Electronics Guide

Energy Storage and the Continuity Rules

A capacitor stores energy in its electric field and an inductor in its magnetic field. Their defining relations are:

i_C = C dv_C/dt

v_L = L di_L/dt

The energy each one stores is:

w_C = ½Cv_C²

w_L = ½Li_L²

Changing stored energy in zero time would take infinite power, which no physical source can supply. Two continuity rules follow:

• Capacitor voltage cannot change instantaneously. A step in v_C would require an infinite current, because i_C = C dv_C/dt.
• Inductor current cannot change instantaneously. A step in i_L would require an infinite voltage, because v_L = L di_L/dt.

Nothing else is protected: capacitor current, inductor voltage, and every resistor voltage and current can jump. Capacitor voltages and inductor currents, the state variables, carry the circuit's memory across a switching instant, and every other quantity is recomputed from them.

Idealized schematics can break these rules. Connect a capacitor charged to V across an identical uncharged capacitor with ideal wires, and both voltages must jump to V/2, because charge is conserved. The stored energy falls from ½CV² to 2 × ½C(V/2)² = ¼CV². In hardware the missing half is dissipated in the wiring resistance or radiated, however small that resistance is; a smaller resistance only makes the current pulse larger and briefer. When a model forces such a jump, restore the resistance or inductance it left out.

Initial and Final Conditions at a Switching Instant

Transient problems label three moments. The instant just before switching is t = 0⁻, the instant just after is t = 0⁺, and the new steady state, reached long afterward, is t → ∞. With DC sources, each moment has a simple equivalent circuit, and four steps find everything a transient solution needs:

1. Before switching (t = 0⁻). Assume the circuit has sat in its present configuration long enough to reach DC steady state. Replace each capacitor with an open circuit and each inductor with a short circuit, then solve for every capacitor voltage and inductor current.
2. Just after switching (t = 0⁺). Redraw the circuit in its new configuration. Replace each capacitor with a voltage source equal to v_C(0⁻) and each inductor with a current source equal to i_L(0⁻). Solve this resistive circuit for any other quantity at 0⁺.
3. Long after switching (t → ∞). With DC sources, replace capacitors with open circuits and inductors with short circuits again, and solve for the final values.
4. Initial slopes, for second-order circuits. Use dv_C/dt(0⁺) = i_C(0⁺)/C and di_L/dt(0⁺) = v_L(0⁺)/L, taking i_C(0⁺) and v_L(0⁺) from step 2.

These open-circuit and short-circuit replacements apply only to DC steady state. With a sinusoidal source, the transient decays toward the sinusoidal steady state found by phasor analysis.

As an example, a 12 V source feeds a node through a switch and R₁ = 4 kΩ. From that node, R₂ = 12 kΩ and C = 10 μF each connect to ground. The switch has been open for a long time and closes at t = 0.

• At 0⁻: with the source disconnected, the capacitor has discharged through R₂, so v_C(0⁻) = 0.
• At 0⁺: the capacitor acts as a 0 V source, so no current flows in R₂, and the full 12 V across R₁ drives 3 mA into the capacitor. Its current has jumped from zero while its voltage has not moved, and the voltage starts rising at i_C/C = 300 V/s.
• At ∞: the capacitor is an open circuit, and the resistors form a divider, so v_C(∞) = 12 V × 12 kΩ / (4 kΩ + 12 kΩ) = 9 V.

First-Order Natural Response

A first-order circuit contains one capacitor or one inductor, or several that combine into one, together with resistors and sources. Its natural response is what happens when stored energy drains away with no source driving the circuit.

Consider a capacitor charged to V₀ and connected at t = 0 across a resistor R. Kirchhoff's current law at the top node gives C dv/dt + v/R = 0, or

dv/dt + v/(RC) = 0

The solution is an exponential decay:

v(t) = V₀ e^−t/τ for t ≥ 0, where τ = RC

Substituting confirms it: the derivative is −v/τ, and C(−v/τ) + v/R = 0 exactly when τ = RC. Ohms times farads is (volts per ampere) × (coulombs per volt), or seconds.

An inductor carrying I₀ and connected at t = 0 across a resistor R obeys Kirchhoff's voltage law around the loop, L di/dt + Ri = 0, so

i(t) = I₀ e^−t/τ for t ≥ 0, where τ = L/R

Henries per ohm also reduce to seconds. Note the inversion: a larger resistance slows the RC discharge, because it limits the current that drains the capacitor, but speeds the RL decay, because it develops more opposing voltage for a given current.

The time constant is the time a response takes to complete 63.2 percent of its total change, since 1 − e⁻¹ ≈ 0.632. It has two other useful readings: a tangent drawn at t = 0 reaches the final value in exactly one time constant, and in every interval of length τ the remaining distance to the final value shrinks by the same factor, e ≈ 2.718.

Engineers usually treat a response as settled after five time constants, when less than 1 percent of the change remains. Precision work needs more. Settling to 0.1 percent takes about 6.9τ, and settling to one part in 65,536, one least significant bit of a 16-bit converter after a full-scale step, takes about 11τ.

First-Order Step Response and the General Solution

When a DC source remains connected after switching, the response decays not toward zero but toward a new final value. Every voltage and current in a first-order circuit then follows one form:

x(t) = x(∞) + [x(0⁺) − x(∞)] e^−t/τ for t ≥ 0

Here x is any voltage or current in the circuit, x(0⁺) is its value just after switching, and x(∞) is its final value. The first term is the forced response, set by the sources. The second is the natural response, whose shape the circuit fixes and whose size depends on how far the starting point lies from the final value. If switching happens at t₀ rather than at zero, replace t with t − t₀. Three numbers therefore settle any first-order problem: the initial value, the final value, and the time constant.

The time constant uses the resistance that the energy-storage element sees. Remove the capacitor or inductor, set every independent source to zero (voltage sources become short circuits and current sources become open circuits), and find the Thévenin resistance R_Th between the two terminals the element occupied. Then τ = R_ThC or τ = L/R_Th. Dependent sources stay in the circuit; when they are present, find R_Th by applying a test source and dividing its voltage by its current.

Completing the example from the previous section:

1. Initial value: v_C(0⁺) = v_C(0⁻) = 0 V.
2. Final value: v_C(∞) = 9 V.
3. Time constant: with the 12 V source shorted, the capacitor sees R₁ and R₂ in parallel, so R_Th = 3 kΩ and τ = 3 kΩ × 10 μF = 30 ms.
4. Response: v_C(t) = 9 V × (1 − e^{−t/30 ms}) for t ≥ 0.

The initial slope agrees with the earlier result: 9 V divided by 30 ms is 300 V/s. Using R₁ alone, a common error, would give a 40 ms time constant. Now suppose the switch reopens at t₁, long after the capacitor has charged. The source and R₁ drop out, the capacitor discharges through R₂ alone, and

v_C(t) = 9 V × e^{−(t − t₁)/120 ms} for t ≥ t₁

The same capacitor charges four times faster than it discharges, because it sees a different resistance in each switch position.

Timing circuits frequently need the moment a response crosses a threshold, x_th. Solving the general form for t gives

t = τ ln{[x(0⁺) − x(∞)] / [x_th − x(∞)]}

For example, a capacitor charging from 0 V toward 5 V through 10 kΩ, with C = 100 nF and τ = 1 ms, crosses 3.3 V at t = 1 ms × ln[(0 − 5)/(3.3 − 5)] = 1 ms × ln 2.94 ≈ 1.08 ms.

Rise time links the time constant to bandwidth. The 10 to 90 percent rise time of a first-order step response is t_r = τ ln 9 ≈ 2.2τ. Viewed in the frequency domain, the same circuit is a single-pole low-pass filter with a −3 dB cutoff frequency f_c = 1/(2πτ). Eliminating τ gives

t_r ≈ 2.2/(2πf_c) ≈ 0.35/f_c

The relation is exact only for a single pole but serves as a common estimate; a 100 MHz bandwidth implies a rise time of about 3.5 ns.

Inductive Kick and Flyback

Continuity of inductor current matters most when a switch opens. Consider a relay coil, modeled as an inductance L in series with its winding resistance R, connected between a DC supply V_S and a transistor that switches it to ground. After turn-on, the coil current rises as i(t) = I₀(1 − e^−tR/L) toward I₀ = V_S/R. At turn-off, the inductor must still carry I₀ the instant after switching. With no path through the switch, the coil voltage reverses and climbs until the current finds one, driving the transistor's terminal far above the supply. This is the inductive kick.

The ideal model predicts an unbounded voltage. A real coil forces its current into the stray capacitance C_p of the winding, the wiring, and the switch, forming a parallel LC circuit. Neglecting losses, energy conservation sets the peak: ½LI₀² = ½C_pV_pk², so

V_pk ≈ I₀ √(L/C_p)

The quantity √(L/C) is the characteristic impedance of the LC pair, and it recurs in every second-order problem here. For illustration, take a coil of 100 mH and 240 Ω on a 12 V supply, so that I₀ = 50 mA and the stored energy is ½ × 0.1 H × (0.05 A)² = 125 μJ. With 100 pF of stray capacitance, the formula predicts a peak near 1,600 V, ringing at about 50 kHz. In practice the transistor avalanches, a contact arcs, or insulation fails first, so the kick needs a controlled path.

If a clamp holds the coil at a constant voltage V_clamp after turn-off, the loop equation L di/dt = −(Ri + V_clamp) gives a current that reaches zero at

t_z = (L/R) ln(1 + I₀R/V_clamp)

A higher clamp voltage stops the current sooner but stresses the switch more. The table compares common choices for the illustrative coil, whose own time constant, L/R, is about 0.42 ms, treating each diode clamp as a constant voltage.

A plain diode protects the switch best but lets the current, and with it the magnetic force, decay most slowly. TE Connectivity's application note Coil Suppression Can Reduce Relay Life (13C3264) explains the risk: a slowly released armature may lack the momentum to break the microweld that can form when contacts close onto a medium- or high-current load, leaving them welded. The note recommends a Zener diode in series with a general-purpose diode as the best compromise. A resistor clamp also works but draws current whenever the coil is energized, 12 mA in this example.

The same kick can be put to work. A boost converter stores energy in an inductor while its transistor conducts; at turn-off, the inductor current, which cannot stop, flows through a diode into an output capacitor held above the input voltage. A flyback converter delivers the stored energy through a coupled inductor to an isolated secondary winding, and an automotive ignition coil interrupts its primary current to produce the thousands of volts that fire a spark plug. The term flyback comes from cathode-ray-tube televisions, whose horizontal deflection circuits generated the picture tube's high voltage during the beam's rapid retrace, or flyback.

The exponential rise of coil current also assumes a constant L. If a magnetic core saturates, its inductance collapses and the current climbs far faster, so inductors in switching circuits must stay below their rated saturation current. Transistor Switching Circuits applies this result to choosing a clamp for a transistor-driven coil.

Second-Order Circuits: Series and Parallel RLC

A circuit whose energy storage reduces to one inductor and one capacitor is second order. Energy can now move back and forth between the magnetic field and the electric field, and the question becomes whether resistance dissipates it before it can oscillate.

In the series RLC circuit, one current, i, flows through R, L, and C. With no source, Kirchhoff's voltage law gives Ri + L di/dt + v_C = 0. Substituting i = C dv_C/dt and dividing by LC gives

d²v_C/dt² + (R/L) dv_C/dt + v_C/(LC) = 0

In the parallel RLC circuit, one voltage, v, appears across R, L, and C. Kirchhoff's current law gives v/R + i_L + C dv/dt = 0, and substituting v = L di_L/dt gives

d²i_L/dt² + [1/(RC)] di_L/dt + i_L/(LC) = 0

Both equations share one standard form,

d²x/dt² + 2α dx/dt + ω₀²x = 0

where α is the neper frequency, or damping coefficient, in reciprocal seconds, and ω₀ is the undamped natural frequency in radians per second. Trying x = Ae^st turns the differential equation into the characteristic equation

s² + 2αs + ω₀² = 0

whose roots are

s₁, s₂ = −α ± √(α² − ω₀²)

These roots are the circuit's natural frequencies; in transfer-function language they are its poles. The damping ratio, ζ = α/ω₀, compares the rate at which resistance removes energy with the rate at which the inductor and capacitor exchange it, and it alone decides the character of the response.

Resistance acts in opposite ways in the two circuits. Series resistance carries the loop current, so a larger R dissipates more energy per cycle; parallel resistance shunts the tank, so a larger R draws less current and damps less, and an ideal parallel tank with no shunt resistance rings forever. The quality factor, Q, equals 1/(2ζ) in both circuits, tying a circuit's ringing in time to the sharpness of its resonance in frequency.

Overdamped, Critically Damped, and Underdamped Responses

Whether ζ is greater than, equal to, or less than 1, which is the sign of α² − ω₀², divides second-order responses into three kinds.

Overdamped: ζ > 1

The roots are real, negative, and distinct, and the response is the sum of two decaying exponentials:

x(t) = A₁e^s₁t + A₂e^s₂t

No oscillation occurs. The root closer to zero decays slowly and dominates the tail; the other dies away quickly.

Critically Damped: ζ = 1

The roots coincide at s = −α, and the solution needs a second, independent term:

x(t) = (A₁ + A₂t) e^−αt

Critical damping is the boundary between the other two cases. For a circuit stepped from rest, it gives the fastest response that never overshoots the final value.

Underdamped: ζ < 1

The roots form a complex-conjugate pair, s = −α ± jω_d, where

ω_d = √(ω₀² − α²) = ω₀√(1 − ζ²)

is the damped natural frequency. The response is a sinusoid inside a decaying exponential envelope:

x(t) = e^−αt(B₁ cos ω_dt + B₂ sin ω_dt)

The envelope shrinks by a factor of e every 1/α seconds while the circuit rings at ω_d, slightly below ω₀. When damping is light, the ringing takes about Q cycles to fall to e^−π, roughly 4 percent, of its starting amplitude, which gives a quick way to read Q from a waveform. With ζ = 0, as in an ideal LC circuit, the oscillation never decays, and the inductor current peaks each time the capacitor voltage passes through zero.

Fitting the Constants

In each case, the two constants come from the two initial conditions, x(0⁺) and dx/dt(0⁺). Evaluating each form and its derivative at t = 0 gives:

• Overdamped: A₁ + A₂ = x(0⁺) and s₁A₁ + s₂A₂ = dx/dt(0⁺).
• Critically damped: A₁ = x(0⁺) and A₂ − αA₁ = dx/dt(0⁺).
• Underdamped: B₁ = x(0⁺) and ω_dB₂ − αB₁ = dx/dt(0⁺).

A series RLC circuit with L = 1 mH and C = 10 μF shows how resistance moves a circuit between the cases. Here ω₀ = 10,000 rad/s (about 1.59 kHz), √(L/C) = 10 Ω, and critical damping occurs at R = 20 Ω.

Raising the resistance past the critical value slows the circuit, because the slow root moves toward zero; the next section quantifies the effect.

Step Response of RLC Circuits

When a DC source switches into a second-order circuit, the complete response is again the final value plus a natural response:

x(t) = x(∞) + x_n(t)

Here x_n(t) takes the overdamped, critically damped, or underdamped form of the circuit after switching. The constants are fitted to the complete response, not to the natural part alone, so the initial conditions for x_n become x_n(0⁺) = x(0⁺) − x(∞) and dx_n/dt(0⁺) = dx/dt(0⁺).

The most common case is a series RLC circuit, initially at rest, connected at t = 0 to a DC voltage V_S, with the output taken across the capacitor. The capacitor starts at v_C(0⁺) = 0, and because the inductor current also starts at zero, dv_C/dt(0⁺) = i(0⁺)/C = 0. The final value is V_S. Fitting the constants gives:

• Overdamped: v_C(t) = V_S[1 + (s₂e^s₁t − s₁e^s₂t)/(s₁ − s₂)]
• Critically damped: v_C(t) = V_S[1 − (1 + αt)e^−αt]
• Underdamped: v_C(t) = V_S{1 − e^−αt[cos ω_dt + (α/ω_d) sin ω_dt]}

Each expression starts at zero with zero slope, as the inductor requires, and each approaches V_S.

Only the underdamped response overshoots. Its first peak comes when the inductor current returns to zero, at the peak time t_p = π/ω_d, and the overshoot, as a fraction of the step, is

M_p = e^−αt_p = e^{−πζ/√(1 − ζ2)}

Overshoot depends on ζ alone, which makes it the easiest way to recognize damping on a waveform.

At ζ = 0.707, the damping of a second-order Butterworth low-pass filter, the overshoot is only 4.3 percent. As ζ approaches zero, the overshoot approaches 100 percent, so a lightly damped LC circuit stepped from zero rings up to nearly twice the applied voltage.

Returning to the series example with L = 1 mH and C = 10 μF, a 5 V step produces three distinct waveforms:

• R = 10 Ω (ζ = 0.5): the capacitor voltage first reaches 5 V after 0.24 ms, peaks at 5.82 V after 0.36 ms, rings at 1.38 kHz, and stays within 2 percent of 5 V after about 0.81 ms.
• R = 20 Ω (ζ = 1): there is no overshoot; the voltage reaches 90 percent of its final value at 0.39 ms and 98 percent at 0.58 ms.
• R = 50 Ω (ζ = 2.5): there is no overshoot, but 90 percent takes 1.12 ms and 98 percent takes 1.90 ms.

For light to moderate damping, the envelope gives a quick estimate of the 2 percent settling time, t_s ≈ 4/α = 4/(ζω₀), which predicts 0.8 ms for the 10 Ω case. The underdamped circuit arrives first but overshoots and rings, and the overdamped one avoids both but is slow, so designs that value speed and overshoot together commonly aim for ζ between about 0.5 and 1.

In the dual case, a DC current source I_S switched into a parallel RLC circuit at rest drives the inductor current from zero toward I_S along the same three curves, with α = 1/(2RC).

Practical Cases

Each of the following circuits reduces to a first-order or second-order core.

Measuring Time Constants and Damping with an Oscilloscope

A square-wave source and an oscilloscope are enough to measure τ and ζ and to estimate the parasitic values behind them.

Setting Up the Test

Drive the circuit with a square wave whose half-period spans at least five to ten time constants, or comfortably exceeds the expected settling time, so that each edge starts from a settled state. The generator is part of the circuit. Most bench function generators have a 50 Ω output impedance, which adds to the circuit's Thévenin resistance. Many also display the amplitude they would deliver into a 50 Ω load unless their load setting is changed, so an output driving a high-impedance circuit swings about twice the displayed value. Use DC coupling, and trigger on the input edge.

Reading a First-Order Response

For a clean exponential, place cursors on the starting and final levels, then find the time at which the waveform completes 63.2 percent of the change; that interval is τ. Alternatively, measure the 10 to 90 percent rise time and divide by 2.2. For the most careful work, plot the natural logarithm of the distance to the final value against time. A true first-order response gives a straight line with slope −1/τ, and a bend in the line reveals a second time constant, such as dielectric absorption in the capacitor or a second pole in the circuit.

Reading an Underdamped Response

For ringing, measure the period, T_d, between successive peaks, which gives ω_d = 2π/T_d. Then measure x₁ and x₂, the heights of two successive peaks of the same polarity above the final value. Their ratio gives the logarithmic decrement, δ = ln(x₁/x₂), from which

ζ = δ / √(4π² + δ²)

For light damping, ζ ≈ δ/(2π). If two successive peaks sit 1.20 V and 0.45 V above the final value, δ = ln 2.67 ≈ 0.98, ζ ≈ 0.15, and Q ≈ 3.2. When the ringing lasts many cycles, measuring across n periods, with δ = (1/n) ln(x₁/x_n+1), improves accuracy. When only a single overshoot is visible, the overshoot table gives ζ directly. Finally, ω₀ = ω_d/√(1 − ζ²) and α = ζω₀, and if either L or C is known, ω₀ = 1/√(LC) yields the other.

Measurement Pitfalls

• Probe loading. A typical 10× passive probe adds about 10 MΩ in parallel with a capacitance on the order of 10 pF. The resistance shortens the time constant of a high-impedance RC circuit, while the capacitance adds to any small circuit capacitance, lengthening a time constant or lowering a ringing frequency. Probes and Adapters covers probe types and loading in detail.
• Ground-lead ringing. The long ground clip on a passive probe forms a series RLC circuit with the probe's input capacitance, and a fast edge rings it. A Tektronix patent on low-inductance ground leads, U.S. Patent 4,838,802, issued in 1989, puts the inductance of conventional ground leads on the order of 200 to 300 nH and the capacitance of typical high-frequency probes at 10 to 15 pF, a combination that resonates between about 75 and 113 MHz. Use the short ground spring supplied with the probe, and distrust any ringing that changes when the ground lead moves.
• Instrument rise time. The oscilloscope and probe have their own rise time, about 0.35 divided by their bandwidth, or somewhat more for a flat, sharply falling response. Rise times combine approximately as the square root of the sum of their squares, so an instrument three times faster than the signal adds about 5 percent to the measured rise time, and one five times faster adds about 2 percent.
• Nonideal components. Electrolytic capacitors have significant equivalent series resistance and leakage, and Class II ceramic capacitors lose capacitance under DC bias, so a measured τ that disagrees with the nominal RC product is often reporting the truth about the part.

Simulating Transients with SPICE

SPICE transient analysis solves the same differential equations numerically, one time step at a time, including nonlinear devices beyond the reach of hand analysis. The following netlist simulates the series RLC example with R = 10 Ω. Its element lines and .tran statement use the classic Berkeley SPICE syntax that ngspice and most other simulators accept.

* Series RLC step response, zeta = 0.5
V1 in 0 PULSE(0 5 0.1m 1n 1n 10m 20m)
R1 in n1 10
L1 n1 out 1m
C1 out 0 10u
.tran 1u 3m
.end

The source steps from 0 to 5 V at 0.1 ms with 1 ns edges, and .tran 1u 3m requests 3 ms of simulated time with a 1 μs output step. A few details decide whether such a simulation can be trusted:

• Time step. The full ngspice form is .tran tstep tstop <tstart <tmax>> <uic>. The maximum internal step, tmax, defaults to the smaller of tstep and (tstop − tstart)/50. Set it explicitly when a long run must resolve fast ringing.
• Initial conditions. By default the simulator first computes a DC operating point, with capacitors open and inductors shorted, and starts from it; the source's initial value of 0 V gives the at-rest start used in the hand analysis. A .ic v(out)=2.5 line forces that node voltage during the operating-point solution and then releases it when the run begins. Adding uic to the .tran line skips the operating point altogether: the run starts from IC= values on individual capacitors and inductors, or from node voltages on a .ic line, and takes any node given neither as zero, a convenient way to set up a pure natural response.
• Edges. If a pulse source omits its rise and fall times, ngspice uses tstep for both. Choose edges much faster than the circuit's response but not needlessly sharp, since sharp edges force tiny time steps.
• Integration artifacts. By default, ngspice integrates with the trapezoidal method, and it offers Gear integration through .options method=gear. Trapezoidal integration adds no numerical damping, so a time step that is long compared with a fast time constant, typically just after an abrupt edge, can produce trap ringing: a spurious sawtooth that alternates between successive time steps. Reducing tmax removes it, while real ringing survives. Gear integration also suppresses it, but its numerical damping can hide ringing that a lightly damped circuit really has.
• Sanity checks. Compare the result with hand analysis. For this netlist, the output should peak near 5.82 V about 0.36 ms after the edge and ring at about 1.38 kHz. A large mismatch usually means a units error, and the classic one is writing 1M for one megohm: SPICE reads M as milli, so a megohm must be written 1meg.

When Laplace Methods Are Quicker

The time-domain method suits circuits with one or two energy-storage elements, step inputs, and a single switching instant. Beyond that, the Laplace transform usually reaches the answer sooner:

• Initial conditions become sources. In the s-domain, a charged capacitor or a current-carrying inductor becomes an impedance plus a source, so nodal or mesh analysis absorbs the initial conditions without separate derivative conditions at 0⁺.
• Algebra replaces calculus. Circuits of third and higher order, or with several coupled loops, lead to simultaneous algebraic equations rather than simultaneous differential equations.
• Inputs beyond the step. Ramps, exponentials, and sinusoids switched on at t = 0 are handled like steps, and the transient and steady-state parts emerge from one calculation.
• Poles tell the story. The characteristic roots found here are the poles of the circuit's transfer function; their positions in the complex plane show whether a response decays, rings, or grows, and tie the transient response to the frequency response.

For first-order circuits the classical method is still quicker, and it remains the better teacher, because every step has a physical meaning.

Summary

Transient response comes down to energy that cannot move instantly. Capacitor voltage and inductor current are continuous, so they carry a circuit across each switching instant while every other quantity is free to jump. A first-order circuit then moves exponentially toward its final value with a time constant of R_ThC or L/R_Th, completing 63 percent of the change in one time constant and more than 99 percent in five. A second-order circuit follows the roots of s² + 2αs + ω₀² = 0, and its damping ratio decides whether a step response rises slowly, rises as fast as possible without overshoot, or rings at ω_d and overshoots by an amount set by ζ alone.

The same results set the timing of a power-on reset and a debounce filter, guide the choice of a relay-coil clamp, size an RC snubber, and explain why hot-plugging a supply into ceramic capacitors can nearly double the input voltage. An oscilloscope measures τ and ζ directly, SPICE checks circuits too large to solve by hand, and Laplace methods take over as circuits grow in order.

Related Topics