Push down on the corner of a parked car and let go. A car in good order rises once and stops; a car with worn shock absorbers bobs two or three times before it settles. In an electronics lab the same two behaviors appear on an oscilloscope when a capacitor discharges through a coil: a smooth return, or a ringing that dies out. Both are the transient between one steady state and another, and both obey the same small family of linear differential equations — first order for a single energy store, second order for two that exchange. This chapter solves them once and for all, names their parameters (time constant, natural frequency, quality factor), and shows how the same numbers describe a circuit and a suspension.
A coil, a capacitor and a resistor on a breadboard, and on the oscilloscope the decaying oscillation that follows a step: the pseudo-periodic regime of this chapter.
7.1 First-order systems
Definition 7.1(First-order linear system)
A quantity x(t) obeys a first-order linear equation with constant coefficients when
τdtdx+x=x∞,
with τ>0 the time constant and x∞ a constant (the value imposed by the source). Its steady state is x=x∞; the transient regime is the approach to it from the initial value x(0)=x0.
Theorem 7.2(Solution)
The unique solution with x(0)=x0 is
x(t)=x∞+(x0−x∞)e−t/τ.
The gap to the steady state shrinks by the factor e−1≈0.37 every τ: 63% of the way at τ, 95% at 3τ, 99% at 4.6τ. The tangent at the origin reaches x∞ at t=τ.
Proof.y=x−x∞ obeys τy′+y=0, whose solutions are y=Ke−t/τ (the mathematics course on linear differential equations proves there are no others); K=x0−x∞. The slope at 0 is −(x0−x∞)/τ, which would close the gap in time τ. ∎
The first-order step response from x0=0: the initial tangent meets the asymptote at t=τ; 63% of the way at τ, 95% at 3τ. Every first-order transient is this curve, stretched by τ and shifted by x0 and x∞.
Proposition 7.3(RC and RL circuits)
A capacitor C charged through a resistorR from an ideal source E obeys RCduC/dt+uC=E: τ=RC, uC,∞=E. A coil L in series with R across E obeys (L/R)di/dt+i=E/R: τ=L/R, i∞=E/R. With the source removed (E→0) the same equations describe the discharge toward zero.
The voltage across a capacitor and the current through an inductor are continuous functions of time: at a switching instant t0, uC(t0+)=uC(t0−) and iL(t0+)=iL(t0−). Currents in resistors and voltages across coils may jump.
Proof. A jump of uC would require i=CduC/dt infinite, a jump of iL would require u=Ldi/dt infinite; both are impossible in a circuit of finite voltages and currents (equivalently, the stored energies 21CuC2 and 21Li2 cannot change instantaneously without infinite power). ∎
Method 7.5(Solving a first-order transient)
Write the loop and node laws after the switch, with the component laws; reduce to one equation in uC or iL and put it in the form τx′+x=x∞; read τ and x∞ (the latter is also the steady state found by replacing C by an open circuit, L by a wire).
Find x0 from the continuity of uC or iL across the switching instant, using the circuit before it.
Write x=x∞+(x0−x∞)e−t/τ; deduce the other quantities by differentiation and the loop law.
When the capacitor sees a network of resistors and sources, replace that network by its Thévenin equivalent (ETh,RTh) first (Chapter 6): τ=RThC, u∞=ETh.
Proposition 7.6(Energy balance of the RC charge)
Charging a capacitor from 0 to E through any resistorR, the source supplies CE2, the capacitor stores 21CE2, and the resistor dissipates the other 21CE2 — whatever R.
Proof.i=(E/R)e−t/τ; the source supplies ∫0∞Eidt=E⋅(E/R)τ=CE2; the resistor takes ∫0∞Ri2dt=(E2/R)∫0∞e−2t/τdt=(E2/R)(τ/2)=21CE2; the difference is the stored 21CE2. A smaller R dissipates the same energy faster. ∎
Example 7.7(A relay coil)
L=20mH, R=5.0Ω, E=10V: τ=4.0ms, i∞=2.0A; the contacts pull in when i reaches 1.5A, i.e. at t=−τln(1−0.75)=5.5ms. Opening the circuit, i must fall from 2A; a “freewheeling” diode across the coil gives it a path through R alone, τ=4ms, instead of a spark.
7.2 Second-order systems
Definition 7.8(Canonical second-order form)
A quantity x(t) is a second-order linear system when
A capacitor C, a coil L and a resistorR in series across a source E: the capacitor voltage obeys the canonical form with
ω0=LC1,Q=R1CL,uC,∞=E.
Proof.Loop law: E=Ldi/dt+Ri+uC with i=CduC/dt: LCuC′′+RCuC′+uC=E; divide by LC and identify ω02=1/LC, ω0/Q=R/L. ∎
The two faces of the second-order system: the series RLC circuit (ω0=1/LC, Q=L/C/R) and the mass–spring–damper (ω0=k/m, Q=km/α). Same equation, same three regimes.
Proposition 7.10(The mechanical oscillator)
A mass m on a spring of stiffness k with a viscous damper of coefficient α (force −αx˙), displaced by x from equilibrium, obeys mx¨+αx˙+kx=0: the canonical form with
ω0=mk,Q=αkm.
The correspondence x↔q, x˙↔i, m↔L, α↔R, k↔1/C maps every result of one system onto the other.
Proof. Newton’s second law (Chapter 12) with the spring force −kx and the damping force; divide by m. ∎
Theorem 7.11(The three regimes)
The solutions of the homogeneous equation x′′+(ω0/Q)x′+ω02x=0 are governed by the roots of r2+(ω0/Q)r+ω02=0, of discriminant Δ=ω02(1/Q2−4):
Q<21, aperiodic: two real negative roots r±=−2Qω0(1∓1−4Q2), x=Aer+t+Ber−t, a return without oscillation;
Q=21, critical: a double root r=−ω0, x=(A+Bt)e−ω0t, the fastest return without overshoot;
Q>21, pseudo-periodic: complex roots r=−1/τ±iω with
τ=ω02Q,ω=ω01−4Q21,x=e−t/τ(Acosωt+Bsinωt),
a damped oscillation of pseudo-periodT=2π/ω inside the envelope ±A2+B2e−t/τ.
The full solution with a constant right-hand side is x∞ plus the homogeneous solution; A and B follow from x(0) and x′(0).
Proof. Characteristic equation of a linear equation with constant coefficients (mathematics course, second-order linear equations): the roots are r=−ω0/2Q±21Δ. For Q>21, Δ=iω04−1/Q2, giving the real part −ω0/2Q=−1/τ and the imaginary part ±ω01−1/4Q2; the real solutions are the combinations of e−t/τcosωt and e−t/τsinωt. ∎
Step response of a second-order system from rest, for three quality factors. Small Q: a slow creep; Q=21: the quickest return without overshoot; large Q: ringing that lasts about Q oscillations.
Proposition 7.12(Reading Q on a pseudo-periodic trace)
gives Q; the number of oscillations visible before the amplitude falls below 5% is about Q. For Q≫1, ω≈ω0 and T≈2π/ω0.
Proof. The envelope e−t/τ multiplies by e−T/τ per period; T/τ=(2π/ω)(ω0/2Q). The envelope reaches e−3≈5% at 3τ=6Q/ω0≈QT, i.e. after about Q periods. ∎
Free pseudo-periodic decay, x=x0e−t/τcosωt: successive maxima shrink by the constant factor e−T/τ, and ln(xn/xn+1)=T/τ≈π/Q measures the quality factor from the trace.
Example 7.13(An RLC ringing)
L=10mH, C=100nF, R=100Ω: ω0=1/10−9=3.16×104rad/s (f0=5.03kHz), Q=L/C/R=316/100=3.16: pseudo-periodic, τ=2Q/ω0=0.20ms, ω=0.987ω0, T=0.20ms, three visible oscillations. The critical resistance is Rc=2L/C=632Ω; above it the discharge is aperiodic.
Method 7.14(Solving a second-order transient)
Write the equation; put it in canonical form; read ω0, Q and x∞.
Decide the regime from Q and write the general solution as x∞ plus the homogeneous solution.
Find x(0) and x′(0) from the continuity of uC and iL (for the RLC: uC(0+)=uC(0−) and uC′(0+)=i(0−)/C); solve for the two constants.
Example 7.15(Initial conditions)
The capacitor of the previous example is charged to U0 and, at t=0, closed on L and R (no source): x∞=0, uC(0)=U0, uC′(0)=i(0)/C=0 since the coil current was zero. Hence A=U0 and −A/τ+Bω=0, B=U0/(ωτ)=U0/4Q2−1: uC=U0e−t/τ[cosωt+sin(ωt)/4Q2−1] — for Q≫1 simply U0e−t/τcosω0t, and the current i=−CduC/dt peaks near U0C/L=U0/(ω0L).
Remark 7.16(Energy)
The stored energy 21CuC2+21Li2 (or 21kx2+21mx˙2) decreases at the rate Ri2 (αx˙2): differentiate and use the equation. Each pseudo-period the energy shrinks by the factor e−2T/τ; with Q=3 about 88% of it is lost per oscillation. A high Q means a slow leak relative to the oscillation — a good clock, a sharp resonance (Chapter 8) — and a low Q a quick return: the right choice for a suspension or a meter needle.
7.3 Exercises
Exercise 7.1★
An RC circuit: R=10kΩ, C=47nF, charged from E. Compute τ, uC(τ)/E, and the time to reach 99% of E.
For the same L and C, what resistance gives the critical regime? What is Q then? What happens for R=2kΩ?
Solution
Solution of Exercise 7.4.
Rc=2L/C=632Ω, Q=21. At 2kΩ, Q=0.16: aperiodic, a slow creep governed by the small root ∣r+∣≈ω0Q=5×103s−1 (time scale 0.2ms but no oscillation).
Exercise 7.5★★
A charged capacitor C=1.0µF discharges through an unknown R; the voltage halves every 3.5ms. Show that the half-life is τln2 and find R.
Solution
Solution of Exercise 7.5.
u=U0e−t/τ halves when e−t/τ=21: t1/2=τln2. τ=3.5/0.693=5.05ms, R=τ/C=5.1kΩ.
Exercise 7.6★★
Show by direct integration that charging a capacitor from 0 to E through R dissipates exactly 21CE2 in R, independently of R. What becomes of the efficiency of the transfer, and how could one do better?
Solution
Solution of Exercise 7.6.
i=(E/R)e−t/τ: ∫0∞Ri2dt=(E2/R)(τ/2)=21CE2, no R in sight. The source supplies CE2: efficiency 50% for any resistor. Better: charge through an inductor (the current is then not proportional to the voltage gap) or in several voltage steps — each step of E/n loses only 21C(E/n)2.
Exercise 7.7★★
On an oscilloscope, a free RLC decay shows successive maxima of 2.0V, 1.5V, 1.12V at intervals of 0.40ms. Deduce Q, ω0, and, with C=1.0µF, the values of L and R.
Solution
Solution of Exercise 7.7.
δ=ln(2.0/1.5)=0.29 (and ln(1.5/1.12)=0.29: consistent); Q≈π/δ=11. T=0.40ms, so ω0≈2π/T=1.57×104rad/s (the correction 1/8Q2 is 0.1%). L=1/(ω02C)=4.1mH; R=ω0L/Q=5.8Ω.
Exercise 7.8★★
A capacitor charged to U0=100V (C=10µF) is switched onto a coil L=1.0mH of small resistance (Q≫1). Write uC(t) and i(t), and estimate the peak current. Where does the energy sit at the peak?
Solution
Solution of Exercise 7.8.
uC≈U0e−t/τcosω0t; i=−CduC/dt≈CU0ω0sinω0t (for t≪τ), peak U0C/L=100×0.1=10A, reached a quarter period after closing, when uC=0: all the energy (21CU02=50mJ) is then magnetic, 21Li2=50mJ.
Exercise 7.9★★
A quarter of a car: m=400kg on a spring k=2.0×104N/m. Compute ω0, f0 and the critical damping coefficient. With a damper α=2000Ns/m, give Q, the regime, the pseudo-period and the decay time τ.
A coil L=0.50H carries I0=2.0A; the source is disconnected and the coil is left on a bleed resistorR=100Ω. Give i(t), τ, the peak voltage across the coil and the energy dissipated. What would happen without the resistor?
Solution
Solution of Exercise 7.10.
i=I0e−t/τ, τ=L/R=5.0ms; peak voltageRI0=200V at t=0+; energy 21LI02=1.0J dissipated in R. Without the resistor the current has no path: di/dt is huge, the coil voltage reaches kilovolts, and an arc jumps the switch.
Exercise 7.11★★★
A capacitor C=1.0µF is charged from E=12V through R1=10kΩ while R2=20kΩ is connected across it. Using a Thévenin equivalent, find the time constant and the final voltage; write uC(t) from uC(0)=0.
Solution
Solution of Exercise 7.11.
Seen from C: ETh=ER2/(R1+R2)=8.0V, RTh=R1∥R2=6.67kΩ; τ=RThC=6.7ms; uC=8.0(1−e−t/τ) V.
Exercise 7.12★★★
Critical regime, step from rest (x(0)=x′(0)=0): show that x=x∞[1−(1+ω0t)e−ω0t], and find numerically the time to reach 95% of x∞ (in units of 1/ω0). Compare with the settling time of the envelope for Q=5 and for the slow root when Q=0.1; conclude on why “critical” is the engineer’s target.
Solution
Solution of Exercise 7.12.
Double root −ω0: x=x∞+(A+Bt)e−ω0t; x(0)=0 gives A=−x∞, x′(0)=0 gives B=ω0A: x=x∞[1−(1+ω0t)e−ω0t]. 95%: (1+s)e−s=0.05, s≈4.7: t95=4.7/ω0. For Q=5 the envelope e−t/τ with τ=10/ω0 needs 3τ=30/ω0; for Q=0.1 the slow root is ∣r+∣≈ω0Q and 3/∣r+∣=30/ω0. Both are six times slower: critical damping is the fastest no-overshoot return, which is what a meter, a door closer or a suspension wants.
7.4 Problem: Designing a suspension
Problem 7.1
Weekend problem — the same differential equation on a lab bench and under a car: how many bounces a worn shock absorber allows, what the designer chooses instead, and how one number tells the two apart
Part I studies a series RLC circuit, L=10mH, C=100nF; Parts II–IV a quarter-car model: a mass m=350kg resting on a spring k=22kN/m and a shock absorber exerting −αx˙. Overshoot of a step response in the pseudo-periodic regime (given): xmax/x∞−1=exp(−πξ/1−ξ2), ξ=1/2Q.
Part I — The circuit.
Write the loop law for the series RLC across a source E, and the differential equation for uC.
Put it in canonical form; express ω0 and Q.
Compute ω0, f0 and the critical resistance Rc.
Write the characteristic equation and its discriminant; state the three regimes in terms of Q.
In the pseudo-periodic case, give τ and ω and the form of uC(t).
For R=100Ω compute Q, τ, the pseudo-period, and the number of visible oscillations.
The capacitor, charged to U0, is switched onto L and R at t=0. Give uC(0+) and duC/dt(0+), with the physical reason for each.
Part II — The quarter car.
Measuring x from the equilibrium position, write Newton’s second law for the body and show that weight drops out.
Put the equation in canonical form; give ω0 and Q in terms of m, k, α, and list the electromechanical correspondences.
Compute ω0, f0 and the critical coefficient αc.
The designer fits α=4000Ns/m: compute Q, ξ, and name the regime.
Compute τ and the pseudo-period.
Compute the overshoot after a step (a kerb): is the car comfortable?
What resistance R would give the circuit of Part I the same Q as this suspension?
Part III — The worn damper. Oil has leaked: α=1200Ns/m.
Compute Q, the pseudo-period and τ.
By what factor do successive maxima shrink? How many bounces are visible before the amplitude is below 5%?
The bounce test: one pushes the corner down by x0=5.0cm and releases. Give the initial conditions, the energy stored in the spring, and say where that energy goes.
Estimate the peak velocity of the body during the first return (≈x0ω0 for Q≫1) and the peak damper force.
Why is the exactly critical damping (Q=21) not the designer’s choice, although it gives no overshoot? (Think of the last few millimeters of the return.)
Compute the overshoot for the worn damper and compare with Part II.
Part IV — Diagnosing Q from a trace. An accelerometer on the body records the free decay.
For the worn car, successive maxima read 1.00, 0.25, 0.062 (arbitrary units). Compute the logarithmic decrement, deduce ξ exactly from δ=2πξ/1−ξ2, then Q; compare with Part III.
For the healthy car no second maximum is visible; only the overshoot, 3.8%, can be read. Invert the overshoot formula to find ξ and Q.
What fraction of the mechanical energy does the worn damper dissipate per pseudo-period?
A technician who has no car at hand builds the RLC of Part I with R chosen so that Q=2.3: what R, and why is the oscilloscope trace, plotted against ω0t, identical in shape to the car’s?
Summarize: which dimensionless number decides the shape of any second-order transient, what value a suspension designer aims for, and what a driver should conclude from “more than one bounce”.
16.e−T/τ=e−1.39=0.25: each bounce is a quarter of the previous; 0.253=1.6%: three visible bounces.
17.x(0)=−x0, x˙(0)=0; energy 21kx02=0.5×22000×0.0025=28J, dissipated as heat in the damper’s oil.
18.v≈x0ω0=0.05×7.93=0.40m/s; damper force αv≈1200×0.40=480N.
19. At Q=21 the return is (1+ω0t)e−ω0t, which creeps asymptotically: the last millimeters take long. A slightly underdamped system (ξ≈0.7) overshoots by a few percent but enters and stays inside a tolerance band sooner.
20.exp(−π×0.22/0.976)=exp(−0.70)=50%, against 3.8%: the worn car bounces halfway back past equilibrium.
21.δ=ln4=1.39 (twice, consistent); ξ=δ/4π2+δ2=1.39/6.43=0.215, Q=1/2ξ=2.3: the worn value.
22.−ln0.038=3.27=πξ/1−ξ2, so ξ=3.27/π2+3.272=0.72, Q=0.69.
23. Energy ∝amplitude2: remaining fraction per period 0.252=6%, so 94% dissipated per pseudo-period.
24.R=316/2.3=137Ω. In the variable ω0t the canonical equation contains only Q: two systems with the same Q trace the same curve.
25. The quality factorQ (or ξ) alone fixes the shape; a designer aims at Q≈0.7 (ξ≈0.7); more than one visible bounce means Q well above 1: the damper is worn.