Coupling Path
The Source-Path-Receiver Model
Every electromagnetic interference event requires three ingredients at once: an energy source, a victim (receiver) that is susceptible at the offending frequency, and a coupling path that links them. Remove or sufficiently weaken any one of the three and the interference stops. In practice the source is often a switching regulator, a clock, or an intentional transmitter that cannot be changed without redesign, and the victim is frequently a circuit that has already passed functional qualification. That leaves the coupling path as the lever engineers reach for first, because it is usually an accessible external feature: a cable, a connector, a slot in an enclosure, a shared ground return, or simply the spacing between two traces.
The four canonical paths split by which field carries the energy. Conductive coupling shares a physical conductor or a common return impedance, so a noise current modulates the voltage that the victim sees. Capacitive coupling is driven by voltage and the stray capacitance between conductors; the coupled current grows with frequency and with the rate of voltage change. Inductive coupling is driven by current loops and mutual inductance, with the coupled voltage growing with the rate of current change. Radiated coupling involves a genuine propagating wave once the source dimensions become comparable to a wavelength, and it follows free-space path loss rather than a simple near-field gradient. Near-field capacitive and inductive coupling fall off rapidly with distance, which is why a few millimeters of separation can change a result.
A useful diagnostic distinction is impedance. Capacitive paths matter most when the victim node is high impedance, since a small displacement current develops a large voltage there; the cure is an electrostatic shield, more spacing, or a lower node impedance. Inductive paths matter most when the victim circuit is low impedance and carries a current loop; the cure is reducing loop area, twisting conductors, or adding a high-permeability shield. Conductive paths are attacked with filtering and separated returns, and radiated paths with a shielded, gasketed enclosure whose seams and apertures are kept far below a quarter wavelength.
Quantifying a Coupling Path
Vvictim = Vsource × H(f) where H(f) = path transfer function
Capacitive (electric-field) coupling:
Ic ≈ Cm × dV/dt → Vnoise ≈ ω × Cm × Rvictim × Vsource
Inductive (magnetic-field) coupling:
Vnoise = M × dI/dt = ω × M × Isource, M ∝ loop area
Noise margin at the victim:
NM = Vthreshold − Vcoupled (NM > 0 immune, NM < 0 susceptible)
Where Cm = mutual capacitance, M = mutual inductance, ω = 2πf, Rvictim = victim node impedance. Example: Cm = 1 pF, Rvictim = 10 kΩ, dV/dt of a 3.3 V edge in 1 ns at f ≈ 100 MHz couples on the order of tens of mV.
Comparison of EMI Coupling Paths
| Path type | Carrier | Worst when | Primary control | Dominant range |
|---|---|---|---|---|
| Conductive | Shared wire / return Z | Common return impedance | Filter, separate returns, feed-through cap | DC to ~30 MHz |
| Capacitive | Electric field (dV/dt) | High-impedance victim node | Spacing, electrostatic shield, lower Z | ~1 MHz and up |
| Inductive | Magnetic field (dI/dt) | Low-impedance current loop | Twist, reduce loop area, μ-metal shield | ~10 kHz to MHz |
| Radiated | Propagating wave | Aperture ≈ λ/4 | Gasketed enclosure, cable shields | > 30 MHz |
| Common mode | Cable vs. chassis | Cable ≈ λ/4 antenna | Common-mode choke, ferrite, bonding | Broadband |
Frequently Asked Questions
How do I identify which coupling path is causing an EMI failure?
Change one variable at a time and watch the response. A clamp-on ferrite on a suspect cable that drops the emission confirms a conductive or common-mode path on that cable. Interference that falls as 1/r in the near field or 1/r² for radiation points to a field-based path. A temporary copper-tape shield with a single ground bond that yields a large drop indicates radiated or capacitive coupling. Compare the problem frequency to cable lengths too, since a cable near λ/4 is an efficient radiator. A near-field probe on a spectrum analyzer localizes the dominant path before you commit to a fix.
Why is attenuating the coupling path usually cheaper than fixing the source or victim?
Fixing the source often means redesigning a converter, slowing edge rates, or swapping an IC, which ripples through the product. Hardening the victim means adding immunity to an already-qualified circuit. The coupling path is usually an external feature you can attack with a low-cost part: a feed-through capacitor, a common-mode choke, a gasketed seam, twisted-pair wiring, or a few millimeters of separation. A sub-dollar ferrite or a shielded cable often resolves an issue that would otherwise force a board respin, so EMC practice attacks the path first.
What is the difference between capacitive and inductive coupling paths?
Capacitive (electric-field) coupling is driven by voltage and stray capacitance; the coupled current rises with frequency and dV/dt, so it dominates at high-impedance, high-voltage nodes and is cured with spacing, an electrostatic shield, or lower node impedance. Inductive (magnetic-field) coupling is driven by current loops and mutual inductance, with coupled voltage rising with dI/dt, so it dominates in low-impedance, high-current circuits and is cured by reducing loop area, twisting wires, and using a shorted or high-permeability shield. Capacitive coupling worsens at high victim impedance; inductive coupling worsens at low victim impedance.