The standard second-order system and transient specifications
Two numbers, ζ and ωn, fix the poles, and the poles fix the overshoot, the peak time and the settling time.
THE SHORT ANSWER
The standard second-order system is T(s) = ωn²/(s² + 2ζωn s + ωn²): the damping ratio ζ sets the shape of the step response and the natural frequency ωn sets its time scale. For 0 < ζ < 1 the poles are −ζωn ± jωd with ωd = ωn√(1 − ζ²). Overshoot depends only on ζ, peak time on ωd, and settling time on the decay rate ζωn.
Given a second-order transfer function, compute ζ, ωn, the poles, tr, tp, Mp and ts by hand; given Mp and ts specifications, find the required ζ, ωn and pole region.
The previous chapter read rise time, overshoot and settling time off a recorded trace. For the standard second-order system the same numbers can be computed from the transfer function before any experiment, and specifications can be turned back into pole locations.
Review the previous lesson: Read a step responseFirst, picture it
Picture a mass on a spring when a constant force is suddenly applied. With little damping it shoots past its new resting point and rings; with heavy damping it creeps there without ringing. Stiffness and mass set how fast it swings, and damping sets how quickly the swinging dies. The standard second-order form separates these two effects: the natural frequency ωn sets the time scale and the damping ratio ζ sets the shape. For a given ζ every step response is the same curve, stretched or compressed along the time axis by ωn. That is why a designer who hears ζ = 0.5 knows at once that the overshoot is about 16%, whatever the hardware.
Build the idea step by step
Start by writing the transfer function in the standard form ωn²/(s² + 2ζωn s + ωn²). Compare coefficients: the constant term gives ωn, then the s term gives ζ. For 0 < ζ < 1 the poles are a complex pair at −σ ± jωd, with σ = ζωn and ωd = ωn√(1 − ζ²). In the s-plane the pair lies on a circle of radius ωn, at an angle β from the negative real axis with cos β = ζ. Each coordinate controls one specification. The real part σ sets the decay envelope and therefore the settling time. The imaginary part ωd sets the oscillation period and therefore the peak time. The angle β alone sets the overshoot. At ζ = 1 the two poles merge at −ωn, and for ζ > 1 they are real and distinct; in both cases the step response has no overshoot. All formulas in this chapter are for the unit-step response of this exact form with zero initial conditions.
Read the equations
Standard second-order form with unit DC gain. Match your denominator to it: the constant term gives ωn², then the coefficient of s gives ζ. If the numerator is a multiple of ωn², every output value is scaled by that factor and the percent overshoot and all the times stay the same.
Poles for 0 < ζ < 1. The real part −σ sets how fast the oscillation decays; the imaginary part ωd is the frequency you see in the response.
The s-plane picture. The poles lie on a circle of radius ωn centred at the origin, and the line from the origin to a pole makes the angle β with the negative real axis. A smaller β means a larger ζ.
Damping classes: underdamped, critically damped and overdamped. Underdamped responses overshoot and oscillate; the other two rise without overshoot. At ζ = 0 the poles are ±jωn and the oscillation never decays.
Unit-step response for 0 < ζ < 1 with zero initial conditions. It is 1 minus a sinusoid of frequency ωd whose amplitude shrinks along the envelope e^(−ζωn t)/√(1 − ζ²).
Rise time, defined here as the first time the response reaches its final value (0–100%), with β in radians. For overdamped systems the 10–90% definition is used instead.
Peak time: the first maximum, half a damped period after the step. It depends only on ωd, the imaginary part of the poles.
Maximum overshoot, measured from the final value c(∞) and usually quoted in percent. It depends only on ζ: 0.4 gives 25%, 0.5 gives 16%, 0.6 gives 9.5% and 0.7 gives 4.6%.
Settling time. Both formulas are approximations taken from the decay envelope, whose time constant is 1/(ζωn). The 2% band takes about four time constants and the 5% band about three.
The overshoot formula solved for ζ, with Mp as a fraction (0.10 for 10%). Use it to turn an overshoot specification into a minimum damping ratio.
Specifications as s-plane boundaries. An overshoot limit is a wedge around the negative real axis, a settling-time limit is a vertical line, and a peak-time limit is a horizontal line. The poles must lie inside all of them.
Critically damped step response. For a given ωn it is the fastest response of this family that has no overshoot.
A zero at −z, with the DC gain kept at 1, adds a scaled derivative of the original step response. The derivative is positive during the rise, so the response is faster and overshoots more. The effect is small when z is at least 5ζωn.
A third real pole at −p, with the DC gain kept at 1, slows the response and reduces the overshoot. When p is at least five times ζωn the complex pair dominates and the second-order formulas remain good estimates.
What the symbols mean
- damping ratio, dimensionless
- undamped natural frequency, in rad/s
- damped natural frequency, in rad/s
- decay rate ζωn, the distance of the poles from the imaginary axis, in s⁻¹
- angle between the negative real axis and the line from the origin to a pole, in rad
- the two poles of T(s), in s⁻¹
- closed-loop transfer function from reference to output
- Laplace transforms of the reference and the output
- output response to a unit step, in output units
- final value of the output
- rise time (0–100%), in s
- peak time, in s
- maximum overshoot as a fraction of c(∞); multiply by 100 for percent
- settling time for the stated band (2% or 5%), in s
- specification limits: smallest damping ratio, longest settling time, longest peak time
- added zero at s = −z, in s⁻¹
- added third pole at s = −p, in s⁻¹
- transfer function and step response with the added zero
- transfer function with the added third pole
- loop gain of the running example, in s⁻²
The formulas for tr, tp and Mp are exact for the standard form with 0 < ζ < 1. The settling-time formulas are approximations taken from the exponential envelope. Overshoot is measured from the final value c(∞), which equals the reference only when the steady-state error is zero. A zero or a third pole near the dominant pair changes every one of these numbers.
Exact response of the lectures' running example (ζ = 0.5, ωn = 2 rad/s). 10–90 % rise time tr = 0.82 s, peak 1.163 at tp = 1.81 s, so Mp = 16.3 %. From ts = 4.04 s on, the output stays inside ±2 % of the final value.
The running example T(s) = K/(s² + 2s + K) with K = 4
The plant G(s) = K/[s(s + 2)] under unity negative feedback gives the closed loop below. Take K = 4.
Match the denominator to the standard form. The constant term gives ωn; the s coefficient then gives ζ.
Locate the poles. Since 0 < ζ < 1 the system is underdamped.
Find the pole angle in radians, then the rise time and the peak time.
Compute the maximum overshoot. The final value is T(0) = 1, so the peak value is 1.163.
Estimate the settling time from σ. A numerical simulation gives 4.04 s for the 2% band and 2.64 s for the 5% band, so treat these as estimates.
Now raise the gain to K = 16. The real part stays at σ = 1 s⁻¹, so the settling time does not improve. The response peaks sooner and overshoots much more.
Reference tables
| Specification | Formula | Remark |
|---|---|---|
| Poles | |s| = ωn and cos β = ζ | |
| Rise time | from 0 to 100% of the final value | |
| Peak time | time of the first peak | |
| Maximum overshoot | fraction of the final value; depends on ζ only | |
| Settling time (2%) | envelope estimate | |
| Settling time (5%) | envelope estimate | |
| Damping from overshoot | inverse of the overshoot formula |
| Mp (%) | β = cos⁻¹ζ (°) | ωn·ts (2%) | |||
|---|---|---|---|---|---|
| 0.1 | 72.9 | 84.3 | 1.68 | 3.16 | 40.0 |
| 0.2 | 52.7 | 78.5 | 1.81 | 3.21 | 20.0 |
| 0.3 | 37.2 | 72.5 | 1.97 | 3.29 | 13.3 |
| 0.4 | 25.4 | 66.4 | 2.16 | 3.43 | 10.0 |
| 0.5 | 16.3 | 60.0 | 2.42 | 3.63 | 8.0 |
| 0.6 | 9.5 | 53.1 | 2.77 | 3.93 | 6.7 |
| 0.7 | 4.6 | 45.6 | 3.29 | 4.40 | 5.7 |
| 0.8 | 1.5 | 36.9 | 4.16 | 5.24 | 5.0 |
| 0.9 | 0.2 | 25.8 | 6.17 | 7.21 | 4.4 |
Divide the last three columns by ωn in rad/s to get seconds. The settling column uses the 4/(ζωn) envelope estimate, so it is approximate; the other columns are exact for the standard second-order system.
More detail and model assumptions
Where the step-response formulas come from
The unit-step response is C(s) = ωn²/[s(s² + 2ζωn s + ωn²)]. Expand it as in the chapter on inverse Laplace transforms: C(s) = 1/s − (s + 2ζωn)/[(s + σ)² + ωd²]. Splitting the numerator into (s + σ) + σ and inverting gives c(t) = 1 − e^(−σt)[cos ωd t + (σ/ωd) sin ωd t]. Because σ/ωd = ζ/√(1 − ζ²), the cosine and sine combine into one sine of amplitude 1/√(1 − ζ²) and phase β, which is the formula in the list.
Differentiate: ċ(t) = [ωn/√(1 − ζ²)] e^(−σt) sin ωd t. The slope is zero whenever ωd t = nπ, and the first maximum is at n = 1, so tp = π/ωd. Substituting back gives c(tp) = 1 + e^(−σπ/ωd), and the excess over the final value 1 is Mp = e^(−πζ/√(1−ζ²)). The response first reaches 1 when sin(ωd t + β) = 0, that is when ωd tr + β = π.
The response stays between the envelope curves 1 ± e^(−σt)/√(1 − ζ²). Asking when e^(−σt) falls to 0.02 gives t = 3.91/σ, rounded to 4/σ; for 0.05 it gives 3.00/σ. These are estimates. The true settling time depends on which swing is the last to leave the band: for ζ = 0.5 and ωn = 2 rad/s the exact values are 4.04 s (2%) and 2.64 s (5%), against 4 s and 3 s from the formulas.
From specifications to a pole region
Each transient specification is a boundary in the s-plane. An overshoot limit Mp ≤ Mp,max requires ζ ≥ ζmin, so the poles must lie inside the wedge β ≤ cos⁻¹ ζmin around the negative real axis. A settling-time limit ts ≤ ts,max requires σ ≥ 4/ts,max, so the poles must lie to the left of a vertical line. A peak-time limit tp ≤ tp,max requires ωd ≥ π/tp,max, so the poles must lie above a horizontal line, with the mirror image below the real axis.
Example: Mp ≤ 10% and ts ≤ 2 s (2% band). The overshoot formula inverted gives ζ ≥ 0.591, that is β ≤ 53.8°. The settling requirement gives σ ≥ 2 s⁻¹. The corner of the allowed region is s = −2 ± j2.73, which corresponds to ωn = 3.38 rad/s. Any pole pair inside the region meets both specifications.
Apply this to the running example T(s) = K/(s² + 2s + K). Its real part is σ = 1 s⁻¹ for every K > 1, so the gain moves the poles straight up and down the line Re(s) = −1. Raising K increases ωd, which shortens the peak time, and lowers ζ, which raises the overshoot. The settling time stays near 4 s. To settle faster you must change the plant or the controller structure; gain alone will not do it.
Dominant poles, a third pole and a zero
A higher-order response is a sum of modes, one per pole, and each mode decays as fast as its pole's real part. Poles far to the left die out quickly and usually carry small coefficients. The pair nearest the imaginary axis therefore shapes most of the response: these are the dominant poles. As a working rule, when every other pole and zero is at least five times farther left than σ, the second-order formulas give good estimates.
Add a third pole to the worked example while keeping the DC gain at 1: T₃(s) = 4p/[(s + p)(s² + 2s + 4)]. The second-order values are Mp = 16.3% and tp = 1.81 s. With p = 10 the simulation gives 15.9% and 1.92 s, almost unchanged. With p = 2 it gives 8.1% and 2.46 s, and with p = 1 the overshoot disappears. An extra pole acts as a lag: it slows the response and lowers the overshoot.
Now add a zero instead: T_z(s) = (4/z)(s + z)/(s² + 2s + 4). The step response becomes c(t) + ċ(t)/z, the original response plus a scaled copy of its derivative. With z = 10 the overshoot is 16.7%; with z = 2 it is 29.8% at tp = 1.21 s; with z = 1 it reaches 69.9%. A left-half-plane zero speeds up the rise and raises the overshoot, while the poles and the decay rate stay the same.
Critical damping and overdamping
At ζ = 1 both poles sit at −ωn and the step response is c(t) = 1 − e^(−ωn t)(1 + ωn t). It rises without overshoot; for a given ωn, no response of this family without overshoot is faster. Peak time and maximum overshoot are not defined here, and the rise time is measured from 10% to 90% of the final value.
For ζ > 1 the poles are real: s = −ζωn ± ωn√(ζ² − 1). Take ωn = 4 rad/s and ζ = 1.25: the poles are −2 and −8, and partial fractions give c(t) = 1 − (4/3)e^(−2t) + (1/3)e^(−8t). The e^(−8t) term is gone within about half a second, and the slow pole at −2 controls the rest. Increasing the damping further moves the slow pole toward the origin and lengthens the tail.
Connect the example to the lab
Open the mass–spring–damper step experiment: m = 1 kg, k = 16 N/m, c = 1.6 N·s/m, with a 2 N step. Then ωn = √(k/m) = 4 rad/s and ζ = c/(2√(km)) = 0.2. The formulas predict Mp = 52.7%, tp = 0.80 s, a peak of 0.125 × 1.527 = 0.191 m, and ts ≈ 5 s for the 2% band. Run it: the overshoot reading should be about 52.7% and the 2% settling time about 4.9 s. The lab reports the 10–90% rise time (about 0.30 s), which is a different definition from the 0–100% formula (0.45 s).
Check the overshoot formula on a mass–spring–damperMake a prediction before moving a control.
- PREDICTFor m = 1 kg, k = 16 N/m, c = 1.6 N·s/m and a 2 N step, compute ζ, ωn, the percent overshoot and the peak displacement before you run anything.
- CHANGE ONE THINGRun the baseline and pin it. Then change only the damping to c = 3.2 N·s/m, which doubles ζ to 0.4, and predict the new overshoot before you run again.
- OBSERVECompare the overshoot readings with 52.7% and 25.4%, and the peak times with 0.80 s and 0.86 s. Both runs should settle at the same 0.125 m.
- EXPLAINUse the Mp formula to explain why doubling ζ more than halved the overshoot, and use the static balance kx = F to explain why the final displacement did not move.
Check three things before you use these formulas: the numerator is a constant (no zero), no other pole lies near the dominant pair, and 0 < ζ < 1. Measure percent overshoot from the final value c(∞), not from the reference; the two differ whenever there is steady-state error.
Check your understanding+
Two designs have closed-loop poles at −2 ± j2 and at −4 ± j4. Which one has the larger overshoot, and which one settles faster?
The overshoot is the same. Both pairs lie on the ray β = 45°, so ζ = 0.707 and Mp = e^(−π) ≈ 4.3%. The second pair has twice the σ and settles in half the time: ts ≈ 4/4 = 1 s against 4/2 = 2 s.Solve these before opening the solutions
Work each problem on paper first. Open the answer to check yours, then the full solution if you need the steps, and mark how it went.
For the unit-step response of G(s), find ωn, ζ, the poles, the peak time, the percent overshoot and the 2% settling time.
Show answer
AnswerAn underdamped system with poles at −3 ± j4.
Show full solution
Solution
Match coefficients with the standard form.
Compute the real and imaginary parts of the poles.
Use ωd for the peak time and the ratio σ/ωd for the overshoot.
Use σ for the 2% settling time.
A unit step is applied to G(s), whose DC gain is not 1. Find the final value, the percent overshoot, and the value and time of the first peak.
Show answer
AnswerThe output settles at 2 and overshoots that final value by 35.1%.
Show full solution
Solution
The poles are in the left half-plane, so the final value of the unit-step response is G(0).
Read ωn and ζ from the denominator, then the pole coordinates.
Overshoot and peak time depend only on the denominator.
Overshoot is a fraction of the final value, so the peak is 2 times 1.351.
A mass–spring–damper has m = 2 kg, c = 8 N·s/m and k = 50 N/m. It is at rest when a constant 10 N force is applied. Find ζ, ωn, the final displacement, the peak displacement and its time, and the 2% settling time.
Show answer
AnswerThe mass settles at 0.20 m after peaking at about 0.251 m.
Show full solution
Solution
Divide the transfer function by m to expose the standard denominator.
The final displacement is the static balance kx = F.
Compute ωd, the peak time and the overshoot.
Scale the overshoot by the final displacement and estimate the settling time from σ = 2 s⁻¹.
A recorded unit-step response settles at 0.8. Its first and highest peak is 1.0 at t = 0.5 s. Assume a second-order system with no zeros and identify ζ, ωn and the transfer function.
Show answer
AnswerOvershoot gives ζ, peak time gives ωd, and the final value gives the DC gain.
Show full solution
Solution
Measure the overshoot from the final value, not from 1.
Invert the overshoot formula to get ζ.
The peak time gives ωd, and ζ converts it to ωn.
The DC gain is the final value 0.8, so the numerator is 0.8ωn².
A plant G(s) = K/[s(s + 4)] is placed under unity negative feedback. (a) Find K for 10% overshoot. (b) Find the resulting peak time and 2% settling time. (c) Can a 2% settling time of 1 s be met by changing K alone? If the plant pole is moved from −4 to −a, find a and K that give 10% overshoot and ts ≈ 1 s.
Show answer
AnswerK ≈ 11.4 gives tp ≈ 1.15 s and ts ≈ 2 s. Gain alone cannot shorten ts because σ = 2 s⁻¹ for every K; a = 8 with K ≈ 45.8 meets both specifications.
Show full solution
Solution
Match the closed-loop denominator to the standard form. The s coefficient fixes σ whatever the gain.
Turn the 10% overshoot specification into a damping ratio.
With σ = 2 fixed, ζ determines ωn and therefore K.
Compute ωd, then the peak time and the settling time.
Gain cannot change σ, so ts stays near 2 s. A settling time of 1 s needs σ = 4 s⁻¹. With the denominator s² + as + K that means a = 2σ = 8; the same ζ then gives ωn and K.
Mark each exercise after you check your answer. Marks stay in this browser.
Terms this lecture teaches
- second-order system2차 시스템
- overshoot (%OS)오버슈트
- peak time피크시간
- settling time정착시간
- rise time상승시간
- natural frequency고유진동수
- damping ratio감쇠비
- step response계단응답
- zero영점
- DC gain직류 이득
- underdamped부족감쇠
- critically damped임계감쇠
- overdamped과감쇠
- damped natural frequency감쇠 고유진동수
- dominant poles지배극점
- mass–spring–damper질량–스프링–댐퍼
- transient response과도응답
Frequently asked questions
What damping ratio gives 10% overshoot?
Invert the overshoot formula: ζ = −ln Mp / √(π² + ln² Mp), with Mp as a fraction. For Mp = 0.10 this gives ζ ≈ 0.591, so the poles must lie within about 53.8° of the negative real axis to keep overshoot at or below 10%.
Why does raising the gain increase overshoot but not shorten settling time?
In the loop K/(s² + 2s + K), the real part of the poles stays at −1 for every K > 1, so the gain only moves them vertically. K = 4 gives ζ = 0.5 and 16.3% overshoot, K = 16 gives ζ = 0.25 and 44.4%, while the estimate ts ≈ 4/(ζωn) stays at 4 s in both cases.
Is percent overshoot measured from the reference or from the final value?
From the final value c(∞). The formula Mp = e^(−πζ/√(1 − ζ²)) describes how far the peak exceeds the value the response settles at, and that value equals the reference only when there is no steady-state error.
Which rise-time definition should I use, 0–100% or 10–90%?
The formula tr = (π − β)/ωd uses 0–100%: the first time an underdamped response reaches its final value. A critically damped or overdamped response approaches the final value without crossing it, so use 10–90% there. Always state which definition a reported rise time uses.
When do the second-order formulas work for a higher-order system?
When one complex pole pair dominates. As a working rule, every other pole and zero should be at least five times farther from the imaginary axis than that pair. A nearby extra pole slows the response and lowers the overshoot; a nearby left-half-plane zero speeds up the rise and raises the overshoot.
Why the next lesson follows
Every formula here assumes the poles are in the left half-plane, so that the transient dies out. The next chapter asks that question directly: when is a loop stable, and how much error remains once the transient is gone?
Further reading
Finished the reading and the exercises? Mark the lecture as done to tick it on the syllabus. Progress stays in this browser.
EIGENROOM · https://eigenroom.com/theory/second-order-specs · Updated 2026-11-15