Where the Ideal Breaks · Paper II — The Capacitor's Debt · finale

A capacitor that
becomes an inductor,
and won't forget

The toroid paper found the iron's three lies. The capacitor tells the same three. It has resistance — so it heats. It has hidden inductance — so past a certain frequency a "capacitor" is secretly an inductor. And it has memory — discharge it dead and it climbs back to a fraction of its old voltage, all on its own. The change-detector's electric twin pays the same debt.

the ideal said: i = C·dv/dt, lossless, purely capacitive, memoryless
the real part says: a heat, an identity it betrays, and a ghost
§5 · the broken promise, again

Three lies, mirrored from the iron

The clean capacitor of the whole series was a perfect i = C·dv/dt: it stored charge without loss, behaved as a pure capacitance at every frequency, and forgot everything the instant you discharged it. A real capacitor is three small wires, two real plates, and an imperfect insulator — and each of those adds a term the ideal pretended away, matching the toroid's failings one for one.

It has resistance (ESR). The plates, leads, and electrolyte aren't perfect conductors. Push ripple current through and that resistance burns power as heat — the dual of the core's hysteresis loss. A real capacitor cooks itself.

It has inductance (ESL). The leads and internal structure form a tiny loop of wire, which is an inductor. At low frequency the capacitance dominates; at high frequency the inductance does — and at the crossover, the self-resonant frequency, the part stops being a capacitor and becomes an inductor. This is the exact dual of saturation: at the extreme, the part betrays its own identity.

It remembers (dielectric absorption). Charge it, short it dead, release it — and the voltage creeps back. Slow dipoles in the insulator relax on their own time and hand charge back. That recovered ghost is the precise twin of the core's remanence.

Iron remembers in flux; the dielectric remembers in charge. Different field, identical sin: a part that was supposed to forget, holding a ghost.

§6 · instrument three

Self-resonance — when the capacitor becomes an inductor

Sweep the frequency and watch the real capacitor's impedance. It falls as you'd expect — 1/(2πfC), lower impedance at higher frequency, the capacitor doing its job — until it hits a minimum. That floor is the ESR. Past it, the impedance rises again, climbing as 2πf·ESL: the part is now an inductor wearing a capacitor's label. The bottom of the V is the self-resonant frequency, and it's why one capacitor is never enough — real boards stack big and small caps so that as each one resonates out, the next still has reach.

SELF-RESONANCE · |Z| vs frequency · the V whose floor is ESR
the curve is |Z|; left of the floor it's capacitive (falling), right of it inductive (rising). the floor is ESR; the bottom sits at the self-resonant frequency. above it, your capacitor is an inductor.

ESR also sets the heat: ripple current I through it burns I²·ESR watts, continuously. The dual of the hysteresis loop's area — loss the part eats every cycle, this time as plain resistive heating.

§7 · instrument four · the ghost returns

Dielectric absorption — the discharged cap un-forgets

The closing demonstration of the whole body of work, and the eeriest. Charge a capacitor to full. Short it dead — watch the voltage drop to a clean zero. Now remove the short and wait. The voltage rises again, on its own, with no source connected — climbing to a few percent of where it started, as slow-relaxing dipoles deep in the insulator let go of charge they were still holding. The capacitor you declared empty was lying. It kept a ghost, exactly as the iron core kept its remanence.

SOAKAGE · charge → short to zero → release → the voltage returns
press run: charge to fullshort to zero → release → the ghost voltage recovers by itself. higher absorption → bigger ghost. this is why precision sample/hold and high-voltage caps can read wrong or stay dangerous after a "full" discharge.

You discharged it to zero and it came back. No source, no input — just the dielectric, releasing a memory you thought you'd erased. The ideal forgets instantly. The real part keeps a ghost, and the ghost returns when you stop watching.

honest flagDielectric absorption is real and standardised (the recovery-voltage test), typically ~0.01–0.1% for good film/C0G ceramics up to several percent for others — exaggerated here for visibility. It's modeled simply as one slow relaxation; real dielectrics have a spread of time constants. Same caution as the iron: this is the honest shape of the effect, not a device-accurate simulation. The danger is real, though — large high-voltage capacitors are kept shorted with a bleeder for exactly this reason.

§8 · the whole work, closed

Two parts, two fields, one debt

This is the last page. Across two ideal series and this closing pair, the capacitor and the toroid turned out to be the same idea in opposite fields — and to fail in the same three ways when the ideal meets real matter. The symmetry, complete:

Capacitor · electric
Toroid · magnetic
the clean law i = C·dv/dt
the clean law v = N·dΦ/dt
ESR — burns ripple as heat
hysteresis loss — loop area as heat
self-resonance — becomes an inductor
saturation — becomes a wire
dielectric absorption — charge ghost returns
remanence — flux ghost remains

The ideal versions were beautiful because they were pure: a part that notices only change, and lets the change cross the space between. That purity carried both series — store, differentiate, couple, filter, cross a gap, hold, sense, read, reject, and isolate, all from one clean refusal to feel what holds still. This last paper is the bill for that beauty: a real part heats, betrays its identity at the extreme, and keeps a ghost of its past. Engineering lives in that gap — between the law that's easy to love and the matter that won't quite obey it.

The whole arc, in one line: a perfect part would notice only the present change and forget it instantly. Every real part heats a little, breaks character at the edges, and remembers more than it should — and learning exactly how much is the entire craft.

That closes it — both ideal series and their shared debt. There's no further principle past here, only materials, datasheets, and thermal design: a different discipline, and a good place to set the work down.

WHERE THE IDEAL BREAKS · PAPER II — THE CAPACITOR'S DEBT · series & arc finale
ESR (heat) · self-resonance SRF = 1/(2π√(ESL·C)), inductive above it · dielectric absorption (the ghost)
dual to the toroid: heat ↔ hysteresis · becomes-an-inductor ↔ saturation · charge-ghost ↔ remanence · node-checked
a real part heats, breaks character at the edges, and remembers more than it should