second-order ODE damping
A weight hanging on a spring swings and settles; a radio circuit rings at one particular frequency. These two look nothing alike, yet they obey exactly the same equation. The mass, damper, and spring give m x'' + c x' + k x = 0, and the coil, resistor, and capacitor give L q'' + R q' + qC = 0: only the letters differ, and the structure is identical. This is a 2nd-order linear differential equation. In the last lesson, a 1st-order equation flowed along a direction field. In 2nd order, acceleration enters the picture and oscillation becomes possible. Whether that oscillation swings and stops or glides smoothly back is decided by two numbers called the characteristic roots. And when those roots are complex, the e(σ+iω)t damped spiral from the complex numbers thread pops out as the answer. In this lesson you move the roots yourself and watch how a single root position decides the fate of a vibration.
First, check that a spring and a circuit really are the same equation. The top of the figure shows a mass held against a wall by a spring and a damper; the bottom shows an RLC circuit made of a coil, a resistor, and a capacitor. Pick the pairs one at a time with the buttons below, and the two matching parts light up together, top and bottom. Mass m pairs with coil L. Both are inertia that resists change: the mass tries to keep its velocity and the coil its current. Damper c pairs with resistor R. Both bleed energy away as heat and slow the motion. Spring k pairs with the capacitor's 1C, and both are restoring forces that push back as far as they were pushed. So the two equations, m x'' + c x' + k x = 0 and L q'' + R q' + qC = 0, are twins with the letters swapped. Solve one and you have solved the other. Mechanical and electrical engineering share the same mathematics not by chance but because nature repeats the same structure.
How do we solve this equation? The key idea is to guess that the answer has the form est. The exponential is its own derivative, so recalling the last lesson's y'=ky, substituting x=est turns each derivative into simply multiplying by s. Then m x'' + c x' + k x = 0 shrinks to the tidy quadratic s² + 2ζs + 1 = 0 (normalized for convenience). The two roots of this quadratic are the characteristic roots, and they hold the whole character of the answer. Plot the roots on the complex plane, which electrical engineers call the s-plane: when underdamped, the two roots sit as a conjugate pair on the dashed circle of radius ω₀. The real part −ζ is the decay, how far left the roots have moved, and the imaginary part ±√(1−ζ²) is the frequency, how far apart they are vertically. Press the different ζ buttons and the roots slide along the circle. Solving has turned into finding these two points.
What happens when the characteristic roots are complex? The answer is exactly the damped spiral from the complex numbers thread. If the roots are s = σ ± iωd, the answer is eσt multiplied by the oscillation cos(ωd t). So the time response swings like a cosine while its amplitude shrinks along the eσt envelope. It is exactly the shape of a bell that rings and fades after one strike. The real part σ is the decay rate, how fast it dies out, and the imaginary part ωd is the frequency, how fast it swings. Use the ζ buttons to move the roots close to the imaginary axis and the damping weakens, so it rings for a long time. Conversely, the farther the roots are from the imaginary axis, the faster it fades. Look at the earlier spiral from the side and you see exactly this swinging curve. One pair of complex characteristic roots is one damped oscillation.
What happens as the damping grows? At some point the oscillation disappears entirely. When the damping is strong enough (ζ ≥ 1), the characteristic roots turn from complex into real. On the s-plane, the conjugate pair that sat on the circle meets on the negative real axis, and both become real roots. With no complex part, there is no cos oscillation. The answer is a monotonic curve that dies away exponentially, sliding back into place without a single swing. Compare the two cases with the two buttons: critical damping (ζ = 1) is the exact boundary state that returns fastest without oscillating. Overdamping (ζ > 1) is so stiff that it creeps back slowly. Car shock absorbers and the closers that keep doors from slamming are deliberately designed for this region, the side that does not oscillate. A system that must not shake keeps its characteristic roots on the real axis.
Now bring everything into one scene. The top is the s-plane where the characteristic roots live; the bottom is the time response those roots produce. Raise ζ slowly from 0 on the track below: at first the two roots sit as a conjugate pair on the circle near the imaginary axis, and the response rings for a long time. As ζ grows, the roots slide along the circle toward the real axis and the ringing dies out faster and faster. At ζ = 1 the two roots meet exactly on the real axis, and that moment is critical damping, the boundary where oscillation vanishes. The tick on the track marks that spot. Raise it further and the roots split apart along the real axis while the response becomes a slow, monotonic return. The two points on top and the curve below move together as one. That is everything about a 2nd-order system. A single root position decides whether it oscillates, how fast it dies out, and how fast it swings. Control engineering is, in the end, the craft of moving these characteristic roots where you want them.