Control Systems and Signals for Students: Time Constants, Damping, Stability and Sampling with Worked Examples

Control systems and signal processing sit behind thermostats, motor drives, cruise control, audio equipment and almost every instrument you will meet as an engineering student. The subject can feel abstract, with Laplace transforms and pole-zero plots. But most exam problems and most practical intuition come from a short list of results: time constants, damping ratio, overshoot, stability tests, steady-state error, decibels and the sampling theorem. This guide gives each with a worked example, so you can check the numbers by hand or with the Free Technology Calculators app.

It builds on earlier topics. For the circuit part see Basic Circuit Calculations and Ohm’s Law, and for thermal systems see Thermodynamics and Heat Transfer Formulas.

1. What a control system does

A control system measures an output, compares it with a desired value (the setpoint) and adjusts an input to reduce the error. In a closed-loop (feedback) system, the controller acts on the error. In open-loop systems there is no feedback, so disturbances are not corrected. Typical goals are speed (respond quickly), accuracy (small steady-state error) and stability (do not oscillate or blow up).

2. Transfer functions in one paragraph

A transfer function G(s) is the ratio of the Laplace transform of the output to that of the input, assuming zero initial conditions. Its poles (roots of the denominator) set the system’s natural behaviour. Poles in the left half of the s-plane give decaying responses (stable), and poles in the right half give growing ones (unstable).

3. First-order systems and the time constant

A first-order system has G(s) = K / (τs + 1). The step response is y(t) = K(1 − e−t/τ). Useful values:

TimeFraction of final value
t = τ63.2%
t = 2τ86.5%
t = 3τ95.0%
t = 4τ98.2% (a common 2% settling time)
10% to 90% rise timeabout 2.2τ

Example 1: an RC low-pass filter. R = 10 kΩ and C = 100 µF. The time constant τ = RC = 10,000 × 0.0001 = 1 s. The output reaches about 63% of a step after 1 s and settles to within 2% after about 4 s. The cutoff frequency is fc = 1 / (2πRC) = 1 / (2π × 1) = 0.159 Hz, where the output amplitude falls to 70.7% of the input (−3 dB).

Example 2: closing the loop. A plant G(s) = 8 / (s + 2) in unity feedback has closed-loop transfer function T(s) = 8 / (s + 2 + 8) = 8 / (s + 10). The pole moves from −2 to −10, so the time constant shrinks from 0.5 s to 0.1 s: feedback made the system five times faster. The DC gain is 8 / 10 = 0.8, so the output settles at 80% of the setpoint, which leads to steady-state error.

4. Steady-state error

For a unity-feedback system with a step input, the steady-state error is ess = 1 / (1 + Kp), where Kp = lims→0 G(s). In Example 2, Kp = 8 / 2 = 4, so ess = 1 / 5 = 0.2, or 20% of the setpoint, which agrees with the 80% final value above. A system with an integrator (a 1/s term in the loop, called type 1) has zero steady-state error for a step. That is the reason controllers include an integral term.

5. Second-order systems: ωn and ζ

The standard form is G(s) = ωn2 / (s2 + 2ζωns + ωn2), with natural frequency ωn and damping ratio ζ.

Damping ratioBehaviour
ζ = 0Undamped: sustained oscillation
0 < ζ < 1Underdamped: oscillates and decays, with overshoot
ζ = 1Critically damped: fastest without overshoot
ζ > 1Overdamped: slow, no overshoot
Poles:                 s = −ζωn ± jωd
Damped frequency:      ωd = ωn × √(1 − ζ²)
Percent overshoot:     %OS = 100 × exp(−πζ / √(1 − ζ²))
Peak time:             tp = π / ωd
Settling time (2%):    ts ≈ 4 / (ζωn)

Example 3. ωn = 10 rad/s and ζ = 0.5.

  • √(1 − 0.25) = 0.866, so ωd = 10 × 0.866 = 8.66 rad/s.
  • Poles at s = −5 ± j8.66.
  • %OS = 100 × exp(−π × 0.5 / 0.866) = 100 × exp(−1.814) = 16.3%.
  • Peak time = π / 8.66 = 0.363 s.
  • Settling time ≈ 4 / (0.5 × 10) = 0.8 s.

Trade-off: raising ζ reduces overshoot but slows the response, while raising ωn speeds it up but demands more from actuators and can excite unmodelled dynamics. Many designers aim for ζ around 0.5 to 0.8 as a compromise, depending on the application.

6. Stability: the Routh-Hurwitz test

To check stability without finding the roots, build the Routh array from the characteristic polynomial. For a cubic s3 + a2s2 + a1s + a0, all coefficients must be positive and a2a1 > a0.

Example 4. A unity-feedback loop has characteristic equation s3 + 6s2 + 11s + K = 0. The conditions are K > 0 and 6 × 11 > K, so the system is stable for 0 < K < 66. At K = 66 the system is marginally stable and oscillates continuously. Check: s3 + 6s2 + 11s + 66 = (s + 6)(s2 + 11), with roots s = −6 and s = ±j√11, so the imaginary pair confirms sustained oscillation at about 3.3 rad/s.

7. PID control in plain words

  • Proportional (P): output proportional to the present error. Raises speed but leaves some steady-state error and can cause overshoot if the gain is high.
  • Integral (I): accumulates past error and removes steady-state error, but can add overshoot and slow oscillation (integral windup in saturated actuators).
  • Derivative (D): reacts to the rate of change, adds damping, but amplifies measurement noise.

Tuning methods such as Ziegler-Nichols give starting values, and real systems usually need refinement and limits on actuator output.

8. Signals: amplitude, RMS and decibels

For a sine wave, the RMS value equals the peak divided by √2 (about 0.707 of the peak). Example 5. Mains at 230 V RMS has a peak of 230 × 1.414 = 325 V. The period of a 50 Hz wave is 1/50 = 20 ms.

Decibels express ratios on a log scale. For amplitude (voltage) ratios, dB = 20 log10(ratio), and for power ratios dB = 10 log10(ratio).

Amplitude ratiodB
0.707−3 dB
2+6 dB
10+20 dB
100+40 dB
0.1−20 dB

9. Sampling and the Nyquist limit

To represent a signal digitally without losing information, the sampling frequency must exceed twice the highest frequency in the signal: fs > 2fmax. Example 6. Audio up to 20 kHz needs more than 40 kHz, which is why 44.1 kHz is a standard rate.

Example 7: aliasing. A 30 kHz tone sampled at 44.1 kHz is above the Nyquist frequency of 22.05 kHz, so it folds back and appears at |44.1 − 30| = 14.1 kHz, a false frequency that cannot be removed afterward. This is why an anti-aliasing low-pass filter is placed before the sampler.

10. Time domain versus frequency domain

Any periodic signal can be written as a sum of sine waves (Fourier series), and the frequency view shows which components carry the energy. A low-pass filter keeps low frequencies and removes high ones, a high-pass filter does the opposite, and a band-pass keeps a band. The bandwidth of a first-order system is 1/τ in rad/s, so a faster time constant means wider bandwidth. That link between speed in time and bandwidth in frequency is central to both control and communications.

Quick reference

QuantityFormula
RC time constantτ = RC
First-order settling (2%)≈ 4τ
Cutoff frequencyfc = 1 / (2πRC)
Overshoot100 exp(−πζ / √(1 − ζ²))
Settling time (2%)≈ 4 / (ζωn)
Step error, type 01 / (1 + Kp)
Nyquistfs > 2fmax

Common mistakes

  • Using the overshoot and settling formulas for systems that are not well approximated as second order.
  • Forgetting to convert Hz to rad/s (multiply by 2π).
  • Using 10 log for amplitude ratios instead of 20 log.
  • Declaring stability from the sign of one coefficient, rather than checking the full Routh array.
  • Ignoring actuator saturation, delays and noise, which real controllers must handle.
  • Sampling at exactly twice the signal frequency in practice, with no margin or anti-alias filter.

How to study this topic

Work each example by hand, then verify with the calculators in Free Technology Calculators. Sketch the step response before computing, so you notice when a number does not make sense. Practise with units, as covered in Unit Conversions and Dimensional Analysis, and see Free Calculators for Engineering Students for more tools. If your lab data needs statistical treatment, read Choosing the Right Statistical Test.

Frequently asked questions

What is the difference between open-loop and closed-loop control?

Closed-loop systems measure the output and correct errors, while open-loop systems apply a fixed input and cannot correct disturbances.

Why does a system with high gain become unstable?

Higher loop gain moves closed-loop poles toward the right half-plane in many systems, as Example 4 shows with K = 66.

Why 4 time constants for settling?

The response is within about 2% of the final value after 4τ. Some courses use 5τ or a 5% band, so follow your textbook.

What does Nyquist rate mean?

It is the minimum sampling rate, twice the highest signal frequency, below which aliasing occurs.

Conclusion

Control and signals reward a small toolkit used well: time constants for first-order behaviour, ζ and ωn for second-order behaviour, Routh for stability, ess for accuracy, dB for gain and Nyquist for sampling. Practise each with numbers, then confirm in Free Technology Calculators.

Disclaimer: This article is for educational purposes. Real control systems involve nonlinearities, delays and safety considerations that these idealised formulas do not cover. Consult qualified engineers for safety-critical design.

Leave a Reply

Share Your Love