Nyquist criterion, gain and phase margins
The Nyquist plot and criterion Z = P + N, open-loop unstable plants, gain and phase margins with their crossover frequencies, and the effect of transport delay on phase margin.
Drafted with Aria, reviewed by the AiCanCode.org team. Spotted an error? Use Give Feedback at the bottom of the page.
Why it matters
Routh needs a polynomial, but real machines include transport delays, measured frequency-response data and open-loop unstable plants (a magnetic levitator, an inverted pendulum). The Nyquist criterion decides closed-loop stability from the open-loop frequency response in all these cases, and the gain and phase margins it produces are the standard robustness numbers written into controller specifications.
Key ideas
Nyquist plot. The polar plot of the open-loop frequency response G(jω)H(jω) as ω runs from −∞ to +∞ (the negative-frequency half is the mirror image of the positive half about the real axis). It is the image of the Nyquist contour — the whole jω-axis closed by a large semicircle around the right half-plane, with small detours around any open-loop poles on the jω-axis.
Why −1 matters. Closed-loop poles are the zeros of 1 + G(s)H(s). Encirclements of the origin by 1 + GH are the same as encirclements of the point (−1, j0) by GH. By the argument principle:
Z = P + N
- P: open-loop poles of G(s)H(s) in the right half-plane (known from the model).
- N: net number of clockwise encirclements of (−1, j0) by the full Nyquist plot (anticlockwise counts negative).
- Z: closed-loop poles in the right half-plane. Stable if and only if Z = 0.
So for an open-loop stable system (P = 0) the closed loop is stable if the plot does not encircle −1. For an open-loop unstable system the plot must encircle −1 anticlockwise P times (N = −P).
Poles at the origin. Type-N systems start at infinite magnitude. The small detour around s = 0 maps to N clockwise half-circles of infinite radius that close the plot — needed to count encirclements correctly.
Relative stability — margins. For minimum-phase, open-loop stable systems, how close the plot comes to −1 measures robustness.
- Phase crossover frequency ω_pc: where ∠GH = −180° (the plot cuts the negative real axis).
- Gain margin GM = 1/|GH(jω_pc)|, or in dB GM = −20·log₁₀|GH(jω_pc)|. It is the factor by which the gain can be raised before instability. If the plot cuts the negative real axis at −a (a < 1), GM = 1/a.
- Gain crossover frequency ω_gc: where |GH| = 1 (0 dB; the plot cuts the unit circle).
- Phase margin PM = 180° + ∠GH(jω_gc). It is the extra phase lag (for example from a delay) that would bring the system to the verge of instability.
- Positive GM (dB) and positive PM ⇒ stable (for minimum-phase systems). Typical design targets: GM ≥ 6 dB and PM between 30° and 60°.
- First- and second-order systems without delay never reach −180°, so their GM is infinite.
Link to time response. For a dominant second-order loop, PM ≈ 100ζ (degrees) for PM up to about 60°, so a 45° PM corresponds to ζ ≈ 0.45 and roughly 20 % overshoot.
Delay. A delay e^(−sT_d) rotates every point of the plot clockwise by ωT_d radians without changing its magnitude. It reduces PM by ω_gc·T_d (converted to degrees), so the largest tolerable delay is PM/ω_gc (PM in rad).
Formulas
Z = P + N — Nyquist criterion (N clockwise positive).
∠G(jω_pc)H(jω_pc) = −180° — phase crossover.
GM = 1/|G(jω_pc)H(jω_pc)|; GM_dB = −20·log₁₀|G(jω_pc)H(jω_pc)|
|G(jω_gc)H(jω_gc)| = 1 — gain crossover.
PM = 180° + ∠G(jω_gc)H(jω_gc) (degrees)
T_d,max = PM(rad)/ω_gc — largest added delay before instability (s); ω_gc in rad/s.
PM ≈ 100·ζ — rough rule for a dominant second-order loop, PM ≤ about 60°.
Worked examples
Example 1 (standard — margins). A unity-feedback loop has G(s) = 12/((s + 1)(s + 2)(s + 3)). Find ω_pc, GM, ω_gc and PM, and state whether the closed loop is stable.
- P = 0 (all open-loop poles in the LHP).
- Phase: ∠G = −[tan⁻¹ω + tan⁻¹(ω/2) + tan⁻¹(ω/3)]. Setting this to −180° gives ω² = 11, so ω_pc = 3.32 rad/s (check: the angles 73.2° + 58.9° + 47.9° = 180°).
- |G(jω_pc)| = 12/√((1 + 11)(4 + 11)(9 + 11)) = 12/√3600 = 12/60 = 0.2. The plot cuts the negative real axis at −0.2.
- GM = 1/0.2 = 5 = 13.98 dB.
- Gain crossover: (1 + ω²)(4 + ω²)(9 + ω²) = 144, solved numerically: ω_gc = 1.22 rad/s.
- ∠G(jω_gc) = −(50.7° + 31.4° + 22.2°) = −104.4°, so PM = 75.6°.
- The plot does not encircle −1 (N = 0), so Z = 0: stable. The gain could be raised fivefold, to 60, before instability.
Example 2 (GATE level — open-loop unstable plant). A unity-feedback loop has G(s) = K/(s − 1), K > 0. Use the Nyquist criterion to find the range of K for stability.
- Open-loop pole at s = +1, so P = 1. Stability needs Z = 0, i.e. N = −1: one anticlockwise encirclement of −1.
- G(jω) = K/(jω − 1) = K(−1 − jω)/(1 + ω²).
- ω = 0: G = −K. ω → ±∞: G → 0. For ω > 0 the imaginary part is negative, for ω < 0 positive.
- The plot is a circle with diameter from −K to 0 (centre −K/2, radius K/2). Going from ω = −∞ to +∞, it passes from the origin through the upper half to −K and back through the lower half: an anticlockwise circle.
- If K > 1 the circle contains −1 and is traversed once anticlockwise: N = −1, Z = 1 − 1 = 0, stable. If K < 1 it does not contain −1: N = 0, Z = 1, unstable.
- Stable for K > 1. Check: closed-loop characteristic equation s − 1 + K = 0 gives the pole s = 1 − K, which is negative only for K > 1.
Common mistakes
- Using "no encirclement means stable" for an open-loop unstable plant; always find P first.
- Counting encirclements of the origin instead of −1.
- Mixing the sign conventions: in Z = P + N, N counts clockwise encirclements positive.
- Reading GM from |GH| at the gain crossover, or PM at the phase crossover — they are the other way round.
- Quoting 10 dB GM as a tenfold gain increase: 10 dB is a factor of 10^(10/20) = 3.16.
- Applying GM/PM rules to non-minimum-phase or open-loop unstable systems without checking with the full Nyquist plot.
For GATE ME
Expect: GM from the point where a polar plot crosses the negative real axis; PM from the phase at gain crossover; phase crossover of a third-order type-0 or type-1 loop; Z = P + N counting for a given sketch; the effect of delay on phase margin; and the range of K for stability using the Nyquist plot. Practise solving tan⁻¹ sums to −180° quickly — for (s + a)(s + b)(s + c) type-0 loops the crossover is at ω² = ab + bc + ca.
Quick check
- The Nyquist plot crosses the negative real axis at −0.5. What is GM in dB?
- The open-loop phase at gain crossover is −120°. What is PM?
- P = 0 and the plot encircles −1 twice clockwise. How many closed-loop RHP poles are there?
- What gain factor does a GM of 20 dB allow?
- A loop has PM = 45° at ω_gc = 2 rad/s. What added delay makes it marginally stable?
Answers: 1. 6.02 dB. 2. 60°. 3. Two. 4. A factor of 10. 5. 0.785 rad / 2 rad/s = 0.393 s.
Interview questions
All Control Systems interview questionsTry answering each one aloud before you open it.
1.What is the Nyquist criterion in control systems?Concept
The Nyquist criterion is a graphical method used in control systems to determine the stability of a closed-loop system. It involves plotting the Nyquist plot, which is a frequency response plot of the open-loop transfer function, and analyzing the encirclements of the critical point (-1,0) in the complex plane. The criterion helps in assessing whether the system will remain stable or become unstable when feedback is applied.
2.Explain the significance of gain margin and phase margin in control systems.Concept
Gain margin and phase margin are measures of the stability of a control system. Gain margin indicates how much the system gain can increase before the system becomes unstable, while phase margin indicates how much the phase can decrease before instability occurs. A higher gain margin and phase margin generally imply a more stable system with better tolerance to parameter variations and disturbances.
3.How do you determine the gain margin from a Nyquist plot?Application
To determine the gain margin from a Nyquist plot, locate the point where the Nyquist plot crosses the negative real axis. The gain margin is the reciprocal of the magnitude of the open-loop transfer function at this point. It is usually expressed in decibels (dB) as 20 log10(1/M), where M is the magnitude at the crossing point.
4.Why is the Nyquist criterion preferred over the Bode plot for certain stability analyses?Application
The Nyquist criterion is preferred over the Bode plot when dealing with systems that have right-half plane poles or zeros, as it provides a more comprehensive view of the system's stability by considering the entire frequency range in the complex plane. It is particularly useful for analyzing systems with time delays or non-minimum phase characteristics, where Bode plots might not provide clear stability information.
5.What happens if the Nyquist plot encircles the critical point (-1,0) in the clockwise direction?Application
If the Nyquist plot encircles the critical point (-1,0) in the clockwise direction, it indicates that the closed-loop system is unstable. The number of clockwise encirclements, combined with the number of poles of the open-loop transfer function in the right-half plane, determines the stability of the system according to the Nyquist stability criterion.
6.How can phase margin be determined from a Bode plot?Application
Phase margin can be determined from a Bode plot by finding the frequency at which the gain is 0 dB (the gain crossover frequency). At this frequency, the phase margin is the difference between the phase angle and -180 degrees. A positive phase margin indicates a stable system, while a negative phase margin indicates instability.
7.What is the impact of a negative gain margin on system stability?Application
A negative gain margin indicates that the system is already unstable or very close to instability. It means that the system gain needs to be reduced to achieve stability. Operating with a negative gain margin can lead to oscillations or unbounded output in response to disturbances or changes in input.
8.Calculate the gain margin if the Nyquist plot crosses the negative real axis at -0.5.Numerical
The gain margin is calculated as the reciprocal of the magnitude at the crossing point. Since the Nyquist plot crosses at -0.5, the magnitude is 0.5. The gain margin is 1/0.5 = 2. In decibels, this is 20 log10(2) ≈ 6.02 dB.
9.A system has a phase margin of 30 degrees. What does this imply about the system's stability?Application
A phase margin of 30 degrees implies that the system is stable but may not have a large safety margin against instability. It indicates that the system can tolerate a phase decrease of up to 30 degrees before becoming unstable. While the system is stable, a higher phase margin is generally preferred for better robustness and performance.
10.If a system has a gain margin of 10 dB, what does this indicate about the system's robustness?Application
A 10 dB gain margin means the loop gain can rise by a factor of 10^(10/20) ≈ 3.16 before the closed loop reaches the verge of instability, not tenfold — 20 dB would be tenfold. That is a comfortable margin; designers usually ask for at least about 6 dB (a factor of 2). Gain margin alone is not enough, though: the phase margin must also be adequate, because a loop with a good GM but a small PM still rings badly and is sensitive to delay.
Finished this topic? Mark it so your progress, study plan and readiness keep up.
Stuck on something here?