(G(j\omega)H(j\omega)\rightarrow 0.5K\).
    • There is a imaginary intersection when $ω ≈ 0.7 $. The intersection is between \(0.1j\) and \(j\).
    • There is a real intersection when \(1 <\omega < 2\). Calculation shows that the exact frequency is \(\omega = \sqrt{2}\) and the crossing is at \(0.5K\angle -180^\circ\).
    • When \(\omega \rightarrow \infty\), \(G(j\omega)H(j\omega)\rightarrow 0\). Calculation shows that the exact frequency is \(\omega = 1/\sqrt{2}\) and the crossing is at \(\sqrt{2}K\angle -270^\circ\).
    • The phase is increasing around above points. Therefore the plot goes counterclockwise around the above points.

Sketch Nyquist Plot

  • Segment 2: Since the system is strictly proper, Segment 2 is the origin.
  • Segment 3: Mirror reflection of segment 1.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Determine \(N\) and \(P\)

  • For the open-loop system, the poles are at \(-1.47,\,0.23\pm0.79j\). Therefore, \(P = 2\)
  • If \(K < 2\), then \(-0.5K > -1\), the Nyquist plot does not encircle \(-1\). Therefore \(N = 0,\,Z = 2\). The system has 2 unstable poles.
  • If \(K > 2\), then \(-0.5K < -1\), the Nyquist plot encircle \(-1\) counterclockwise twice. Therefore \(N = -2\) and \(Z = 0\). The system is stable.

Example 4

Consider a feedback system with open-loop transfer function \[G(s)H(s) = \frac{K}{(s-1)(s+2)},\,K>0\] Determine the range of \(K\) such that the feedback system is stable.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Sketch Nyquist Plot

Sketch Nyquist Plot

  • Segment 1:
    • When \(\omega \rightarrow 0\), \(G(j\omega)H(j\omega)\rightarrow -K/2\).
    • There is no real intersection for \(0 < \omega < \infty\).
    • There is no imaginary intersection for \(0 < \omega < \infty\).
    • When \(\omega \rightarrow \infty\), \(G(j\omega)H(j\omega)\rightarrow 0\).
    • The phase is always increasing at \(\omega = 0\), therefore the plot goes counterclockwise at \(-K/2\).
  • Segment 2: Since the system is strictly proper, Segment 2 is the origin.
  • Segment 3: Mirror reflection of segment 1.

Sketch Nyquist Plot

Sorry, your browser does not support SVG.

Determine \(N\) and \(P\)

  • For the open-loop system, the poles are at \(1,\,-2\). Therefore, \(P = 1\)
  • If \(K < 2\), then \(-0.5K > -1\), the Nyquist plot does not encircle \(-1\). Therefore \(N = 0,\,Z = 1\). The system has 1 unstable poles.
  • If \(K > 2\), then \(-0.5K < -1\), the Nyquist plot encircle \(-1\) counterclockwise once. Therefore \(N = -1\) and \(Z = 0\). The system is stable.

Summary

  • Cauchy's Argument Principle: \(Z = N+P\).
  • Nyquist Stability Criterion: Closed-loop system is stable if and only if \[N+P = 0.\]
  • Sketch Nyquist plot from Bode plot
  • Determine stability of the closed-loop system using Nyquist stability criterion.