Skip to content

Parasitic capacitance and self-resonant frequency (SRF)

Every physical inductor or transformer winding contains unintended capacitive elements distributed across its conductor geometry. These parasitic capacitances β€” formed between adjacent turns, between winding layers, and between the winding and its core β€” establish a resonant circuit with the component's own inductance. The frequency at which inductive and capacitive reactances balance is the Self-Resonant Frequency (SRF): the hard operational ceiling beyond which the component ceases to behave inductively.

Definition

The Self-Resonant Frequency ($f_{\text{SRF}}$) is the frequency at which a wound electromagnetic component's inductive reactance ($X_L$) equals its effective parasitic capacitive reactance ($X_C$) in magnitude. At this frequency the net reactive impedance disappears and the component presents a purely resistive impedance. Above $f_{\text{SRF}}$ the component becomes net-capacitive.

Three distinct capacitive mechanisms combine into the effective parasitic capacitance $C_p$:

  • Inter-turn capacitance ($C_{\text{turn}}$): Between adjacent turns within the same layer, separated only by the enamel insulation wall. For 0.10 mm wire with Grade 2 insulation the dielectric gap is approximately 0.012 mm β€” generating significant capacitance over a high turn count.
  • Inter-layer capacitance ($C_{\text{layer}}$): Between successive winding layers. Typically the dominant contributor in multi-layer coils.
  • Winding-to-core capacitance ($C_{\text{core}}$): Present whenever wire is deposited directly onto a semi-conductive core. Negligible at low frequencies for ferrite cores ($\rho \approx 10^4\text{–}10^{10}\,\Omega\cdot\text{cm}$) but significant above 1 MHz.

Key properties

The governing equation derives from setting $X_L = X_C$ across the parallel LC tank formed by $L$ and $C_p$:

$$f_{\text{SRF}} = \frac{1}{2\pi\sqrt{L \cdot C_p}}$$
Parameter Relationship to SRF Engineering implication
Inductance ($L$) $f_{\text{SRF}} \propto 1/\sqrt{L}$ Doubling $L$ reduces SRF by 29%
Parasitic capacitance ($C_p$) $f_{\text{SRF}} \propto 1/\sqrt{C_p}$ Halving $C_p$ raises SRF by 41%
Layer count More layers β†’ higher $C_p$ Each additional layer reduces SRF nonlinearly
Winding topology Single-layer, bank, or pi winding β†’ lower $C_p$ Topology is the primary SRF lever available to the designer
Insulation permittivity ($\varepsilon_r$) Higher $\varepsilon_r$ β†’ higher $C_p$ PTFE ($\varepsilon_r \approx 2.1$) vs polyurethane ($\varepsilon_r \approx 3.5$) yields ~30% SRF gain

Typical $C_p$ values by winding topology, for a representative 10 Β΅H coil wound with 0.10 mm wire:

Winding topology Typical $C_p$ Estimated SRF (10 Β΅H)
Single-layer solenoid 0.5–2 pF 35–70 MHz
Multi-layer orthocyclic winding 2–15 pF 13–35 MHz
Multi-layer wild winding 5–30 pF 8–22 MHz
Bank (sectional) winding 1–5 pF 22–50 MHz
Toroidal single-layer 0.3–1.5 pF 41–92 MHz

Physics and governing equation

The impedance of a real inductor passes through three distinct regions as frequency increases:

Frequency region Dominant behaviour Impedance characteristic
$f \ll f_{\text{SRF}}$ Inductive $|Z| = 2\pi f L$ β€” rising linearly. Full nominal inductance available.
$f \approx 0.5 \cdot f_{\text{SRF}}$ Inductive, $C_p$ influence rising Apparent inductance begins to inflate above nominal $L$. Circuit models become unreliable.
$f = f_{\text{SRF}}$ Purely resistive $|Z|$ reaches maximum. Component behaves as high-impedance RF choke only.
$f > f_{\text{SRF}}$ Capacitive $|Z|$ falls with increasing frequency. Component passes high-frequency signals.

The Quality factor ($Q$) degrades as operating frequency approaches $f_{\text{SRF}}$ according to:

$$Q(f) = \frac{2\pi f L}{R_s} \cdot \left(1 - \left(\frac{f}{f_{\text{SRF}}}\right)^2\right)$$

At $f = 0.7 \cdot f_{\text{SRF}}$ the correction factor reduces to 0.51 β€” Q is already halved even though the component still appears nominally inductive. For applications requiring $Q > 50$, limit operation to below $0.3 \cdot f_{\text{SRF}}$. For $Q > 100$, limit to below $0.2 \cdot f_{\text{SRF}}$.

When to use

SRF specification is critical whenever the operating frequency exceeds approximately 10% of the component's nominal SRF, or wherever the following conditions apply:

  • RF and high-frequency inductors: Any inductor operating above 1 MHz where apparent inductance inflation and Q degradation would corrupt filter or matching network performance.
  • Gate drive and pulse transformers: Fast switching transients contain harmonics at multiples of the fundamental frequency. A 500 kHz switching converter generates harmonics at 1.5 MHz, 2.5 MHz and beyond β€” each must remain below $f_{\text{SRF}}$.
  • Wireless power transfer coils: Resonant WPT systems operating at 6.78 MHz or 13.56 MHz require SRF well above the operating frequency to maintain the designed resonant network behaviour.
  • Precision sensor coils: Coils in impedance-based sensing systems where capacitive parasitics would corrupt the measurement transfer function.

Limitations

  • SRF versus inductance trade-off: Increasing $L$ by adding winding layers always reduces SRF. The two specifications cannot be simultaneously maximized for a given wire gauge and winding geometry.
  • Measurement dependency: SRF is sensitive to test fixture parasitics. A poorly calibrated fixture can shift the apparent SRF by 5–15%, producing incorrect component qualification decisions. Per IEC 62024-1, open and short calibration at the component reference plane is mandatory.
  • Temperature sensitivity: $C_p$ increases with temperature as insulation permittivity rises. An inductor meeting SRF specifications at 25Β°C may fail at 125Β°C operating temperature, particularly for polyurethane-insulated windings.
  • Winding variation: $C_p$ is sensitive to small variations in conductor placement during manufacturing. SRF can vary Β±10–20% between production batches for the same nominal winding specification unless tight winding tension and pitch controls are enforced.

Comparison to alternatives

Where SRF constrains a design, three architectural alternatives exist:

Architecture SRF vs multi-layer wound inductor Inductance per unit volume Typical application
Single-layer solenoid 3–10Γ— higher SRF Low β€” longer winding length required RF inductors above 10 MHz
Bank (sectional) winding 2–5Γ— higher SRF Moderate HF transformers, broadband inductors
Multilayer with PTFE insulation 1.3–1.8Γ— higher SRF Moderate β€” comparable to standard winding Precision RF coils where winding geometry is fixed
Chip inductor (SMD, ferrite) Highest SRF (500 MHz – 10 GHz typical) Low per unit β€” standard values only High-frequency PCB decoupling, RF matching

The key difference between a custom wound inductor and an SMD chip inductor is design freedom: a custom winding can be optimised for a specific $L$, $Q$, current rating, and SRF simultaneously, whereas chip inductors offer fixed standard values with no winding topology control available to the system designer.

Design failure modes

  • Apparent inductance inflation: Below $f_{\text{SRF}}$, $C_p$ causes the measured inductance to appear higher than the DC value. Circuits designed to nominal $L$ will exhibit resonant frequency errors if this effect is not modelled. Prevention: Measure inductance at operating frequency, not at 1 kHz.
  • Q collapse near SRF: Operating a resonant circuit at $f > 0.3 \cdot f_{\text{SRF}}$ causes Q to fall below specification, broadening filter bandwidths and reducing WPT efficiency. Prevention: Validate Q at operating frequency during design qualification.
  • Capacitive bypass at high frequency: Above $f_{\text{SRF}}$, the inductor passes high-frequency interference rather than blocking it β€” the opposite of its intended EMI filter function. Prevention: Specify $f_{\text{SRF}} \geq 3 \times$ the highest frequency to be attenuated.
  • Batch SRF spread: If winding tension or traverse pitch varies between production batches, $C_p$ shifts and SRF moves outside the qualified range. Prevention: Include SRF as an incoming inspection parameter with acceptance limits of Β±15% of nominal.

For definitions of SRF, Q factor, skin effect, and proximity effect referenced in this article, see the mycoil.info Engineering Glossary.