Non-Ideal Components: Parasitic R, L, C Effects

Every resistor, capacitor, and inductor on a schematic is an idealization. The moment you place a real part on a PCB, it acquires an ESR, an ESL, a lead inductance, or a parallel leakage path that the datasheet's nominal value doesn't capture. At DC and low frequency these parasitics are negligible; above a few MHz — or during fast switching edges regardless of frequency — they often dominate circuit behavior. Understanding which parasitic matters where is the difference between a design that works on the bench and one that only works in simulation with ideal parts.

Why Ideal Components Fail

An ideal resistor has impedance Z = R at all frequencies. An ideal capacitor has Z = 1/(jωC), monotonically decreasing to zero. An ideal inductor has Z = jωL, monotonically increasing without bound. None of this holds for real parts because every physical component has:

• Series resistance in leads, plates, or windings (ESR)
• Series inductance from lead length and current loop geometry (ESL)
• Parasitic capacitance between terminals, turns, or to the case/ground plane

The result is that every real component is better modeled as a small RLC network, and that network has its own resonance — a frequency where the "wrong" element takes over.

Real Capacitor Model

        ESL        ESR
  o----[L]-------[R]-------[C]----o
                (ideal capacitance)
        (parallel leakage Rp across the whole thing, usually >> ESR, often ignored above audio)

Impedance magnitude:

|Z(ω)| ≈ √[ ESR² + (ωL − 1/(ωC))² ]

This is a series RLC branch. Below the self-resonant frequency (SRF), the 1/(ωC) term dominates and the part behaves capacitively. Above SRF, ωL dominates and the part behaves inductively. At SRF the two reactances cancel and the impedance drops to a minimum equal to ESR.

Self-resonant frequency:

f_SRF = 1 / (2π·√(L·C))

Worked example: A 100 nF ceramic capacitor with typical package inductance ESL ≈ 1 nH and ESR ≈ 20 mΩ.

f_SRF = 1 / (2π·√(1×10⁻⁹ × 100×10⁻⁹))
      = 1 / (2π·√(1×10⁻¹⁶))
      = 1 / (2π × 1×10⁻⁸)
      ≈ 15.9 MHz

Below 15.9 MHz this part is a good capacitor. Above it, it looks like a small inductor — which is exactly why decoupling networks use multiple capacitor values in parallel (e.g., 10 µF + 100 nF + 10 pF): each covers the frequency band where the others have already gone inductive.

Check: at f_SRF, ωL = 2π(15.9×10⁶)(1×10⁻⁹) ≈ 0.10 Ω and 1/(ωC) = 1/(2π(15.9×10⁶)(100×10⁻⁹)) ≈ 0.10 Ω. They match, confirming resonance, and the net impedance there is just ESR ≈ 20 mΩ — the minimum achievable impedance for that part, not zero.

Real Inductor Model

        ESR                    Cp (winding-to-winding,
  o----[R]-------[L]---o        turn-to-turn capacitance)
                        |
                        +---[Cp]---o (parallel across the ideal L)

Here the parasitic capacitance Cp between windings is in parallel with the ideal inductance, so:

f_SRF = 1 / (2π·√(L·Cp))

Below SRF the part is inductive; above SRF, Cp dominates and the part becomes capacitive — the opposite failure direction from a capacitor. This is why real inductors used as RF chokes or ferrite beads have a usable frequency range with an upper limit, and why "just add a bigger inductor" for filtering can backfire at high frequency: a larger L usually means more turns, more Cp, and a lower SRF.

Real Resistor Model

At DC, a resistor is just R. At high frequency, three effects appear:

• Lead/trace inductance, typically 1–5 nH for a through-hole or 0805-style part with modest lead length — negligible below a few MHz, significant for fast transient current paths (e.g., gate drive resistors, snubbers).
• Parasitic parallel capacitance, from the resistive element's geometry and end-cap terminations — typically fractions of a pF, matters for high-value resistors (>100 kΩ) at RF, where it can shunt the resistance and cause the impedance to fall instead of stay flat.
• Skin effect and proximity effect in wirewound resistors, raising effective R at high frequency (a separate topic on its own).

For most digital and analog signal-path resistors below ~10 MHz, these are second-order. For pulse and RF work, they are first-order.

Comparison Table

Design Implications

• PCB decoupling: place low-ESL capacitors (small case size, short vias) closest to the IC pin; use multiple values in parallel to extend the effective low-impedance band past any single part's SRF.
• High-current switching loops: trace/via inductance in the power loop (not just the capacitor's own ESL) often dominates ringing during MOSFET switching — minimizing loop area matters as much as component choice.
• Filter design accuracy: a simulated LC filter's real cutoff and stopband attenuation will deviate from ideal-component SPICE results once ESR (reduces Q, raises insertion loss) and self-resonance (limits stopband attenuation at high frequency — the "coming back down" of a real low-pass filter's attenuation curve) are included.
• Gate drive and RF matching: parasitic lead inductance in a gate resistor or matching network shifts the intended phase/timing enough to cause overshoot or detuning; always check vendor-supplied SPICE models with ESL/ESR included, not the ideal part.
• Sensing and measurement: current-sense resistors show inductive kick on fast edges — measured V = IR + L·dI/dt, which corrupts fast current measurements unless the sense resistor is a low-inductance (e.g., 4-terminal Kelvin) type.

Key Takeaways

• Every real component is a small RLC network; the "ideal" behavior only holds below its self-resonant frequency.
• Capacitors turn inductive above SRF (ESR + ESL in series with C); inductors turn capacitive above SRF (Cp in parallel with L) — opposite failure modes.
• f_SRF = 1/(2π√(LC)) locates the corner where the parasitic element takes over from the intended one.
• Resistors are usually the most "ideal" component at moderate frequency, but lead inductance and stray capacitance matter for RF and fast pulse applications.
• Practical fixes — multiple decoupling values, minimized loop area, low-ESL/ESR package selection, Kelvin sensing — all follow directly from identifying which parasitic dominates in the frequency range of interest.