14.1 Coupled Oscillators & Normal Modes ended by promising an equation from electronics that matches a damped, driven mass on a spring. Here it is. A resistor, an inductor and a capacitor in series obey the second-order ODE of 5.3 Mechanical Vibrations: Spring–Mass–Damper and 5.6 Forced Oscillations & Resonance, so the damping regimes and resonance carry over with new names. Circuits also give the frequency domain its practical form: instead of solving for , ask what the circuit does to a sine of each frequency. The answer is called impedance, and it is the Laplace transfer function of 6.5 Convolution & Transfer Functions on the imaginary axis.
Transcript
- Two laws govern every circuit. Kirchhoff's current law says the currents into a node sum to zero. Kirchhoff's voltage law says the voltage drops around a closed loop sum to zero. A resistor gives v equals R i, an inductor gives v equals L times di by dt, and a capacitor gives i equals C times dv by dt. Switch a battery onto a resistor and an empty capacitor, and the voltage law gives R C v dot plus v equals V. The capacitor charges as V times one minus e to the minus t over tau, with time constant tau equal to R C. After one time constant it has reached 63.2 percent, and after 4.6 time constants, 99 percent. For a kilo-ohm and a microfarad, tau is 1 ms.
- Add an inductor and the equation turns second order: L q double dot plus R q dot plus q over C equals V of t. Compare it with m x double dot plus c x dot plus k x equals F. The dictionary is exact. Charge is displacement, current is velocity, voltage is force. Inductance is the mass, resistance is the damper, and one over C is the spring. The natural frequency is omega naught equals one over root L C, zeta is R over 2 times root C over L, and the engineer's quality factor is Q equals one over 2 zeta.
- Take R equal to 10 ohms, L equal to 10 millihenries, C equal to 10 microfarads, switched onto a 1 volt battery with the capacitor empty. Then omega naught is 3162.28 radians per second, zeta is 0.1581, and Q is 3.162, so the circuit is underdamped. The current first vanishes at pi over omega d, which is 1.0061 milliseconds, and there the capacitor voltage peaks at 1.6047, an overshoot of 60.47 percent. Raise R to 200 ohms. Zeta is 3.162, the poles are minus 513.2 and minus 19,487, and the capacitor creeps up to 0.9211 at 5 ms with no overshoot. The slow pole sets the rise.
- Drive the circuit with a sine. Differentiating multiplies by j omega, so the components become impedances: R, j omega L, and one over j omega C, and series impedances add. The reactive part vanishes at omega naught, where Z equals R is smallest and the current largest: resonance. Transform the equation and the current transfer function is one over Z of s, so the impedance is the Laplace transform of the impulse response at s equals j omega. At resonance the capacitor voltage is Q times the source, 3.162 for our circuit. The peak sits slightly lower, at 3082.2 radians per second, with height 3.2026.
- Measure the voltage across each component as a fraction of the source, and one circuit gives three filters. The capacitor is a low-pass: gain 1 below omega naught, 10.00 dB at omega naught, and falling at minus 40 dB per decade, minus 39.9 dB at ten times omega naught. The resistor is a band-pass with peak gain exactly 1. The inductor is the mirror image, a high-pass. The band-pass width between the half-power points, minus 3.01 dB, is R over L, which is omega naught over Q. Here the edges are 2701.56 and 3701.56, a width of exactly 1000. An RC low-pass has cutoff one over R C, with phase minus 45 degrees there: 1000 radians per second, 159.15 hertz, for a kilo-ohm and a microfarad.
- Give each loop a charge, and a multi-loop circuit becomes a matrix equation, the analog of the last lesson's mass, damper and stiffness system. Take two identical loops, each with 10 millihenries and 10 microfarads, sharing a coupling capacitor of 100 microfarads. The modes are 1, 1 and 1, minus 1. In the first, nothing flows through the coupling capacitor and omega 1 is 3162.28. In the second it is charged twice over and omega 2 is 3464.10 radians per second. Energy beats between the loops every pi over the difference, 10.41 ms. With a 10 ohm resistor in each loop, every mode is an RLC circuit with alpha 500, and zeta 0.1581 and 0.1443.
- Resistors and capacitors only dissipate, so a two-stage ladder has no oscillation. The matrix is minus 2, 1, 1, minus 1, and its eigenvalues solve lambda squared plus 3 lambda plus 1 equals 0, so lambda is minus 0.382 and minus 2.618, in units of one over R C. Everything here was linear, which is what made impedance possible. The next lesson, pendulums, leaves that world. The pendulum is the oscillator of this lesson only under the small angle approximation, sine theta about theta. Drop it and the syllabus lists the nonlinear pendulum with elliptic integrals and the separatrix, the driven damped pendulum, a Lagrangian introduction, and the double pendulum, with chaos. Superposition fails there, so impedance does not apply.
Kirchhoff’s laws and three components
Two laws govern every circuit (from memory, from physics; they are conservation laws). Kirchhoff’s current law (KCL): the currents into a node sum to zero, because charge does not pile up at a junction. Kirchhoff’s voltage law (KVL): the voltage drops around a closed loop sum to zero, because the electric potential is single valued. Each component relates its voltage to its current ( is charge):
The resistor (ohms) dissipates energy at rate . The inductor (henries) stores and resists changes in current. The capacitor (farads) stores . The number of inductors plus capacitors sets the order of the ODE.
RC and RL: first-order charging
Put a resistor and a capacitor in series with a battery switched on at , with the capacitor empty. KVL gives , and with :
This is the linear first-order equation of 3.4 Linear First-Order Equations & Integrating Factors. After one time constant the capacitor has charged to of , and after to (since ). For and , ms. The series RL circuit is the same equation with the roles swapped: gives with . For mH and , ms again. The rate is the lone eigenvalue.
With and only, is a pure oscillator at , energy sloshing between the magnetic and electric fields. Put all three together and you get the main equation.
The series RLC equation and the analogy
A battery (or any source) in series with , , . KVL: , and with ,
Compare with . The dictionary is exact:
| Mechanical | Electrical |
|---|---|
| displacement | charge |
| velocity | current |
| force | voltage |
| mass | inductance |
| damper | resistance |
| spring | |
| kinetic energy | magnetic energy |
| spring energy | electric energy |
Inductance is inertia, resistance is friction, a small capacitor is a stiff spring. Every result of 5.2 Constant Coefficients & the Characteristic Equation transfers: and
The quality factor is the engineer’s version of , , that is . Energy decays like , so over one period a fraction is lost (large ): a high- circuit rings for many cycles. The poles are , always in the left half plane for : a stable spiral for small , a stable node for large (7.3 Phase Portraits & the Trace–Determinant Plane).
Three regimes and a worked example
With no source, solutions are at those poles. Three cases, set by (or against ):
- , underdamped: oscillation at under an envelope .
- , critically damped: , the fastest return without overshoot.
- , overdamped: two real decays, and the slow one dominates.
Take , mH, F, switched onto a 1 V battery with the capacitor empty. Then rad/s ( Hz), , , , and . So this circuit is underdamped, with rad/s. For the step, the capacitor voltage is , and the current is , which vanishes first at ms. At that instant , so peaks, with
The capacitor voltage is at ms, at ms and at ms. Now raise to : , the poles are and , and the capacitor creeps up to at ms with no overshoot. The fast pole is gone within about ms (five of its time constants) and the slow one, a time constant of ms, sets the rise, close to the ms of a plain RC circuit: a heavily overdamped oscillator behaves like a first-order one.
Driving the circuit: resonance and impedance
Drive with . After the transient (decaying like ) the circuit follows the source at its frequency. Write the source as the real part of (engineers use because means current) and look for a current ; the complex numbers and carry amplitude and phase and are called phasors. Differentiating multiplies by and integrating divides by it, so the three components become
These ratios are impedances. Series impedances add:
Parallel impedances add as reciprocals, like resistors. Here , and the reactive part vanishes at , where is smallest and the current amplitude largest, in phase with the source: resonance.
Here is the link to the Laplace transform. Taking transforms with zero initial conditions, becomes , so the current transfer function is . The impedance is the Laplace transform of the circuit’s impulse response, evaluated at . 13.5 Frequency Response & Bode Plots proves this for the cruise plant. Use for transients and for steady sine.
At resonance, the voltage across the capacitor is , and the inductor has the same magnitude with the opposite sign. For the capacitor sees times the source voltage. No paradox: the circulating energy is large, and the source only supplies the loss in . The peak of sits slightly lower, at rad/s, with height ; for large the two coincide.
Frequency response and the filters
Measure the voltage across each of the three components, as a fraction of the source. Dividing the impedances,
Using and , the first is . The Bode plot shows (decibels) and against . For the circuit above, the capacitor output has gain ( dB) well below , passes through ( dB, since ) at with phase , and falls as above it: dB at , an asymptotic slope of dB per decade. Phase goes from to .
So one circuit, three outputs, three filters:
- Low-pass (): passes slow signals, rejects fast ones.
- Band-pass (): rewrite as , a peak of gain exactly at .
- High-pass (): the mirror image of the low-pass, passing fast signals.
The band-pass bandwidth is the gap between the half-power frequencies, ( dB), where : a quadratic with roots , and their difference is exactly
A high- circuit is selective, as a radio tuner needs. For the circuit above the edges are and rad/s, a width of exactly (and , so is their geometric mean). Change and the band slides without changing width.
The simplest filter has only two parts. The RC low-pass of the charging example, measured across , has , so
The cutoff is , the reciprocal of the time constant, where ( dB) and the phase is . For and F, rad/s ( Hz). Measure across instead and the same circuit is a high-pass with the same cutoff.
Multi-loop circuits as systems
Give each loop a loop charge (the charge circulated around it) and write KVL for each. The result is the matrix equation promised at the end of the last lesson:
with inductance matrix , resistance matrix and “elastance” matrix of terms, the analog of . That expectation was right, with one refinement: was diagonal last time, but need not be, since two coils with overlapping fields have a mutual inductance, a coupling in the mass matrix. A coupling capacitor puts it in , as a coupling spring did. With resistors the modes decouple only if is a combination of and .
Worked example: two coupled LC loops. Two identical loops each have mH and F and share a coupling capacitor F, with voltage . KVL on each loop, with no source:
This is the two-mass system of the last lesson with , , , so the eigenvectors are and . In the first, the two loop currents are equal, nothing flows through , and . In the second, is charged twice over and . So and rad/s, and energy beats between the loops every ms. If each loop also contains a resistor , then is proportional to , so each mode is an RLC circuit with the same and its own : and .
Worked example: a two-stage RC ladder. The source feeds node 1 through , node 1 feeds node 2 through , and each node has a capacitor to ground. KCL at each node:
In matrix form with , whose eigenvalues solve , so and (in units of ). There are no oscillations: resistors and capacitors only dissipate. The slow mode, eigenvector , has both nodes moving together with time constant ; the fast one, , with . It settles at , a stable node in the sense of 7.2 Linear Systems & the Eigenvalue Method.
The lab
The left panel is the time response (input dashed), an exact formula and not a numerical integration. The right panels are the Bode magnitude and phase of the same output, with marked. With the default , mH, F:
- Step input across the capacitor: rad/s, , , overshoot , and the voltage at ms, .
- Press “critical damping”: becomes (steps of ; the exact value is ) and the ringing disappears.
- Sine drive at : the output grows to times the input, phase .
- Measure across the resistor: the band-pass edges are and rad/s, and the width equals . Change and the width stays; set mH and it halves to .
- Across the inductor (high-pass) the peak sits just above , at the same height as the capacitor.
The exercise: gain and step response
The playground reproduces the multi-loop examples. The exercise builds the capacitor gain from impedances and the step response from the ODE.
Concept checks
Problems
A V battery is switched onto a series RL circuit with mH and . Find , the time to reach of the final current, and the energy stored then.
Show worked solution
ms and A, so A. Setting gives ms. The current is then A and the stored energy mJ.
A series RLC has mH, nF and . Find , , the bandwidth of the voltage across the resistor, and the two half-power frequencies.
Show worked solution
rad/s ( kHz), , and the bandwidth is rad/s ( kHz). With , the edges are and rad/s.
A series circuit has , mH and F, driven by a V amplitude source at rad/s. Find , the current amplitude and the capacitor voltage amplitude.
Show worked solution
and . The reactances cancel, so (this is exactly ). The current amplitude is A, in phase with the source. The capacitor voltage is V, larger than the source, as predicts.
For the two-stage RC ladder with , find the eigenvectors of , and use them to write for a unit step from rest when and F. Which mode dominates the late-time approach?
Show worked solution
For , gives , so . For , , so . The steady state is , so write with : and . Then , , and with ms,
.
The slow mode (time constant ms) dominates after about ms; as it should.
What comes next
Everything here was linear, which is what made impedance possible. The next lesson, 14.3 Pendulums: Linear, Nonlinear, Driven & Double, leaves that world. The pendulum becomes the harmonic oscillator of this lesson only under the small-angle approximation . Drop it and the syllabus lists the nonlinear pendulum with elliptic integrals and the separatrix, the driven damped pendulum, a Lagrangian introduction and the double pendulum, with chaos. Superposition fails there, so impedance does not apply.
Key Takeaways
- KCL and KVL plus , , give the ODEs of any circuit; inductors plus capacitors set the order.
- RC and RL charging have or : after , after .
- is the mass-spring-damper with , , . Then , , .
- For , mH, F: rad/s, , , step overshoot at ms, critical .
- Impedance is the Laplace transfer function at . At resonance and the capacitor voltage is times the source.
- One circuit gives three filters: low-pass, band-pass (peak , bandwidth , edges and rad/s here), high-pass. An RC low-pass has cutoff ( dB, ) and rolls off at dB per decade.
- Multi-loop circuits are : normal modes for coupled LC loops ( and rad/s) and real decay rates for RC networks ( and over ).
- Next: 14.3 Pendulums: Linear, Nonlinear, Driven & Double, where the oscillator becomes nonlinear.