RF Design

Coupled Resonator

Q: How is the coupling coefficient between two resonators extracted from measurement?

When two synchronously tuned resonators are weakly probed, the single resonance splits into two peaks at frequencies f1 (lower) and f2 (higher). The coupling coefficient is k = (f2 squared minus f1 squared) divided by (f2 squared plus f1 squared), which simplifies to approximately k = (f2 minus f1) divided by f0 for narrow splitting. In a 3D EM simulator such as HFSS or CST, you place two resonators with weak excitation ports, sweep frequency, identify the two split-mode peaks, and apply the formula. For example, if f1 = 1.98 GHz and f2 = 2.02 GHz, then k is approximately 0.020, or 2 percent. This measured k is then matched to the normalized coupling matrix value M times fractional bandwidth derived from the filter prototype, allowing the physical spacing or iris dimension to be tuned until the target coupling is reached.

Q: What is the difference between coupling coefficient and external Q in a coupled-resonator filter?

The coupling coefficient k describes energy exchange between two adjacent resonators inside the filter and sets the inter-resonator bandwidth, while the external Q (Qe) describes how strongly the first and last resonators are loaded by the input and output ports. Both scale with the fractional bandwidth FBW. The relationships are k(i,i+1) = FBW divided by the square root of (g_i times g_(i+1)) and Qe = g0 times g1 divided by FBW, where g values are the lowpass prototype element values. For a 4-pole 0.5 dB Chebyshev filter with 2 percent FBW (prototype g values g1 = 1.6703, g2 = 1.1926), the input external Q is roughly 83.5 and the first inter-resonator coupling is about 0.0142. Getting k right shapes the passband ripple and return loss; getting Qe right sets the input match and overall insertion loss.

Q: Why do two coupled resonators produce two resonant peaks instead of one?

Two identical resonators each resonate at f0 in isolation, but coupling them allows energy to slosh back and forth, creating two normal modes: an even mode where the fields add in phase and an odd mode where they oppose. These modes resonate at slightly different frequencies, fe and fo, separated by an amount proportional to the coupling coefficient. The lower split frequency corresponds to one mode and the upper to the other, with the original f0 sitting near the geometric mean. This mode splitting is the physical basis of the bandpass response: an N-resonator filter has N coupled modes that together define N poles. The stronger the coupling, the wider the split and the wider the resulting passband.

/KUP-uhld REZ-uh-nay-ter/

Two or more resonant circuits linked by mutual inductive or capacitive coupling form a coupled-resonator structure, the building block of nearly every microwave bandpass filter. When the resonators are brought close enough to exchange energy, the single resonance at f₀ splits into separate normal modes whose spacing is governed by the coupling coefficient k. An N-resonator chain produces N poles, and the filter response is synthesized by assigning each inter-resonator coupling and each input/output external Q from the lowpass prototype. Typical narrowband cavity designs use k values of 0.5% to 5% and external Q of 20 to 200, trading bandwidth against selectivity and insertion loss.

Category: RF Design

Coupling k: 0.5% to 5%

External Q: 20 to 200

How Coupling Splits a Resonance Into Filter Poles

A single resonator stores energy at one frequency and, on its own, behaves like a narrow notch or peak with a bandwidth set by its quality factor. The moment a second resonator is placed nearby so the two share electric or magnetic field, energy begins to slosh between them. This shared energy lifts the degeneracy of the single resonance: instead of one peak at f₀, the pair exhibits two normal modes, an even mode and an odd mode, separated in frequency by an amount proportional to the coupling strength. This mode-splitting behavior is the physical mechanism behind the multi-pole bandpass response, and it is why filter engineers think in terms of coupled resonators rather than isolated tank circuits.

Modern filter design treats a coupled-resonator network as a coupling matrix. Each diagonal entry represents a resonator's self-resonant frequency offset, and each off-diagonal entry M_ij represents the normalized coupling between resonators i and j. Cross-couplings between non-adjacent resonators create transmission zeros that sharpen the skirts, enabling elliptic and quasi-elliptic responses. The designer extracts the required couplings from a lowpass prototype, then maps each normalized value onto a physical dimension: iris width in a waveguide cavity, rod spacing in a combline assembly, or gap and via geometry in a planar resonator.

The coupling can be predominantly magnetic (inductive), predominantly electric (capacitive), or mixed, and the sign matters. Magnetic and electric couplings of opposite sign partially cancel, which designers exploit to realize the negative couplings needed for certain cross-coupled topologies. Because field overlap depends sensitively on geometry, the coupling coefficient is almost always extracted from full-wave electromagnetic simulation rather than closed-form equations, then verified on hardware with a vector network analyzer.

Coupling and Mode-Splitting Equations

Coupling coefficient from split modes:
k ≈ (f₂² − f₁²) / (f₂² + f₁²) ≈ (f₂ − f₁) / f₀

Inter-resonator coupling from prototype:
k_i,i+1 = FBW / √(g_i × g_i+1)

Input/output external Q:
Q_e = (g₀ × g₁) / FBW

Where f₁, f₂ = lower and upper split-mode frequencies, f₀ = center frequency, FBW = fractional bandwidth, g_i = lowpass prototype element values. Example: 4-pole 0.5 dB Chebyshev (g₁ = 1.6703, g₂ = 1.1926), FBW = 2% → Q_e ≈ 83.5, k_1,2 ≈ 0.0142.

Coupling Mechanisms Compared

Coupling type	Dominant field	Sign	Physical control	k vs. spacing	Typical use
Inductive (magnetic)	H-field overlap	Positive	Iris near current max	Increases as gap closes	Cavity, combline mainline
Capacitive (electric)	E-field overlap	Negative	Probe near voltage max	Increases as gap closes	Cross-coupling, zeros
Mixed	Both, net difference	Either	Iris position vs. field	Can null at a point	Sign control, dispersion
Aperture (iris)	H or E by slot orientation	Set by slot	Slot width and offset	Monotonic with width	Waveguide filters
External (port)	Tap or loop loading	n/a (Q_e)	Tap height, loop area	Loading sets Q_e	Input/output match

Common Questions

Frequently Asked Questions

How is the coupling coefficient between two resonators extracted from measurement?

Weakly probe two synchronously tuned resonators; the single resonance splits into peaks at f₁ and f₂. Then k ≈ (f₂² − f₁²) / (f₂² + f₁²), which for narrow splitting reduces to (f₂ − f₁) / f₀. With f₁ = 1.98 GHz and f₂ = 2.02 GHz, k is about 0.020 (2%). Match this to M × FBW from the prototype and tune the spacing or iris until the target is met.

What is the difference between coupling coefficient and external Q in a coupled-resonator filter?

The coupling coefficient k governs energy exchange between adjacent resonators and sets the inter-resonator bandwidth, while external Q (Q_e) sets how strongly the end resonators are loaded by the ports. From the prototype, k_i,i+1 = FBW / √(g_ig_i+1) and Q_e = g₀g₁ / FBW. For a 4-pole 0.5 dB Chebyshev at 2% FBW (g₁ = 1.6703, g₂ = 1.1926), Q_e ≈ 83.5 and k_1,2 ≈ 0.0142. k shapes ripple; Q_e sets the input match.

Why do two coupled resonators produce two resonant peaks instead of one?

Each resonator resonates at f₀ in isolation, but coupling lets energy oscillate between them, forming two normal modes: an even mode (fields in phase) and an odd mode (fields opposed). They resonate at fe and fo, split by an amount set by k, with f₀ near their geometric mean. This mode splitting is the basis of the bandpass response; an N-resonator filter has N coupled modes defining N poles, and stronger coupling widens the passband.

Related Terms

← Coupled Port Coupler Design →

Microwave Filters

Need a Coupled-Resonator Filter Designed?

From narrowband cavity preselectors to cross-coupled elliptic duplexers, our engineering team synthesizes and builds coupled-resonator filters for radar, satellite, and millimeter-wave systems. Tell us your specifications.

Get in Touch