AC Circuit Analysis

AC circuit analysis describes how circuits behave when the sources vary with time rather than holding constant. The shift from DC brings three things that DC analysis never has to confront: a phase relationship between voltage and current, a response that depends on frequency, and elements that store energy and return it instead of dissipating it. Every one of them follows from the fact that inductors and capacitors relate voltage to the rate of change of current, and current to the rate of change of voltage.

The central tool is the phasor, which replaces the differential equations of the time domain with algebra on complex numbers. That substitution is not free. Phasor analysis applies to a circuit that is linear and time invariant, driven by sinusoids at a single common frequency, and observed after the transient has died away. Within those limits it is exact and extraordinarily efficient. Outside them, the circuit needs the time-domain methods described later in this article, or a full Laplace or numerical treatment.

Sinusoidal Waveforms and Time-Varying Signals

A sinusoidal voltage is written

v(t) = V_m sin(ωt + φ)

where V_m is the amplitude in volts, ω is the angular frequency in radians per second, and φ is the phase angle. The angular frequency relates to the ordinary frequency by ω = 2πf, and the period is T = 1/f. Those three parameters define the waveform completely, and the rest of AC analysis is bookkeeping about how a circuit changes them.

Sinusoids earn their central place for a structural reason. Any periodic signal that satisfies the Dirichlet conditions, which are mild enough that every waveform a physical circuit produces meets them, decomposes into a sum of sinusoids at integer multiples of its fundamental frequency. Knowing a circuit's response at each of those frequencies therefore predicts its response to the whole signal. The step that reassembles the parts, however, is superposition, and superposition requires linearity. A nonlinear circuit generates frequencies the input never contained, and no amount of sinusoidal analysis will describe it.

Key Parameters of AC Waveforms

The root-mean-square value carries the most practical weight, because it is the value that determines heating in a resistance. It is defined as the square root of the mean of the squared waveform over one period:

V_rms² = (1/T) ∫₀^T v²(t) dt

Several measurements follow from that definition and from the shape of the waveform:

Peak value (V_m or I_m): The maximum instantaneous value of the waveform.
Peak-to-peak value: The difference between the most positive and most negative excursions, equal to 2V_m for a symmetric waveform.
RMS value: The DC value that would dissipate the same average power in a resistance. For a sinusoid, and only for a sinusoid, V_rms = V_m/√2 ≈ 0.707V_m. A symmetric square wave has V_rms = V_m; a symmetric triangle wave has V_rms = V_m/√3 ≈ 0.577V_m.
Average value: Zero over a complete cycle for any symmetric alternating waveform. The figure quoted in practice is the rectified average, taken over a half cycle, which is 2V_m/π ≈ 0.637V_m for a sinusoid.
Form factor: The ratio of RMS to rectified average, equal to π/(2√2) ≈ 1.11 for a sinusoid, 1 for a square wave, and 2/√3 ≈ 1.15 for a triangle wave.
Crest factor: The ratio of peak to RMS, equal to √2 ≈ 1.414 for a sinusoid, 1 for a square wave, and √3 ≈ 1.732 for a triangle wave.

These distinctions are not academic. An averaging multimeter rectifies the input, measures its average, and multiplies by the sinusoidal form factor of 1.11. Presented with the peaky current that a rectifier-input power supply draws, it reports a value that is simply wrong, and the error grows with the crest factor of the waveform. True-RMS instruments exist because that error is large enough to matter.

Phasor Representation and Complex Numbers

A phasor is a complex number that carries the amplitude and phase of a sinusoid at a known frequency. Representing signals this way turns the differential equations of an AC circuit into linear algebraic equations, which is the single largest simplification available in circuit theory.

The method rests on conditions that deserve stating plainly, because a phasor result obtained outside them is meaningless. The circuit must be linear, so that superposition holds. It must be time invariant, so that one fixed relationship describes it. Every source must be sinusoidal at one common frequency, since phasors belonging to different frequencies cannot be added. And the circuit must have reached sinusoidal steady state, with the natural response decayed to insignificance. A circuit driven at several frequencies at once is handled by solving each frequency separately and superposing the time-domain results, which is legitimate only because the circuit is linear.

The mathematical foundation is Euler's formula, e^jθ = cosθ + j sinθ. Here j is the imaginary unit, defined by the relation

j² = −1

Defining j by that equation rather than as the square root of −1 avoids a genuine ambiguity: the square-root function is single valued only once a branch has been chosen, and both j and −j satisfy the defining relation. Engineering writes j rather than i so that the symbol does not collide with instantaneous current.

By convention phasors are referenced to the cosine. A signal v(t) = V_m cos(ωt + φ) is the real part of V_m e^{j(ωt + φ)}. Because every signal in the circuit shares the factor e^jωt, the analysis divides it out and works with what remains, the phasor

V = V_m e^jφ = V_m∠φ

A signal written as a sine converts first, using sin(ωt + φ) = cos(ωt + φ − 90 degrees). One further convention needs declaring in any piece of work: whether the phasor magnitude is the amplitude V_m or the RMS value V_m/√2. Both are in use and both are consistent. This article writes phasors with the amplitude as magnitude and states RMS values explicitly in the section on power, which is why no factor of one half appears in the power expressions there.

Phasor Arithmetic and Operations

Working with phasors means complex arithmetic in rectangular form (a + jb) and polar form (r∠θ), choosing whichever suits the operation:

Addition and subtraction: Performed in rectangular form, adding real and imaginary parts separately.
Multiplication: Easier in polar form, where magnitudes multiply and angles add.
Division: Also simpler in polar form, with magnitudes dividing and angles subtracting.
Differentiation: Multiplication by jω, which advances the phase by 90 degrees and scales the magnitude with frequency.
Integration: Division by jω, which retards the phase by 90 degrees. The constant of integration belongs to the natural response and does not appear in the steady-state phasor.

Phasor diagrams plot these complex quantities as arrows in the plane, showing at a glance which voltages add in phase, which oppose, and how the resultant compares with the individual contributions. They make the cancellation at resonance and the geometry of power factor correction immediately visible.

Complex Impedance and Admittance

Impedance extends Ohm's law to the phasor domain as the ratio Z = V/I. It is generally complex, with a real part equal to the resistance R and an imaginary part equal to the reactance X:

Z = R + jX

The magnitude |Z| gives the ratio of voltage amplitude to current amplitude, and the angle of Z equals θ_v − θ_i, the angle by which the voltage leads the current. That angle is the φ used throughout the section on power below. Impedance is a complex ratio, not a phasor: it corresponds to no sinusoid of its own and carries no time dependence.

Component Impedances

Each ideal element has a characteristic impedance:

Resistor: Z_R = R, purely real, with voltage and current in phase.
Inductor: Z_L = jωL, purely imaginary and positive, so the voltage leads the current by 90 degrees.
Capacitor: Z_C = 1/(jωC) = −j/(ωC), purely imaginary and negative, so the current leads the voltage by 90 degrees.

Reactance and impedance are not interchangeable, and conflating them is the most common source of sign errors in AC work. Impedance is the complete complex quantity R + jX. Reactance is the imaginary part alone, measured in ohms, and it is signed: an inductor contributes X = +ωL, a capacitor contributes X = −1/(ωC). The symbols X_L and X_C are conventionally used for the unsigned magnitudes ωL and 1/(ωC), which is why the resonance condition is written X_L = X_C even though the two reactances carry opposite signs inside Z. Keep the two usages separate and state which one is in play.

The frequency dependence of reactance drives most of the behavior in this article. Inductive reactance rises in proportion to frequency, so an ideal inductor approaches an open circuit at high frequency. Capacitive reactance falls in inverse proportion, so an ideal capacitor approaches a short. Real components abandon those trends at their self-resonant frequencies: interwinding capacitance makes an inductor look capacitive above its self-resonance, and equivalent series inductance makes a capacitor look inductive above its own. A 100 nF ceramic capacitor with roughly 1 nH of loop inductance self-resonates near 16 MHz, which is exactly the region where a decoupling capacitor is expected to be doing its work.

Worked Example: A Series RC Branch

Take a 100 Ω resistor in series with a 1 µF capacitor at 1 kHz. The capacitive reactance is

1/(ωC) = 1/(2π × 1000 Hz × 1 µF) ≈ 159.2 Ω

so the branch impedance, its magnitude, and its angle are

Z = 100 − j159.2 Ω, |Z| ≈ 188 Ω, ∠Z ≈ −57.9 degrees

The negative angle says the current leads the voltage by 57.9 degrees, as a capacitive branch must. Driven by 1 V RMS the branch draws 1 V / 188 Ω ≈ 5.32 mA RMS. The real power is dissipated entirely in the resistor: (5.32 mA)² × 100 Ω ≈ 2.83 mW, which agrees with V_rmsI_rms cosφ because the phase angle φ is the impedance angle and cosφ = R/|Z| ≈ 0.532. The reactive power is V_rmsI_rms sinφ ≈ −4.51 mvar, negative because the branch is capacitive.

Series and Parallel Impedance Combinations

Impedances combine exactly as resistances do, with complex arithmetic replacing real:

Series: Z_total = Z₁ + Z₂ + Z₃ + …
Parallel: 1/Z_total = 1/Z₁ + 1/Z₂ + 1/Z₃ + …

Admittance Y = 1/Z = G + jB, where G is conductance and B is susceptance, often simplifies parallel work because parallel admittances add directly. One caution: G is not the reciprocal of R unless the element is purely resistive. For Z = R + jX the correct relations are G = R/(R² + X²) and B = −X/(R² + X²).

Power in AC Circuits

Power in AC circuits is harder than in DC circuits because voltage and current are generally out of step, and because the answer depends on how the phase angle is defined. Every expression in this section uses the passive sign convention and defines

φ = θ_v − θ_i

the angle by which the voltage leads the current. With that definition an inductive load has φ > 0 and a capacitive load has φ < 0. The sign of the reactive power, and the entire vocabulary of leading and lagging, follows from this choice, so it must be fixed before anything is written down.

Instantaneous power is the product of the instantaneous voltage and current:

p(t) = v(t) i(t) = V_mI_m cos(ωt + θ_v) cos(ωt + θ_i)

A product-to-sum identity separates it into a constant term and an oscillating one:

p(t) = (V_mI_m/2) cosφ + (V_mI_m/2) cos(2ωt + θ_v + θ_i)

The first term is the real power. The second oscillates at twice the source frequency and averages to zero over a cycle, but it is physically present: a single-phase motor produces a torque pulsation at twice the line frequency for precisely this reason, and the vibration is audible. A balanced three-phase load produces none, because the double-frequency terms of its three phases cancel and the total instantaneous power is constant.

Types of AC Power

Real power (P): Measured in watts (W). P = V_rmsI_rms cosφ. This is the average of p(t), and the only component that transfers energy irreversibly into the load.
Reactive power (Q): Measured in volt-amperes reactive (var). Q = V_rmsI_rms sinφ. Under the convention above it is positive for an inductive load and negative for a capacitive one. It measures energy that shuttles between the source and the reactive elements without net delivery.
Apparent power (S): Measured in volt-amperes (VA). S = V_rmsI_rms. Apparent power sizes conductors, switchgear, and transformers, because those are limited by current and insulation rather than by useful output.

In sinusoidal steady state the three are the sides of a right triangle:

S = √(P² + Q²)

That identity holds only when the voltage and the current are both sinusoidal at a single frequency. With distorted waveforms, which describes the rectifier front end of nearly every modern electronic load, the product V_rmsI_rms exceeds √(P² + Q²), and the shortfall is accounted for by a distortion component. IEEE Std 1459-2010, IEEE Standard Definitions for the Measurement of Electric Power Quantities Under Sinusoidal, Nonsinusoidal, Balanced, or Unbalanced Conditions, sets out the definitions used in that case. An instrument that reports cosφ on a distorted current gives the displacement power factor only, which is not the true power factor P/S.

Power Factor and Correction

The power factor is the ratio P/S, and it equals cosφ under sinusoidal conditions. Unity power factor means the load draws the smallest current that can deliver a given real power. It does not mean maximum power transfer, which is a different condition entirely and is treated below, and it says nothing about how usefully the load converts the energy it absorbs.

An inductive load draws a current that lags the voltage, so its power factor is called lagging and its reactive power is positive. A capacitive load draws a leading current, has a leading power factor, and absorbs negative reactive power. Motors, transformers, and magnetic ballasts are inductive, which is why utility loads are overwhelmingly lagging and why correction almost always means adding capacitance.

Worked Example: Power Factor Correction

Consider a 10 kW load on a 240 V, 60 Hz single-phase supply, operating at a power factor of 0.80 lagging:

S = P/PF = 10 kW / 0.80 = 12.5 kVA, and I = S/V = 12.5 kVA / 240 V ≈ 52.1 A

φ = arccos 0.80 ≈ 36.9 degrees, so Q = S sinφ = 12.5 kVA × 0.60 = 7.5 kvar

Raising the power factor to 0.95 lagging leaves the real power untouched and reduces the reactive demand to Q′ = P tan(arccos 0.95) ≈ 3.29 kvar. A parallel capacitor supplies the difference of 4.21 kvar:

C = Q_C/(ωV²) = 4.21 kvar / (2π × 60 Hz × (240 V)²) ≈ 194 µF

The line current falls from 52.1 A to 10.53 kVA / 240 V ≈ 43.9 A, a reduction of about 16 percent. Because feeder loss scales with the square of the current, the I²R loss falls by about 29 percent. The load performs exactly the same work; only the current required to deliver it changes.

Complex Power

Complex power unifies the three quantities in one expression:

S = V I*

where V and I are RMS phasors and I* is the complex conjugate of the current phasor. The real part is P, the imaginary part is Q, and the magnitude is the apparent power. With amplitude phasors instead of RMS phasors the expression carries a factor of one half, S = ½ V I*, and forgetting it is a frequent source of results wrong by a factor of two. Real and reactive power are each conserved across a network, so the sums of P and of Q over all elements are zero. Apparent power is not conserved and does not add.

Maximum Power Transfer

For a source of fixed internal impedance Z_S = R_S + jX_S driving a load Z_L, the load absorbs the greatest power when the load impedance is the complex conjugate of the source impedance:

Z_L = Z_S* = R_S − jX_S

The conjugate, not the equal. The two reactances cancel, which maximizes the current for a given resistance, and the surviving resistive condition R_L = R_S then follows from the DC result. At the match the source dissipates as much power internally as it delivers, so the efficiency is 50 percent.

That penalty explains where conjugate matching belongs and where it does not. It is the right goal for a receiving antenna, an interstage network in an RF amplifier, or any situation in which the source impedance is fixed and the available signal power is small. It is the wrong goal wherever efficiency matters. Power distribution never matches the load to the source impedance, and neither does audio power amplification: a power amplifier is built to behave as a voltage source with an output impedance of a small fraction of an ohm, and it drives a 4 or 8 ohm loudspeaker deliberately unmatched. Matching there would discard half the power inside the amplifier and would make the frequency response follow the loudspeaker's strongly frequency-dependent impedance. Loudspeaker impedance ratings exist so that the amplifier can supply the required current safely, not to satisfy a maximum power transfer condition.

Resonance in AC Circuits

Resonance is the condition in which inductive and capacitive reactances cancel, leaving a circuit that looks purely resistive at one particular frequency. The cancellation produces sharp frequency selectivity and, in high-quality circuits, voltage or current magnification large enough to matter for component ratings. Resonance underlies radio tuning and oscillator frequency control, and it accounts for a good deal of unintended behavior in power systems and signal paths.

Series Resonance

A series RLC branch has impedance Z = R + j(ωL − 1/(ωC)). The reactive part vanishes when ωL = 1/(ωC), which gives the resonant frequency

f₀ = 1/(2π√(LC))

At that frequency:

The impedance is at its minimum and is purely resistive, Z = R.
The current is at its maximum for a given source voltage.
The voltages across L and across C are equal in magnitude and opposite in phase, each Q times the source voltage.
Voltage and current are in phase, so the phase angle is zero and the power factor is unity.
The net reactive power is zero; energy circulates between the inductor and the capacitor.

The quality factor measures how sharp the resonance is. For the series case

Q = ω₀L/R = 1/(ω₀RC) = (1/R)√(L/C)

and the half-power bandwidth follows from it:

BW = f₀/Q

That bandwidth runs between the two frequencies at which the current falls to 1/√2 of its peak and the dissipated power therefore falls to half, which is 3.01 dB below the peak and is universally rounded to 3 dB.

Worked Example: A Series Resonant Circuit

Take L = 100 µH, C = 100 pF, and a total series resistance R = 5 Ω:

f₀ = 1/(2π√(100 µH × 100 pF)) ≈ 1.59 MHz, so ω₀ = 1.0 × 10⁷ rad/s

Q = ω₀L/R = (1.0 × 10⁷ rad/s × 100 µH) / 5 Ω = 200

BW = f₀/Q ≈ 1.59 MHz / 200 ≈ 8.0 kHz

A 1 V drive at resonance therefore develops 200 V across the inductor and 200 V across the capacitor. The two cancel around the loop, so the source never sees them, but the components experience them and must be rated accordingly. Note also that R in this calculation is the total loss around the loop. The inductor's winding and core loss usually dominates it, which is why the achievable Q is a property of the coil far more than of anything else in the circuit.

Parallel Resonance

The ideal parallel RLC circuit, with R, L, and C all connected across the source, resonates at exactly the same frequency, f₀ = 1/(2π√(LC)), because the pure inductive and capacitive susceptances cancel there regardless of R. Its quality factor inverts the series expression:

Q = R/(ω₀L) = ω₀RC = R√(C/L)

so a larger parallel resistance gives a sharper resonance. The rest of the behavior mirrors the series case:

The impedance is at its maximum and equal to R.
The current drawn from the source is at its minimum.
The circulating current in the L and C branches is Q times the source current.
The circuit appears purely resistive, with unity power factor.
The arrangement is commonly called an anti-resonant or tank circuit.

The practical tank is not the ideal circuit, and here the textbook formula needs qualifying. A real tank is a capacitor in parallel with a coil that carries series resistance R_s, and the frequency at which that combination looks purely resistive is lower than 1/(2π√(LC)). Solving for the frequency at which the susceptances cancel gives

f₀√(1 − 1/Q²), where Q = ω₀L/R_s is the coil's quality factor

The correction is negligible for a good coil and significant for a poor one. At Q = 200 it lowers the frequency by roughly 13 parts per million; at Q = 3 it lowers it by about 6 percent. The same caution applies to the definition itself: for a lossy tank the frequency of unity power factor, the frequency of maximum impedance magnitude, and 1/(2π√(LC)) are three slightly different numbers, and quoting a single figure is safe only when Q is high.

Parallel resonant circuits supply the frequency-selective feedback in LC oscillators, form the tuned loads of RF amplifier stages, and appear throughout impedance matching networks. Their high impedance at resonance also makes them useful as traps, blocking one frequency while passing everything else.

Frequency Response and Filters

Frequency response describes how the gain and phase shift of a circuit vary with frequency. The transfer function H(jω) = V_out/V_in states the relationship as a complex ratio, whose magnitude |H(jω)| is the gain and whose angle ∠H(jω) is the phase shift.

A transfer function exists only for a linear time-invariant circuit. A circuit containing a conducting diode, a saturating magnetic core, or a transistor driven beyond its small-signal range has no transfer function in this sense. Such a circuit can be linearized about an operating point, and the resulting transfer function is valid for small excursions about that point and nowhere else.

The First-Order RC Low-Pass

With a resistor in series and a capacitor to ground, voltage division between the two impedances gives

H(jω) = (1/(jωC)) / (R + 1/(jωC)) = 1/(1 + jωRC)

and the cutoff frequency is the frequency at which ωRC = 1:

f_c = 1/(2πRC)

At f_c the magnitude is 1/√2 ≈ 0.707 and the phase is −45 degrees. Because power goes as the square of voltage, the output power into a fixed load is exactly half its low-frequency value there. That is what the cutoff frequency means: it is the half-power point, 3.01 dB down, conventionally called the 3 dB point. Well above f_c the magnitude falls in inverse proportion to frequency, which is 20 dB per decade or 6 dB per octave, and the phase approaches −90 degrees.

With R = 1 kΩ and C = 100 nF, f_c ≈ 1.59 kHz. At ten times that frequency ωRC = 10, so the magnitude is 1/√101 ≈ 0.0995, which is −20.0 dB, and the phase is −84.3 degrees. The straight-line asymptote is accurate a decade away from the corner and 3 dB in error at the corner itself.

Basic Filter Types

Low-pass: Passes low frequencies and attenuates high ones. The simple RC form is H(jω) = 1/(1 + jωRC) with f_c = 1/(2πRC).
High-pass: Passes high frequencies and attenuates low ones. Exchanging the two elements gives H(jω) = jωRC/(1 + jωRC), with the same cutoff frequency and a phase that runs from +90 degrees at low frequency to zero at high frequency.
Band-pass: Passes a band and attenuates both sides of it. A series RLC branch near resonance gives the simplest form, with center frequency f₀ and bandwidth f₀/Q.
Band-stop, or notch: Attenuates a band and passes both sides of it. Useful against interference at a known frequency, such as mains hum or a local transmitter.

Filter order sets the asymptotic rolloff at 20n dB per decade for order n, so a second-order filter falls at 40 dB per decade and a fourth-order at 80. Higher order buys a steeper transition at the cost of more components, greater sensitivity to component tolerances, and more accumulated phase shift, which matters whenever the filter sits inside a feedback loop.

Filter Design Considerations

Practical design trades one property against another, and the classic approximations differ precisely in which trade they make:

Passband ripple: Variation of gain within the passband. A Butterworth response has none and is maximally flat; a Chebyshev response accepts ripple in exchange for a steeper transition at the same order.
Stopband attenuation: The minimum rejection achieved in the stopband, which sets the order required.
Transition band: The frequency span between the edge of the passband and the start of the stopband. Narrowing it is the main driver of filter complexity.
Phase linearity: Constant group delay, which preserves waveform shape. A Bessel response optimizes it and pays with a gentle transition.
Component sensitivity: How far the response moves when components sit at the edges of their tolerance. High-Q sections are the most sensitive and deserve the tightest parts.

Bode Plots and Graphical Analysis

A Bode plot presents frequency response as two graphs against a logarithmic frequency axis: magnitude in decibels, and phase in degrees. The logarithmic axes serve a purpose beyond compressing a wide frequency range. Because decibels are logarithmic, cascading two stages adds their magnitude curves and adds their phase curves, turning the multiplication of transfer functions into addition on the page.

Construction and Interpretation

An asymptotic magnitude plot is assembled from a few features:

Low-frequency gain: The horizontal starting level, set by the transfer function evaluated as the frequency approaches zero.
Real poles: Each bends the slope down by 20 dB per decade at its corner frequency.
Real zeros: Each bends the slope up by 20 dB per decade at its corner frequency.
Asymptotic approximation: Straight segments that meet at the corners and lie 3 dB from the true curve at each one.

A complex-conjugate pole pair behaves differently, and the straight-line construction misses it. The pair changes the slope by 40 dB per decade at once, and near the natural frequency it produces a peak: at ω = ω_n the magnitude of the standard second-order response is exactly 1/(2ζ), which for a lightly damped pair is very large.

Phase plots show a transition spread over roughly two decades around each corner. A pole contributes −90 degrees in total and a left-half-plane zero contributes +90 degrees, and the phase at any frequency is the sum of all contributions. The right-half-plane zero is the exception that catches designers out: it lifts the magnitude by 20 dB per decade like any zero, but its phase contribution is −90 degrees. Loops that contain one, such as a boost converter in continuous conduction, are correspondingly difficult to compensate.

Stability Analysis

Bode plots support the two margins used to judge feedback stability:

Gain margin: The factor by which the loop gain may rise before instability, read at the phase crossover frequency, where the phase passes −180 degrees.
Phase margin: The additional phase lag the loop can tolerate before instability, read at the gain crossover frequency, where the loop gain magnitude is unity, or 0 dB.

Positive margins indicate a stable closed loop when the open-loop transfer function is itself stable and minimum phase. The general test is the Nyquist criterion, which counts encirclements of the point −1 and correctly handles loops that are open-loop unstable or conditionally stable. A conditionally stable loop can be destabilized by reducing the gain, a failure that the margins alone do not reveal. Typical design targets are a gain margin above 6 dB and a phase margin above 45 degrees; 45 degrees of phase margin corresponds to ζ ≈ 0.42 and about 23 percent overshoot when a dominant pole pair governs the response.

Transfer Functions and System Analysis

Transfer functions describe linear time-invariant systems completely. Written in the Laplace domain as H(s) = Y(s)/X(s), the ratio of the transformed output to the transformed input with zero initial conditions, a transfer function encodes every dynamic property of the system. The frequency response H(jω) is that same function evaluated along the imaginary axis, a substitution that is valid when the system is stable.

Poles and Zeros

Factoring the numerator and the denominator exposes the zeros and the poles:

H(s) = K [(s − z₁)(s − z₂) …] / [(s − p₁)(s − p₂) …]

Their positions in the complex s-plane determine the behavior:

Stability: Every pole must lie strictly in the open left half-plane. A pole on the imaginary axis gives a sustained oscillation, and one in the right half-plane gives a growing response.
Oscillatory response: Produced by complex-conjugate pole pairs; real poles alone give pure exponentials.
Damping: Set by the angle of the pole from the negative real axis, with ζ equal to the cosine of that angle. The real part −ζω_n fixes the decay rate and the imaginary part fixes the ringing frequency.
Natural frequency: Equal to the distance of the pole from the origin.

Standard Forms

The second-order low-pass form recurs throughout control and filter work, parameterized by the natural frequency ω_n and the damping ratio ζ:

H(s) = ω_n² / (s² + 2ζω_ns + ω_n²)

The damping ratio sets the character of the response:

ζ < 1, underdamped: An oscillatory approach with overshoot, the fractional overshoot being e^{−πζ/√(1−ζ²)}.
ζ = 1, critically damped: The fastest approach to the final value with no overshoot.
ζ > 1, overdamped: Two real exponentials, slower and without oscillation.

Two values are worth remembering. A damping ratio of 1/√2 ≈ 0.707 gives the maximally flat magnitude response, the second-order Butterworth case, with about 4.3 percent step overshoot. A damping ratio near 0.5 is a common compromise in control loops, trading a little more overshoot for a faster rise.

Transient Response Analysis

Phasor analysis returns only the forced response, the part of the solution that persists once the circuit has settled. The complete solution adds the natural response, which the circuit contributes on its own and which decays. Transient analysis is the study of that second part, and it governs switching behavior, pulse circuits, startup, and the settling of any system that must reach a new state quickly.

First-Order Circuits

A circuit with one energy-storage element responds exponentially, with a time constant set by that element and the resistance it sees:

RC circuit: τ = RC. Charging toward a step V gives v(t) = V(1 − e^−t/τ).
RL circuit: τ = L/R. Current builds as i(t) = I(1 − e^−t/τ).

After one time constant the response has covered 63.2 percent of the way to its final value, and after five time constants it is within 0.7 percent of it. That five-time-constant rule is the usual working estimate for settling time in digital and switching circuits, although a precision analog circuit settling to a part in a thousand needs closer to seven.

Second-Order Circuits

An RLC circuit has two storage elements and three distinct behaviors, distinguished by the damping ratio in exactly the way the standard form above describes:

Overdamped: Two real exponential terms and no oscillation.
Critically damped: The fastest response that does not overshoot.
Underdamped: A decaying sinusoid at the damped natural frequency ω_d = ω_n√(1 − ζ²), which is always lower than ω_n.

Step and Impulse Response

Two standard test signals expose what a system does:

Step response: The reaction to a sudden change in level, which reveals rise time, overshoot, and settling time directly.
Impulse response: The reaction to a very short pulse, which carries the complete description of a linear time-invariant system.
Relationship: The step response is the integral of the impulse response, and the impulse response is the derivative of the step response.

Because the impulse response h(t) characterizes a linear time-invariant system completely, the output for any input follows from it by convolution:

y(t) = x(t) * h(t) = ∫ x(τ) h(t − τ) dτ

with the integral taken over all τ. The asterisk here denotes convolution, not multiplication, and the distinction matters because the Fourier transform turns one into the other. For a stable system the Fourier transform of h(t) is the transfer function H(jω), and convolution in time becomes an ordinary product in frequency. That correspondence is the formal link between the time-domain and frequency-domain views of one circuit, and, like everything else in this article, it depends on the circuit being linear and time invariant.

Practical Applications and Real-World Considerations

The same small set of ideas reappears across every branch of electrical engineering, at scales separated by many orders of magnitude in frequency and in power.

Power Systems

Utility engineering rests on AC analysis:

Power flow analysis: Solving for voltage magnitudes and phase angles at every bus in a network, given the injected real and reactive powers.
Fault analysis: Predicting the currents and voltages during a short circuit, which sets the interrupting ratings of breakers and the settings of protective relays.
Harmonic analysis: Assessing the distortion that nonlinear loads inject, and its heating effect on transformers and neutral conductors.
Reactive power compensation: Capacitor banks and static compensators that support voltage and reduce line current.
Stability studies: Determining whether generators remain in synchronism after a disturbance.

Signal Processing

Analog signal processing is applied frequency response:

Active filters: Operational amplifier circuits that realize precise responses without inductors.
Equalizers: Deliberate frequency-dependent gain, used to shape audio and to compensate cable loss.
Oscillators: Resonant or phase-shift networks that set a stable frequency of operation.
Modulators and demodulators: Frequency translation, which relies on the filtering that separates the wanted product from the unwanted one.
Phase-locked loops: Feedback systems whose loop filter is designed with exactly the Bode techniques described above.

Measurement and Instrumentation

Measurement turns the same relationships around, inferring an unknown impedance from an observed response:

Impedance analyzers: Measuring magnitude and phase against frequency to characterize components and materials.
Network analyzers: Measuring scattering parameters, the high-frequency description that replaces impedance once connections behave as transmission lines.
Lock-in amplifiers: Recovering a small signal buried in noise by multiplying it with a reference at the signal frequency and low-pass filtering the result.
Bridge circuits: Nulling techniques that compare an unknown impedance against known standards with high precision.

Common Design Challenges and Solutions

Most difficulties in practical AC design come from the gap between the ideal elements of the analysis and the physical components on the board. The gap widens with frequency.

Parasitic Elements

Every real component carries unintended inductance, capacitance, and resistance:

Resistors: Lead and body inductance in series, and a small shunt capacitance across the element, which together give film resistors a useful range that ends well below the gigahertz region.
Capacitors: Equivalent series resistance, which produces loss and heating, and equivalent series inductance, which sets the self-resonant frequency above which the part behaves as an inductor.
Inductors: Winding resistance that rises with frequency through the skin and proximity effects, interwinding capacitance that sets a self-resonant frequency, and core loss that grows with flux swing and frequency.
PCB traces: Distributed inductance and capacitance that make a trace a transmission line once its delay is a meaningful fraction of the signal's edge rate.

Mitigation Strategies

Component selection: Choose parts whose self-resonant frequency and loss suit the operating band, rather than judging by nominal value alone.
Layout optimization: Shorten current loops, keep return paths adjacent to their outgoing paths, and provide an uninterrupted reference plane.
Impedance matching: Terminate transmission lines in their characteristic impedance to suppress reflections.
Shielding: Enclose or separate sensitive nodes to limit coupled interference.
Decoupling: Place local capacitance close to each load, chosen and mounted so that its self-resonance falls where the supply current demand is largest.

Advanced Topics and Modern Developments

Several specialized fields extend these fundamentals rather than replacing them, and each is worth knowing exists.

Switched-Mode Power Conversion

Switching converters are nonlinear and time varying, so they have no transfer function in the ordinary sense. The standard approach recovers one anyway:

State-space averaging, which replaces the switching waveform with its average over a cycle.
Small-signal linearization about the operating point, which yields a transfer function valid for modest perturbations.
Control loop compensation designed from the resulting Bode plot, where the right-half-plane zero of the boost and flyback topologies limits the achievable bandwidth.
Input and output filter design, which must account for the interaction between the filter's output impedance and the converter's negative input resistance.

RF and Microwave Circuits

Once physical dimensions approach a wavelength, lumped analysis fails and distributed effects take over:

Transmission line theory, reflection coefficients, and the Smith chart.
Scattering parameters, which describe multiport networks in terms of traveling waves rather than terminal voltages and currents.
Impedance matching networks, where the conjugate match described earlier is the usual design objective.
Distributed element filters built from transmission line sections instead of discrete inductors and capacitors.

Digital Signal Integrity

High-speed digital design is an AC problem wearing digital clothes:

The relationship between edge rate and bandwidth, which determines whether a trace must be treated as a transmission line.
Reflections and terminations on PCB traces.
Crosstalk between adjacent traces, and ground bounce in shared return paths.
Power distribution network impedance, kept low across a wide band so that switching current does not modulate the supply rail.

Troubleshooting AC Circuits

Diagnosing an AC circuit means measuring magnitude and phase against frequency, not voltage alone. A systematic approach isolates the cause faster than probing at random.

Common AC Circuit Problems

Oscillation: Unintended feedback, often through a shared supply impedance or a stray coupling path, that leaves the loop with insufficient phase margin.
Distortion: Nonlinear behavior that generates harmonics absent from the input, usually from clipping, slew-rate limiting, or core saturation.
Noise: Random or interfering signals, whose spectrum usually identifies the source.
Loading effects: A change in behavior when one stage is connected to the next, which the source and load impedances predict.
Unexpected resonance: Peaks or nulls in the response caused by parasitic inductance and capacitance rather than by intended components.

Diagnostic Tools and Techniques

Oscilloscope: Observing waveform shape and measuring the phase between two channels directly.
Spectrum analyzer: Identifying frequency components, harmonics, and interference that a time-domain view hides.
LCR meter: Measuring component values, along with Q and equivalent series resistance, at the frequency of interest rather than at an arbitrary one.
Signal generator: Applying controlled test signals and sweeping frequency.
Vector network or frequency response analyzer: Measuring magnitude and phase against frequency, including loop gain measured by signal injection.

Systematic Troubleshooting Approach

Verify supply voltages and the integrity of ground connections.
Confirm the DC bias conditions before investigating AC behavior, since a wrong operating point invalidates every small-signal assumption.
Isolate sections to identify which stage is responsible.
Compare measurements against calculated values rather than against expectations.
Account for parasitic elements and component tolerances before concluding that a component has failed.
Record what was measured and under what conditions, so that the result can be reproduced.

Conclusion

AC circuit analysis rests on a small number of ideas that repay careful handling. The phasor converts calculus into algebra, but only for a linear, time-invariant circuit in sinusoidal steady state at one frequency. Impedance extends Ohm's law into the complex plane, provided that reactance and impedance stay distinct and their signs are respected. Power splits into a real component that does work, a reactive component whose sign depends on a phase convention that must be stated, and an apparent component whose relationship to the other two holds only for undistorted waveforms. Resonance and frequency response follow directly from the frequency dependence of reactance.

The recurring failures in practice are failures of scope rather than of arithmetic: applying a phasor result to a nonlinear circuit, quoting an RMS conversion factor derived for a sinusoid, matching an impedance where efficiency was the real objective, or trusting the ideal parallel resonance formula for a lossy tank. Knowing where each result stops being true is as much a part of the subject as the result itself, and it is what allows these same tools to carry over into control systems, communications, power electronics, and signal integrity.