How does a box-section filter differ from an inline filter?

In a standard inline filter (like a Chebyshev), the resonators are arranged in a straight line. Energy flows sequentially from resonator 1, to 2, to 3, to 4. Because there is only one path for the signal to take, the filter's rejection skirts slope downward gradually. To make the skirts steeper, you must add more resonators, which increases the physical size and insertion loss of the filter. A box-section topology arranges four resonators in a square. Energy flows from 1 to 2 to 3 to 4, but a deliberate gap or probe is added so that energy also jumps directly from resonator 1 to resonator 4. This cross-coupling creates a second, parallel path for the signal.

Why does cross-coupling create steeper rejection skirts?

When a signal takes two different paths through a filter and recombines at the output, the phase difference between those two paths determines what happens. At the center of the passband, the two paths are designed to be in-phase, adding constructively to allow the signal through. However, at a specific frequency just outside the passband, the phase of the primary path (1-2-3-4) and the phase of the cross-coupled path (1-4) will be exactly 180 degrees apart. The two signals destructively cancel each other out entirely. This creates a 'transmission zero'—a deep notch of nearly infinite attenuation right next to the passband. This makes the rejection skirt drop off like a cliff, rather than a gradual slope, without requiring extra resonators.

Why are box-section filters harder to tune?

Standard inline filters can be tuned sequentially. A technician can tune resonator 1, then 2, then 3, isolating each stage. In a box-section topology, the cross-coupling creates a feedback loop. Adjusting the tuning screw on resonator 4 affects the resonance of resonator 1 because they are directly coupled. This high degree of interaction means tuning cannot be done sequentially; it requires an iterative matrix approach. Technicians often use specialized VNA software with time-domain reflectometry (TDR) or phase-based extraction algorithms to untangle the coupled responses and tune the filter correctly.

Passive Components

Box-Section Topology

A base station filter must pass a transmit signal at 1990 MHz but block a receiver signal at 1980 MHz. The bands are so close together that a standard inline Chebyshev filter would need 12 resonators to drop off fast enough, resulting in massive size and unacceptable insertion loss. Instead, the designer arranges the resonators in a square—a box-section. They cut a small window allowing energy to bypass the middle resonators and jump directly from resonator 1 to resonator 4. At exactly 1980 MHz, the signal traveling the main path (1-2-3-4) arrives exactly 180 degrees out of phase with the signal traveling the bypass path (1-4). The two paths violently cancel each other out, creating a "transmission zero"—a deep notch of infinite attenuation right on the receiver frequency. By using multipath phase cancellation instead of brute-force sequential filtering, the box-section provides cliff-edge selectivity with half the resonators.

Category: Passive Components

Mechanism: Non-adjacent cross-coupling

Benefit: Transmission zeros (Notches)

Inline vs. Box-Section Filter Topologies

Feature	Standard Inline Topology	Box-Section (Cross-Coupled)
Coupling Path	Sequential (1→2→3→4)	Multipath (1→2→3→4 AND 1→4)
Transmission Zeros	None (All at infinity)	Finite zeros (Placed near passband)
Skirt Selectivity	Gradual slope (6 dB/octave per pole)	Cliff-edge drop-off
Group Delay	Peaks heavily at band edges	Can be self-equalized (flatter delay)
Tuning Difficulty	Low (Sequential tuning possible)	Very High (Iterative, highly interactive)

Transmission Zero Generation:
For a zero to exist at ω_z, the transfer function of the main path H_m(ω_z) must equal the negative of the cross-coupled path transfer function H_c(ω_z).
H_m(ω_z) + H_c(ω_z) = 0

Coupling Sign (Capacitive vs. Inductive):
If the main path couplings are magnetic (inductive, positive), making the 1-4 cross-coupling electric (capacitive, negative) will place transmission zeros on the high-frequency side of the passband.

Common Questions

Frequently Asked Questions

How does it differ from an inline filter?

In an inline filter, energy has only one path to take: from resonator 1 to 2, to 3, to 4. In a box section, resonators are folded into a square, and a direct path is opened between 1 and 4. Energy now has two parallel paths through the filter.

Why does cross-coupling make skirts steeper?

Because the two parallel paths have different phase lengths. By carefully tuning the coupling strengths, the designer ensures that at a specific frequency just outside the passband, the two paths arrive exactly 180 degrees out of phase. They cancel each other out, creating a deep attenuation notch (a transmission zero). This creates a much steeper rejection skirt than a standard filter.

Why are they so hard to tune?

Cross-coupling creates a feedback network. In an inline filter, you can tune resonator 1, and it barely affects resonator 4. In a box section, resonator 1 is directly coupled to resonator 4. Turning the tuning screw on cavity 4 instantly detunes cavity 1. This requires highly skilled technicians or advanced VNA software using phase-extraction algorithms to align.

Related Terms

← Bond Wire Inductance Breakdown (PA) →

Filter Design

Cross-Coupling Matrix Analyzer

Enter your required passband and rejection notch frequencies. Calculate the exact N-by-N coupling matrix, including the main-line and cross-coupling coefficients needed to synthesize transmission zeros.

Calculate Coupling Matrix