Damped & Driven Oscillations · Guide · glidepathscience.org
GuideDamped & Driven Oscillations
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook takes them. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about a quarter of an hour to read. Each section ends with a worked example that is not one of the sim’s own problems. The ideal oscillator, its period and energy, are in Simple Harmonic Motion & Pendulums; this Guide starts where real oscillators lose energy and are pushed.
Part 1 · The essentials
Math you’ll need: algebra, square roots and percentages; two optional boxes use a derivative.
The damping force
A block on a spring, pulled aside and let go, would oscillate forever if nothing took its energy. Real oscillators always lose energy: to air, to friction in the spring itself, or on purpose to a damper. A car’s shock absorber is a damper: a piston that must push oil through small holes. The faster the piston moves, the harder the oil resists, and for a well-made damper the force is simply proportional to the velocity:
$F_{damper} = −bv$
The constant $b$, the damping coefficient, is measured in kilograms per second (the same as N·s/m), and the minus sign says the force always opposes the motion. Compare it with the spring’s force, $−kx$, which opposes the displacement. The two peak at opposite points of the cycle. At a turning point the block stops for an instant, so the damper pushes not at all while the spring pulls hardest; passing through the middle the block is fastest, so the damper pushes hardest while the spring does nothing.
Whatever work the damper does becomes heat in its oil. That energy comes from the oscillation, so the swing shrinks. For light damping it shrinks slowly, by the same fraction every cycle, and the rhythm barely changes.
Three regimes, and critical damping
How much damping a spring and mass need depends on both. Push a block down and let go, and three different things can happen:
- Underdamped: it overshoots and oscillates, each swing smaller, until it settles. A guitar string and a car with worn shock absorbers are underdamped.
- Overdamped: it creeps back without ever passing its rest position, slowed by a damper so strong that it also resists the spring’s pull back.
- Critically damped: the boundary between the two, the least damping that returns without overshooting, and the fastest return without a bounce.
The boundary falls at the critical damping coefficient
$b_c = 2\sqrt{mk}$
and any damper is compared with it by the damping ratio $ζ = b/b_c$ (zeta). Below 1 the block bounces; at 1 it is critically damped; above 1 it is overdamped. A heavier block or a stiffer spring needs a stronger damper to reach the same $ζ$: load a car with passengers and the same shock absorbers give a slightly bouncier ride.
Critical damping is not always what a designer wants. A door closer is set near it, so the door shuts quickly without slamming past the frame. A car’s shock absorbers are usually set well below it, at a damping ratio of roughly 0.2 to 0.4 in typical passenger cars: the car rises past level once, a little, and then stops. The reason is comfort, and the Isolation section below explains it. The size of that first rebound, as a share of the push, is $e^{−πζ/\sqrt{1−ζ^2}}$: 73 % for a worn $ζ$ = 0.10, 10 % at $ζ$ = 0.59, and none at all from $ζ$ = 1.
Energy lost per cycle
At a peak of its swing the block is momentarily still, so all of the oscillation’s energy is in the spring: $E = \tfrac12kA^2$. Comparing two successive peaks therefore measures the energy lost in one cycle:
fraction lost $= 1 − (A_2/A_1)^2$
Because the energy goes as the square of the amplitude, a small fall in amplitude is a larger fall in energy: peaks 10 % lower mean 19 % less energy. With light damping every cycle takes the same fraction, so the peaks fall by the same factor each time, a decay called exponential. The quality factor $Q = 1/2ζ$, which the Simple Harmonic Motion sim reads from a ring-down, says how long that takes: for light damping the amplitude falls to $1/e$ in about $Q/π$ cycles.
Driven oscillations: the drive sets the rhythm
Now push the block with a force that varies like a cosine, $F_0\cos ωt$, at a frequency $f = ω/2π$ of your choosing. The block’s own natural frequency is $f_0 = \frac{1}{2π}\sqrt{k/m}$. At first the motion is a muddle: the push starts the block moving at both frequencies. But the part at its own frequency is a free oscillation, and damping removes it, just as it would after a release. What remains, the steady state, moves at exactly the drive’s frequency, whatever its own frequency is.
The steady state is not in step with the push, in general, and how far behind it lags tells what the push is doing:
- Well below $f_0$ the push is slow, the block barely accelerates, and the spring alone balances the push: the block follows in step, stretched about $F_0/k$.
- At $f_0$ the block lags a quarter cycle behind. Then the push is largest exactly when the block moves fastest in the push’s direction: it is in step with the velocity, and every push adds energy.
- Well above $f_0$ the block cannot keep up; it moves half a cycle behind, against the push, and only a little.
A child on a swing knows this: push at the right moment, as the swing moves away from you through the bottom, and each push adds energy.
How high resonance climbs, and where the energy goes
At resonance something neat happens to the forces. The spring’s force and the force needed to accelerate the block (its inertia, $ma$) are equal and opposite there, so they cancel, and the push is left balancing only the damper. Since the push is then in step with the velocity, their peaks match:
$F_0 = bv_{max}$, so $A = v_{max}/ω_0 = F_0/(bω_0)$
Only the damping limits the swing at resonance: halve $b$ and the amplitude doubles. Away from resonance the spring (below) or the inertia (above) does the limiting, and the damping hardly matters. So damping lowers and broadens the peak of the response curve without changing its far ends.
In the steady state the amplitude stays the same cycle after cycle, so the energy stored, $K + U$, does too. Every joule the drive puts in, the damper turns into heat. The damper’s average power over a cycle is
$P = \tfrac12bv_{max}^2$
(the average of $v^2$ over a cycle is half its peak), and at resonance, with $v_{max} = F_0/b$, $P = F_0^2/2b$. Less damping means more power absorbed at resonance: the motion grows until the damper takes as much as the push gives.
A shaken base, and isolation
Drive the oscillator a different way: shake its base instead of pushing the block. A car body on its springs over a bumpy road, an instrument on a vibrating floor and a washing machine on its feet (seen from the floor’s side) are all oscillators driven through their mounts. The question is how much of the shaking gets through.
Below the natural frequency everything moves together: the block follows the base. Near $f_0$ the block moves more than the base, the resonance a driver feels as the car bobbing on a gently wavy road. But well above $f_0$ the block’s inertia keeps it from following, and it moves less than the base. The crossover is exactly at $\sqrt2\,f_0$: above it, the mount passes less than it receives. So to isolate a vibration, mount it softly, with a natural frequency well below the shaking frequency.
Damping has a surprising role. Near resonance it helps, cutting the peak. Above $\sqrt2\,f_0$ it hurts, because the damper’s force, proportional to the relative speed, carries the base’s fast motion straight to the block. That is why cars are not critically damped: a softer damper passes less of a rough road’s fast jolts, at the cost of a little rebound after a big bump.
Part 2 · Going further
Math you’ll need: calculus, linear differential equations with constant coefficients, complex exponentials, logarithms.
The equation of motion and its solutions
Newton’s second law for the driven, damped block is $m\ddot x + b\dot x + kx = F_0\cos ωt$. It is linear, so its general solution is a particular solution plus the general solution of the free equation. The particular solution is the steady state, $A\cos(ωt − δ)$; the free solution, the transient, carries two constants that fit the initial position and velocity, and it decays. However the block starts, after a few decay times only the steady state is left: it depends on the drive and the oscillator, never on the start.
Trying $x = e^{st}$ in the free equation gives $ms^2 + bs + k = 0$, so
$s = −γ ± \sqrt{γ^2 − ω_0^2}$, with $γ = b/2m$, $ω_0 = \sqrt{k/m}$
Underdamped ($γ < ω_0$), the roots are complex and the motion is $e^{−γt}$ times an oscillation at $ω_d = \sqrt{ω_0^2 − γ^2}$. Critically damped ($γ = ω_0$, the same condition as $b = b_c$), the roots merge and $x = (c_1 + c_2t)e^{−γt}$. Overdamped, both roots are real and negative; the motion is a sum of two decays, and the slower one, $γ − \sqrt{γ^2 − ω_0^2}$, sets how long the return takes. As $ζ$ grows that slow rate falls toward $k/b$, so heavy damping returns ever more slowly: its time constant tends to $b/k$.
Q, exactly
The quality factor is defined as $Q = ω_0/2γ = \sqrt{mk}/b = 1/2ζ$, at any damping. For high Q it has two familiar meanings: the energy lost per cycle is about $2π/Q$ of the energy stored, and the amplitude falls to $1/e$ in about $Q/π$ cycles. Both are approximations. Exactly, successive maxima of $x$ are one damped period $T_d = 2π/ω_d$ apart and fall by $e^{−γT_d}$, so the energy kept from one peak to the next is
$e^{−2γT_d}$, with $2γT_d = \dfrac{2π}{Q\sqrt{1 − 1/4Q^2}}$
Expanding for large Q gives back $1 − 2π/Q$. At low Q the rule fails badly: it would say an oscillator with Q below 2π loses more than all its energy in a cycle.
The response curve, three resonances and the phase
Writing the drive as $F_0e^{iωt}$ and trying $x = Xe^{iωt}$ gives $X = F_0/(k − mω^2 + ibω)$. Its size and angle are the steady amplitude and the lag:
$A = \dfrac{F_0}{\sqrt{(k − mω^2)^2 + (bω)^2}}$, $δ = \text{atan2}(bω,\ k − mω^2)$
The amplitude is largest where the denominator is smallest, at $ω_{peak} = ω_0\sqrt{1 − 2ζ^2}$, below both $ω_0$ and the free ringing frequency $ω_d = ω_0\sqrt{1 − ζ^2}$; for $ζ ≥ 1/\sqrt2$ there is no peak at all. The velocity amplitude, $ωA = F_0/\sqrt{b^2 + (mω − k/ω)^2}$, peaks exactly at $ω_0$ for any damping, and so does the power absorbed, $\tfrac12bω^2A^2$; the acceleration amplitude $ω^2A$ peaks above $ω_0$. “The resonance frequency” needs saying which.
The lag rises from 0 through exactly 90° at $ω_0$ toward 180°. Note the quadrant: above resonance $k − mω^2 < 0$, and a plain arctangent would give a negative angle. In the phasor picture, the forces on the block are the drive $F_0$, the spring $−kX$ (opposite $X$), the damper $−ibωX$ (a quarter turn behind $−X$) and they add to $ma = −mω^2X$. Below resonance the spring’s arrow dominates and the drive balances it; at resonance the spring’s arrow alone equals $ma$ (the spring supplies exactly the force the inertia needs), so the drive’s and the damper’s arrows cancel; above it the drive mostly fights inertia. With heavy damping the damper’s arrow can be the longest even off resonance.
Ring-up: a sharp resonance is slow
Start an oscillator from rest and drive it at resonance. The steady state is $A\sin ω_0t$ (a quarter cycle behind the cosine drive), and to start from rest the transient must cancel its initial velocity: $x = A[\sin ω_0t − (ω_0/ω_d)e^{−γt}\sin ω_dt]$. For light damping that is an oscillation inside the envelope $A(1 − e^{−γt})$, which reaches 90 % of $A$ at $t_{90} = \ln 10/γ = 2Q\ln 10/ω_0$.
The build-up and the ring-down share one time constant, $1/γ$, and since the resonance’s width is $2γ$, a sharper resonance is a slower one: width times build-up time is fixed. A frequency sweep that crosses the resonance in less than a few $1/γ$ reaches the peak before the motion has built up, so its measured peak comes late and low, and then rings (the Frequency Sweep tab shows it).
Base excitation, isolation and sensors
Shake the base as $u = Y_0\cos ωt$. The spring and damper connect the block to the base, so $m\ddot x + b(\dot x − \dot u) + k(x − u) = 0$, which is the force-driven equation with $F = ku + b\dot u$, an amplitude $Y_0\sqrt{k^2 + (bω)^2}$. The block’s amplitude over the base’s is the transmissibility, with $r = ω/ω_0$:
$T = \dfrac{\sqrt{1 + (2ζr)^2}}{\sqrt{(1 − r^2)^2 + (2ζr)^2}}$
The same function gives the fraction of a mounted machine’s shaking force passed to the floor. Every curve equals 1 at $r = \sqrt2$; above that $T < 1$, and $T$ rises with $ζ$.
A vibration sensor is a mass on a spring inside a case bolted to the moving object; it reads the mass’s motion relative to the case, $z = x − u$, with $z/Y_0 = r^2/\sqrt{(1 − r^2)^2 + (2ζr)^2}$. Well above $f_0$ the mass stays put and $z ≈ −u$: a seismometer, with a soft spring, usually damped near $ζ$ = 0.7 so the reading is within a few percent of the true motion from about twice $f_0$ up. Well below $f_0$, $z ≈ ω^2u/ω_0^2$, the case’s acceleration over $ω_0^2$: an accelerometer, with a stiff spring so the flat range is wide.
Fluid drag, and the same equation in a circuit
A paddle in water feels drag that grows as the speed squared, $F = −cv|v|$. The equation is no longer linear, but for light drag each cycle is nearly sinusoidal, and the energy it loses per cycle, $\oint c|v|^3dt = \tfrac83cω^2A^3$, compared with the energy stored, $\tfrac12kA^2$, gives $dA/dt = −(4cω/3πm)A^2$. So $1/A$ grows linearly with time, $1/A = 1/A_0 + (4cω/3πm)t$: a first-order result, as for an equivalent linear damper $b_{eq} = 8cωA/3π$ that weakens as the swing shrinks. It is exact in the limit where each cycle loses only a small part of its amplitude, $8cA/3m ≪ 1$; the sim integrates the exact equation, and even when the first cycle takes half the swing the steps in $1/A$ stay within 2 % of $8c/3m$. (Linear damping makes $\ln A$ linear in time; dry friction makes $A$ itself linear.)
A series RLC circuit obeys $L\ddot q + R\dot q + q/C = V(t)$, the same equation with $m → L$, $b → R$, $k → 1/C$ and $x → q$. Everything above carries over: $ω_0 = 1/\sqrt{LC}$, $Q = (1/R)\sqrt{L/C}$, ring-down, resonance, the phase of the current against the voltage, the power in the resistor. AC Circuits & Transformers takes the circuit further.
Reference
Common mistakes
Part 1
- Putting the damper’s largest force at the turning points: that is the spring’s; the damper’s follows the speed.
- Thinking the most damping returns fastest: critical damping does, and heavy damping creeps.
- Forgetting the 2 in $b_c = 2\sqrt{mk}$, or that a heavier load lowers $ζ$.
- Comparing amplitudes for energy: energy goes as $A^2$.
- Expecting a driven oscillator to keep its own frequency: the steady state has the drive’s.
- Putting the push in step with the position at resonance: it is in step with the velocity.
- Thinking damping lowers the whole response curve: it sets the peak; the far ends hardly change.
- Choosing stiff mounts to isolate a vibration: soft mounts, $f_0$ well below the shaking.
Part 2
- Using $1/γ$ or $b/k$ for an overdamped return when $ζ$ is near 1: use the slow root.
- Applying $2π/Q$ at low Q: the energy kept per cycle is $e^{−2γT_d}$.
- Speaking of one resonance frequency: displacement, velocity and acceleration peak at three.
- Taking the arctangent’s angle above resonance: the lag is past 90°.
- Adding damping to improve isolation above $\sqrt2\,f_0$: it makes it worse.
The equation sheet
| Part 1 | |
|---|---|
| damping force | $F_{damper} = −bv$ |
| critical damping | $b_c = 2\sqrt{mk}$, $ζ = b/b_c$ |
| first rebound | $e^{−πζ/\sqrt{1−ζ^2}}$ of the push |
| energy at a peak | $E = \tfrac12kA^2$ |
| lost per cycle | $1 − (A_2/A_1)^2$ |
| natural frequency | $f_0 = \frac{1}{2π}\sqrt{k/m}$ |
| at resonance | $F_0 = bv_{max}$, $A = F_0/(bω_0)$ |
| power | $P = \tfrac12bv_{max}^2 = F_0^2/2b$ (resonance) |
| isolation | less gets through for $f > \sqrt2\,f_0$ |
| Part 2 | |
|---|---|
| roots | $s = −γ ± \sqrt{γ^2 − ω_0^2}$ |
| decay rate | $γ = b/2m$ |
| quality factor | $Q = ω_0/2γ$ |
| kept per cycle | $e^{−2γT_d}$ |
| steady state | $A = F_0/\sqrt{(k − mω^2)^2 + (bω)^2}$ |
| lag | $δ = \text{atan2}(bω, k − mω^2)$ |
| displacement peak | $ω_{peak} = ω_0\sqrt{1 − 2ζ^2}$ |
| ring-up | $t_{90} = \ln 10/γ$ |
| a shaken base | $D = \sqrt{(1 − r^2)^2 + (2ζr)^2}$ |
| transmissibility | $T = \sqrt{1 + (2ζr)^2}/D$ |
| a sensor | $z/Y = r^2/D$ |
| fluid drag | $1/A = 1/A_0 + (4cω/3πm)t$ |
| series RLC | $Q = (1/R)\sqrt{L/C}$ |
Symbols: $m$ the mass; $k$ the spring constant; $b$ the damping coefficient, $b_c$ its critical value and $ζ = b/b_c$; $γ = b/2m$; $ω_0 = \sqrt{k/m}$ and $f_0 = ω_0/2π$; $ω_d = ω_0\sqrt{1 − ζ^2}$ and $T_d = 2π/ω_d$; $A$ an amplitude, $v_{max}$ the speed amplitude; $F_0$ the drive’s amplitude and $ω$ its angular frequency; $δ$ the lag; $Q$ the quality factor; $r = ω/ω_0$; $Y$ the base’s amplitude and $z$ a sensor’s relative motion; $c$ a fluid drag coefficient; $L$, $R$, $C$ inductance, resistance, capacitance.
Sources and further reading
- OpenStax, University Physics Volume 1, chapter 15, “Oscillations,” §15.5 “Damped Oscillations” and §15.6 “Forced Oscillations,” CC BY-NC-SA 4.0 (openstax.org): the Intro track follows their order and notation.
- Vibration isolation, transmissibility and seismic sensors are standard in mechanical-vibrations texts; the car damping ratios quoted are a typical range, not a specification.
- This sim’s Field Manual (H) for the model and its limits; Simple Harmonic Motion & Pendulums for the ideal oscillator, Q from a ring-down, dry friction and the bandwidth; AC Circuits & Transformers for the RLC circuit.
Damped & Driven Oscillations v1.0 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org