What Is a Nyquist Plot?

A Nyquist plot is a parametric graph of the frequency response of a linear time-invariant (LTI) system. It plots the real part of the open-loop transfer function G(s)H(s) against its imaginary part as the frequency ω varies from zero to infinity (and often back from zero to negative infinity). Named after the engineer Harry Nyquist, this plot is a cornerstone of control theory and signal processing. Unlike Bode plots, which present magnitude and phase separately, the Nyquist plot consolidates both into a single polar representation, making it especially effective for analyzing closed-loop stability.

The fundamental strength of a Nyquist plot lies in its application of the Nyquist stability criterion. This criterion uses the number of clockwise encirclements of the critical point (−1, 0) to determine whether a closed-loop system is stable. Engineers rely on this graphical technique because it provides a complete picture of the frequency response at a glance, revealing not only stability margins but also the presence of resonance and potential oscillatory behavior. Understanding how to read these plots is essential for designing robust controllers, audio amplifiers, power electronics, and any feedback system where unwanted oscillations can degrade performance or cause failure.

Key Features of a Nyquist Plot and Their Significance

Loop Gain and the Critical Point

The most important feature of a Nyquist plot is its relationship to the point (−1, 0). This point represents the closed-loop gain condition: if the open-loop transfer function equals −1 at some frequency, the closed-loop gain goes to infinity, and the system will oscillate at that frequency. The plot’s trajectory near this point directly indicates stability margins. A plot that passes through (−1, 0) at a specific frequency predicts sustained oscillations. A plot that encircles it in the clockwise direction suggests an unstable closed-loop system, while a plot that does not encircle it indicates stability (provided the open-loop system has no right-half-plane poles).

Direction of the Plot and Phase Information

The direction of the Nyquist contour carries essential phase information. As frequency increases, the plot traces a curve that moves from the positive real axis (at DC gain) toward the origin (at infinite frequency). The direction of the curve reveals whether phase is lagging or leading. For systems with more poles than zeros, the phase eventually approaches −90° per pole-zero excess. A plot that crosses the negative real axis between −1 and 0 indicates positive gain margin; a crossing left of −1 indicates negative gain margin and potential instability.

Nyquist Encirclements and the Stability Criterion

The Nyquist stability criterion states that the number of closed-loop unstable poles is equal to the number of open-loop unstable poles plus the number of clockwise encirclements of the (−1, 0) point. If the open-loop system is stable (no right-half-plane poles), then any clockwise encirclement of the critical point implies an unstable closed-loop system. In practice, even if the plot does not fully encircle −1, the system may still be susceptible to oscillations if the plot approaches the critical point closely. This closeness is quantified by gain margin (the distance from the critical point along the real axis) and phase margin (the angle difference at gain crossover). A gain margin less than 6 dB or a phase margin less than 30° often indicates poor damping and a tendency toward resonant peaks.

Understanding Resonance in Control Systems

What Is Resonance?

Resonance occurs when a system amplifies a particular frequency more than others, often corresponding to the natural frequency of a lightly damped pole. In the frequency domain, resonance appears as a sharp peak in the magnitude response. For example, a second-order system with low damping ratio (ζ) will exhibit a pronounced resonance peak near its undamped natural frequency ωn. This phenomenon can be desirable in tuned circuits or filters but is usually problematic in control systems because it amplifies disturbances and can lead to sustained oscillations at the resonant frequency.

How Resonance Manifests in a Nyquist Plot

In a Nyquist plot, resonance is not directly displayed as a peak but is inferred from the shape of the trajectory. A system with a lightly damped resonant mode will have a Nyquist contour that approaches the (−1, 0) point more closely than a well-damped system. Specifically, as the frequency sweeps through the resonant region, the plot’s magnitude becomes large, and the phase changes rapidly. If the plot comes within a small distance from the critical point, the system exhibits a high sensitivity peak, often quantified by the maximum of the sensitivity function ||S(jω)||∞. This peak directly correlates with the damping ratio: a sensitivity peak above 6 dB (factor of 2) indicates low damping and a high risk of oscillatory behavior.

To identify resonance from a Nyquist plot, look for a region where the curve loops or bulges toward the negative real axis. The closer the loop gets to (−1, 0), the more pronounced the resonance. For instance, if the Nyquist plot passes within a circle of radius 0.5 centered at (−1, 0), the closed-loop system will have a sensitivity peak of at least 6 dB, which is a typical warning threshold. Resonance can also be detected by examining the phase margin: a phase margin below 30° almost always accompanies a resonant peak in the closed-loop response.

Identifying Potential Oscillations

Oscillations and the Nyquist Criterion

Sustained oscillations in a feedback system arise when the closed-loop transfer function has poles on the imaginary axis. This condition is equivalent to the open-loop transfer function equaling −1 at some frequency ωc. On the Nyquist plot, this occurs when the curve intersects the (−1, 0) point. If the intersection happens at a frequency where the plot has positive direction (increasing frequency), the system will oscillate at that frequency. If the plot only approaches but does not exactly cross −1, the system may exhibit decaying or growing oscillations depending on the margin.

Potential oscillations are often indicated by a Nyquist plot that passes very close to the critical point. In such cases, small changes in parameters (e.g., temperature, component aging) can cause the plot to shift and cross (−1, 0), leading to instability. This is why engineers design systems with sufficient gain and phase margins to avoid operating near the critical point. A common rule of thumb is to maintain a gain margin of at least 10 dB and a phase margin of at least 45°, which ensures that even if the plant dynamics vary, the closed-loop system remains stable and free from oscillations.

Encirclements and Clockwise Loops

Any clockwise encirclement of (−1, 0) by the Nyquist plot is a strong indicator of potential oscillations. If the open-loop system is stable, a clockwise encirclement means the closed-loop system has right-half-plane poles, which produce growing exponential signals that eventually become sustained oscillations (limited by nonlinearities). Even if the encirclement is not complete, a large clockwise loop that nearly encloses the critical point suggests that the system is close to the verge of oscillation. This behavior is common in systems with significant phase lag, such as those with large time delays.

For non-minimum-phase systems or systems with pure time delays, the Nyquist plot may exhibit multiple loops or spirals. Each loop that approaches the critical point represents a potential oscillation frequency. Engineers use the Nyquist stability criterion to count encirclements and predict whether the closed-loop response will be stable. In practice, simulation tools generate Nyquist plots and compute stability margins automatically, but understanding how to interpret the plot by hand remains an important skill for troubleshooting and design.

Practical Steps to Analyze a Nyquist Plot for Resonance and Oscillations

Follow these steps to systematically evaluate a Nyquist plot for signs of resonance and potential oscillations:

  1. Plot the Nyquist curve for a range of frequencies from zero to infinity. Include negative frequencies to obtain a closed contour (if the system is proper).
  2. Locate the critical point (−1, 0). Mark this point on the plot.
  3. Examine the proximity to the critical point. Measure the shortest distance from the Nyquist curve to (−1, 0). This distance is inversely related to the sensitivity peak. A distance less than 0.5 indicates a resonance peak greater than 6 dB.
  4. Count clockwise encirclements of (−1, 0). For a stable open-loop system, any clockwise encirclement means closed-loop instability. For unstable open-loop systems, use the Nyquist criterion equation.
  5. Determine gain margin: Find the point where the curve crosses the negative real axis. The gain margin is the reciprocal of the magnitude at that point (if the crossing is to the left of −1, the margin is negative).
  6. Determine phase margin: Identify the frequency where the curve crosses the unit circle (magnitude = 1). The phase margin is 180° plus the phase angle at that frequency.
  7. Check for loops or bulges near the critical point. A loop that closely surrounds (−1, 0) even without fully encircling it suggests low damping and a high risk of oscillatory behavior under parameter variations.
  8. Simulate time-domain response for frequencies near potential resonance points to confirm the presence of oscillations.

These steps provide a quick yet thorough assessment. In many engineering environments, software such as MATLAB or Python (SciPy) is used to generate Nyquist plots and compute margins automatically. However, manual inspection remains valuable for developing intuition and catching subtle features that automated metrics might miss.

Advanced Topics: Sensitivity Peaks and Robustness

Maximum Sensitivity and the Nyquist Plot

The sensitivity function S(s) = 1/(1 + G(s)H(s)) quantifies how errors are propagated through the feedback loop. The largest value of |S(jω)| over all frequencies is called the maximum sensitivity, denoted Ms or ||S||∞. On the Nyquist plot, Ms equals the reciprocal of the shortest distance from the curve to (−1, 0). If the closest distance is d, then Ms = 1/d. A typical design constraint is Ms ≤ 2 (6 dB), which corresponds to d ≥ 0.5. A higher Ms indicates lower damping and more amplification at the resonant frequency, often leading to oscillations or excessive overshoot in the step response.

Engineers use Nyquist-based robustness analysis to ensure that even with modeling uncertainties, the system remains stable and does not develop oscillations. The distance from the Nyquist curve to (−1, 0) serves as a measure of robust stability. A larger distance means the system can tolerate larger gain variations or phase shifts before crossing into instability. This is formalized by the stability margin, which is the maximum allowed additional gain or phase lag that maintains stability.

Relationship with Bode Plots

While Nyquist plots provide a comprehensive view, Bode plots are often used alongside for detailed gain and phase information. A resonance peak in the Bode magnitude plot corresponds to a region in the Nyquist plot where the curve loops near the critical point. The frequency at which the Bode magnitude peaks is the same frequency at which the Nyquist plot is closest to (−1, 0). Combining both plots gives a complete picture: the Nyquist plot reveals stability margins directly, while the Bode plot shows the frequency scale and the shape of the magnitude and phase individually.

For quick design, many engineers first use Bode plots to adjust compensators (e.g., lead-lag filters) to achieve desired margins. Then they verify with a Nyquist plot to ensure no hidden instability from encirclements or multiple loops. This dual-plot approach is standard in industrial control design.

Real-World Examples and Applications

Example 1: Second-Order Resonant System

Consider a simple second-order system with transfer function G(s) = ωn² / (s² + 2ζωns + ωn²). For ζ = 0.2 and ωn = 10 rad/s, the Nyquist plot shows a large loop that swings left of (−1, 0) for certain gains. The distance to (−1, 0) might be only 0.3 at the resonant frequency, giving Ms ≈ 3.3 (over 10 dB). This system is highly likely to oscillate if placed in a feedback loop without compensator. A phase margin of only 15° confirms the risk. By adding damping or reducing gain, the Nyquist plot shrinks away from the critical point, increasing margins and suppressing oscillations.

Example 2: Time-Delay System

Systems with dead time (e.g., chemical processes, networked control) have Nyquist plots that spiral inward. Each time the spiral crosses the negative real axis, there is a potential oscillation frequency. The inner loops may approach (−1, 0) very closely, making the system sensitive to delays. A common method to increase stability is to use a Smith predictor or reduce the loop gain at higher frequencies. Analyzing the Nyquist plot helps decide the required delay margin and gain adjustments.

Example 3: Audio Amplifier Stability

Audio power amplifiers often use feedback to reduce distortion. However, inductive loads (speakers) can introduce phase shifts that cause the Nyquist plot to loop near (−1, 0) at high frequencies. This can lead to parasitic oscillations that damage speakers or amplifiers. Engineers add lead compensation or output inductors to modify the phase response, moving the Nyquist plot away from the critical point. The Nyquist plot is essential for verifying stability across all load conditions.

Conclusion

Nyquist plots provide an intuitive and powerful way to identify resonance and potential oscillations in feedback systems. By examining the proximity and encirclements of the critical point (−1, 0), engineers can assess stability margins, detect lightly damped modes, and predict the likelihood of sustained oscillations. The method is widely used in control system design, from aerospace to consumer electronics. Mastering the interpretation of Nyquist plots—alongside Bode plots and sensitivity analysis—enables engineers to build systems that are both high-performance and robust, avoiding costly redesigns and safety failures. Always verify designs with time-domain simulation, but rely on the Nyquist plot for a deep, frequency-domain understanding of resonance and stability.

For further reading, consult classical control textbooks such as Modern Control Systems by Dorf and Bishop or the original 1932 paper by Harry Nyquist on the work published in the Journal of the Franklin Institute. Additionally, the Nyquist stability criterion article on Wikipedia and the Nyquist Plot tutorial from Swarthmore College offer practical examples and interactive tools.