Coupling Length
How Coupling Length Sets a Coupler's Response
A coupled-line directional coupler is built from two transmission lines placed close enough that the fields of one line excite the other along their shared run. That shared run is the coupling length. The structure supports two propagation modes: the even mode, in which both lines carry equal in-phase voltages, and the odd mode, in which the voltages are equal but out of phase. Each mode sees its own characteristic impedance (Z0e and Z0o) and, in an inhomogeneous medium such as microstrip, its own phase velocity. The coupled-port voltage results from the interference of these two modes as they propagate over the coupling length, so the electrical length of the coupled section, θ = βL, is the variable that ties physical geometry to electrical performance.
Maximum coupling occurs at θ = 90°, which corresponds to a physical coupling length of one quarter guided wavelength. At this length the differential phase between the even and odd modes is exactly right for the forward components to cancel at the isolated port and reinforce at the coupled port, producing a backward-wave coupler whose coupled output emerges at the port nearest the input. The mid-band coupling factor is fixed by the impedance ratio Z0e/Z0o, while the coupling length sets where in frequency that peak lands. Because the coupling response varies as sin(θ), tight couplers (3 dB to 6 dB) sustain only modest fractional bandwidth before coupling flatness degrades, whereas loose couplers (15 dB to 30 dB) hold their nominal coupling across an octave or more.
Quarter-Wave Design and Length Calculation
To build a coupler at a target center frequency f0, the designer first finds the guided wavelength from the effective dielectric constant, then takes one quarter of it as the physical coupling length. In microstrip the even and odd modes propagate at different velocities, so a single λ/4 cannot satisfy both modes exactly; this velocity mismatch is the root cause of limited directivity in microstrip couplers and is why stripline, with its homogeneous dielectric, achieves far better directivity for the same coupling length. A small open-end length correction is added at each end of the coupled section to account for fringing capacitance.
|S31| = C × sin(θ) / √(1 − C2 × cos2(θ)) where θ = βL
Maximum coupling (mid-band, θ = 90°):
C = (Z0e − Z0o) / (Z0e + Z0o) → Coupling (dB) = −20·log10(C)
Physical coupling length (quarter-wave):
L = λg/4 = c / (4 × f0 × √εeff)
Where C = mid-band voltage coupling factor, Z0e/Z0o = even/odd-mode impedances, θ = electrical length, β = phase constant, εeff = effective permittivity. Example: stripline, εr = 2.2, f0 = 10 GHz → λg ≈ 20.2 mm, L ≈ 5.1 mm.
Coupling Length Across Frequency
| Electrical length θ | Physical length | Frequency vs. f0 | Coupling behavior | Notes |
|---|---|---|---|---|
| 45° | λg/8 | 0.5 f0 | Under-coupled (~3 dB below peak) | Low-frequency band edge |
| 90° | λg/4 | f0 | Maximum coupling | Design center, deepest isolation |
| 135° | 3λg/8 | 1.5 f0 | Coupling falling off | High-frequency band edge |
| 180° | λg/2 | 2 f0 | Coupling null (0 power) | Lines effectively decoupled |
| 270° | 3λg/4 | 3 f0 | Coupling peak repeats | Periodic response, spurious passband |
Frequently Asked Questions
Why is the coupling length set to a quarter wavelength?
At θ = 90° (one quarter guided wavelength) the relative phase between the even and odd modes is exactly right for the backward components to add at the coupled port and cancel at the isolated port, giving maximum coupling and the deepest isolation. A shorter section couples more weakly but more flatly; a longer section over-couples at band center and rolls off, reaching a coupling null at θ = 180°.
How do I calculate the physical coupling length for a coupler?
Find the guided wavelength λg = c / (f0 × √εeff), then set L = λg/4. For stripline with εr = 2.2 at 10 GHz, λg ≈ 20.2 mm so L ≈ 5.1 mm. In microstrip the even and odd modes see different effective permittivities, so average the two εeff values (or use the odd-mode value) and add an open-end length correction for fringing at the coupled-section ends.
What happens to coupling and bandwidth if the coupling length is wrong?
Too short shifts the coupling peak up in frequency and reduces coupling (higher dB); too long shifts the peak down and rolls off above f0, hitting a null at 2 f0. A 10 percent length error moves the center frequency about 10 percent and degrades directivity, since the even-mode and odd-mode phases no longer null cleanly at the isolated port. Microstrip is most sensitive because unequal mode velocities cap directivity regardless of length.