Transmission Lines

Coupling Gap

Q: How does coupling gap width affect the coupling coefficient?

The coupling between two conductors is dominated by the fringing-field series capacitance across the gap, which falls off roughly as the natural log of the spacing. Narrowing the gap increases the gap capacitance and therefore the coupling level (less negative S21), while widening it reduces coupling. For an end-coupled microstrip on a 0.508 mm alumina substrate at 10 GHz, a 25 micron gap yields tight coupling around minus 3 to minus 6 dB, while a 200 micron gap drops the coupling toward minus 20 dB. Because the relationship is logarithmic and dominated by edge fields, tight couplings demand very small, tolerance-sensitive gaps that are often limited by the photolithography or milling process resolution.

Q: Why are tight coupling gaps difficult to fabricate accurately?

Coupling is set by the fringing capacitance across the gap, so a fixed absolute etching error becomes a large fractional error when the gap is small. A 10 micron over-etch on a 25 micron target gap is a 40 percent change, which can shift the coupling level by 2 to 3 dB and detune a filter passband. Standard photolithography on thin-film alumina holds roughly plus or minus 5 micron, but PCB etching on soft substrates typically holds only plus or minus 25 to 50 micron, making sub-100 micron gaps unreliable. Designers often switch to broadside or edge-coupled overlap geometries, or add a series interdigital capacitor, when the required gap drops below the achievable process tolerance.

Q: What equivalent-circuit model represents a coupling gap?

A symmetric coupling gap in a single line is modeled as a pi network of three capacitors: a series gap capacitance Cg that carries the coupled energy across the gap, plus two shunt fringing capacitances Cp from each conductor edge to ground. Cg controls the through coupling and Cp accounts for the open-end fringing field of each line. For an edge-coupled pair, the gap instead sets the difference between the even-mode and odd-mode capacitances, which in turn defines the even-mode and odd-mode impedances used in coupled-line filter and directional-coupler synthesis. Full-wave electromagnetic solvers extract these capacitances directly, and they are typically in the femtofarad range at microwave frequencies.

/KUHP-ling gap/

Spacing between two conductor edges or ends across which RF energy crosses by way of a fringing-field series capacitance rather than a direct metal connection. In end-coupled and edge-coupled microstrip, the gap dimension fixes the gap capacitance and therefore the coupling coefficient, the achievable bandwidth, and the through-path S21 level. Because the controlling fringing field is concentrated right at the conductor edges, coupling varies roughly with the logarithm of the gap width, so a small absolute change in spacing produces a large change in coupling. Tight couplings near minus 3 dB call for gaps on the order of 10 to 30 microns, while loose couplings of minus 20 dB or weaker use gaps of several hundred microns. The gap also separates the even-mode and odd-mode capacitances that define coupled-line filter and directional-coupler behavior.

Category: Transmission Lines

Typical gap: 10 µm to 500 µm

Coupling range: −3 dB to −25 dB

How the Gap Sets Coupling Between Lines

When two transmission-line conductors run close together or sit end to end with a small break, the electromagnetic field bridging the break stores energy that behaves electrically like a series capacitor. This gap capacitance is the mechanism by which a signal couples from one conductor to the next without a galvanic connection. The narrower the gap, the higher the capacitance and the more energy transfers, which is why the gap dimension is the primary design knob for setting coupling strength in end-coupled filters, gap-coupled resonators, and many directional couplers. Because the energy is carried almost entirely by the fringing field at the facing edges, the coupling depends only weakly on the absolute conductor length and very strongly on the spacing.

The relationship is not linear. The series gap capacitance scales approximately with the natural logarithm of the inverse gap width for thin-film geometries, so halving a gap does not double the coupling; it changes it by a roughly fixed number of decibels. In practical terms, doubling the gap typically reduces coupling by about 4 to 6 dB across the microwave band. This logarithmic behavior is convenient for trimming a design over a modest range, but it makes very tight couplings extremely sensitive to fabrication error, since the gap must then shrink into the few-tens-of-microns regime where etching and milling tolerances dominate.

For a symmetric edge-coupled pair, the gap is what separates the even-mode and odd-mode field distributions. A narrow gap raises the odd-mode capacitance sharply while leaving the even mode largely unchanged, increasing the spread between the odd-mode impedance and the even-mode impedance. That impedance spread is the quantity coupled-line synthesis equations actually solve for, so the gap is the geometric variable that ties an abstract coupling specification to a physical layout dimension.

Gap Capacitance and Coupling Equations

Series gap (pi) equivalent circuit:
Two open ends across a gap → series C_g plus two shunt C_p

Voltage coupling coefficient:
k = (Z_0e − Z_0o) / (Z_0e + Z_0o)  →  Coupling(dB) = 20·log₁₀(k)

Series-gap coupling magnitude:
|S₂₁| ≈ 2πf × C_g × Z₀  (weak-coupling, C_g in F)

Gap-width dependence (thin-film approximation):
C_g ∝ ln(1 / s)  where s = gap width

Where Z_0e = even-mode impedance, Z_0o = odd-mode impedance, C_g = series gap capacitance, C_p = shunt fringing capacitance. Example: at 10 GHz a 12 fF gap on a 50 Ω line gives |S₂₁| ≈ 0.038, or about −28 dB.

Coupling Gap vs. Coupling Level on Common Substrates

Gap width	Substrate (10 GHz)	Approx. C_g	Coupling level	Fab method	Typical use
10 µm	0.254 mm alumina	~60 fF	−3 to −5 dB	Thin-film, photolith	Tight 3 dB couplers
25 µm	0.254 mm alumina	~30 fF	−6 to −9 dB	Thin-film, photolith	End-coupled filters
100 µm	0.508 mm Rogers 5880	~12 fF	−14 to −18 dB	PCB etch	Loose monitor taps
250 µm	0.508 mm Rogers 5880	~5 fF	−20 to −25 dB	PCB etch / mill	Sampling / detectors
0.5 mm	0.787 mm FR-4	~3 fF	−26 dB or weaker	Standard PCB	Stray coupling budget

Common Questions

Frequently Asked Questions

How does coupling gap width affect the coupling coefficient?

Coupling is dominated by the fringing-field series capacitance across the gap, which falls off roughly as the natural log of the spacing. Narrowing the gap raises the gap capacitance and the coupling (less negative S21). On 0.508 mm alumina at 10 GHz, a 25 µm gap gives tight coupling near −3 to −6 dB, while a 200 µm gap drops toward −20 dB. The logarithmic, edge-dominated behavior means tight couplings need very small, tolerance-sensitive gaps.

Why are tight coupling gaps difficult to fabricate accurately?

A fixed absolute etch error becomes a large fractional error when the gap is small. A 10 µm over-etch on a 25 µm gap is a 40% change and can shift coupling 2 to 3 dB and detune a passband. Thin-film alumina holds about ±5 µm, but PCB etch on soft substrates holds only ±25 to 50 µm, so sub-100 µm gaps are unreliable. Designers then switch to broadside-coupled overlap or add an interdigital capacitor.

What equivalent-circuit model represents a coupling gap?

A symmetric series gap is a pi network: a series C_g carrying the coupled energy plus two shunt fringing C_p from each edge to ground. For an edge-coupled pair the gap instead sets the difference between even-mode and odd-mode capacitances, which define Z_0e and Z_0o used in coupled-line synthesis. Full-wave solvers extract these directly; the capacitances are typically in the femtofarad range at microwave frequencies.

Related Terms

← Coupling (Fundamental) Coupling Length →