Cavity Filter PCB | High-Q Base Station Filter Boards

What Is a Cavity Filter PCB?

A cavity filter PCB is the printed-circuit substrate that forms the floor, feed network and tuning interface of a metallic-cavity bandpass or band-reject filter. The board carries coaxial probe feeds, coupling loops or aperture slots into machined or die-cast aluminum cavities whose unloaded Q routinely exceeds 5 000 at 700 MHz–3.5 GHz. Because the cavity walls, not the PCB traces, set the resonant frequency, the board’s job is to launch and extract energy with minimal added loss while surviving the thermal and mechanical environment of a base station radio unit.

Why Cavity Filters Still Dominate Base Stations

Planar filters on microstrip or stripline are compact but limited to unloaded Q values of roughly 100–300 on standard laminates. A machined-cavity resonator at 900 MHz easily reaches Q_u of 8 000–12 000, which translates directly into steeper skirts, lower passband insertion loss and higher out-of-band rejection—exactly what a macro-cell duplexer or combiner demands. The PCB inside each cavity is the bridge between the high-Q air-dielectric resonator and the rest of the RF signal chain.

Typical Deployment Scenarios

• Macro base-station duplexers – 4 to 12 cavity sections, Tx/Rx isolation > 80 dB.
• Tower-top combiners – multi-operator sharing on a single antenna, requiring < 0.5 dB insertion loss per channel.
• Public-safety repeaters – UHF/VHF cavity notch filters rejecting co-site interference.
• Broadcast transmitters – FM/TV bandpass filters handling kilowatts of continuous power.

Key Performance Parameters

Material Selection for the Cavity Filter PCB

The substrate under the feed probes and coupling structures sees strong RF fields, so both dielectric loss and dimensional stability matter. Common choices, ranked by increasing performance and cost:

FR-4 (Limited Use)

Standard FR-4 (Df ≈ 0.020 at 1 GHz) is acceptable only for the DC control and bias sections of a cavity filter assembly—never for the RF feed network. Its high loss tangent and inconsistent Dk (4.2–4.8 depending on resin content and glass weave) would degrade the match at each probe.

Low-Loss Thermoset Laminates

Materials like Panasonic Megtron 6 (Dk 3.71, Df 0.002 at 1 GHz) or Isola Astra MT77 (Dk 3.00, Df 0.0017) offer a practical mid-range. They can be processed on standard FR-4 press cycles, keep feed-line loss below 0.01 dB/cm at 2 GHz, and hold Dk within ±2 % over –40 °C to +85 °C.

PTFE and Ceramic-Filled Laminates

For the highest performance, Rogers RO4003C (Dk 3.38, Df 0.0027) or RO4350B (Dk 3.48, Df 0.0037) are popular because they combine low loss with standard epoxy-compatible processing. Where the cavity filter operates above 3 GHz or demands the tightest Dk tolerance, Rogers RT/duroid 5880 (Dk 2.20, Df 0.0009) is the benchmark, though it requires specialized drilling and handling. Consult the Rogers RO4003C datasheet for detailed Dk vs. frequency curves.

Material Comparison

Plating and Surface Finish

Skin-depth effects at UHF and low-microwave frequencies mean the surface conductivity of both the cavity walls and the PCB copper is critical. At 900 MHz the skin depth in copper is about 2.2 µm, so the outermost microns of plating dominate loss.

• Silver plating (2–5 µm) – highest conductivity (6.30 × 10⁷ S/m), preferred on cavity walls and sometimes on PCB pads. Tarnish can raise loss over time; a passivation layer helps.
• ENIG (electroless nickel / immersion gold) – convenient for PCB pads, but the nickel barrier layer (3–6 µm, conductivity only 1.4 × 10⁷ S/m) sits right in the skin-depth zone and adds measurable loss. Acceptable for coupling loops where loss budget allows.
• Hard gold (0.75–1.25 µm over nickel) – good for tuning-screw contact pads; the thin gold wears better than ENIG under repeated adjustment.
• Immersion silver – a cost-effective alternative to electrolytic silver for the PCB itself; provides 0.15–0.40 µm of silver directly on copper with no nickel, preserving low loss.

Feed and Coupling Structures on the PCB

The PCB inside a cavity filter typically implements one or more of these structures:

Probe Feed (Capacitive Coupling)

A short copper stub or pad extends from a 50 Ω microstrip into the cavity volume. The probe length and width set the coupling coefficient. Tight control of the pad-to-ground gap (often 0.2–0.5 mm) is essential; a 50 µm etching variation can shift the coupling by 1–2 %, which in a narrow-band filter is the difference between meeting and missing the return-loss spec.

Loop Feed (Magnetic Coupling)

A printed or wire loop on the PCB couples magnetically to the cavity’s H-field. Loop feeds are less sensitive to gap tolerances but occupy more board area. They are common in lower-frequency (VHF/UHF) cavity filters.

Inter-Cavity Coupling Apertures

In multi-pole filters the cavities are coupled through irises or apertures in the shared wall. The PCB may carry a coupling slot or a bypass trace that lets the designer fine-tune the bandwidth electrically, rather than relying solely on mechanical iris dimensions.

Layout Rules and Fabrication Notes

Ground Integrity

The PCB ground plane must make low-impedance contact with the cavity body around the full perimeter. Use a dense ring of plated through-holes (via pitch ≤ λ/20 at the highest frequency of interest) and specify a flat, un-masked copper border for gasket or solder-seal contact. Any gap in the ground seal creates a slot antenna that radiates energy out of the cavity and kills Q.

Dimensional Tolerances

A cavity filter PCB for a 1.8 GHz duplexer typically needs board outline tolerance of ±50 µm (±2 mil) and etched-feature tolerance of ±25 µm (±1 mil). These are tighter than standard IPC-6012 Class 3 minimums, so specify them explicitly on the fabrication drawing.

Thermal Management

High-power duplexers in macro base stations dissipate several watts per cavity. The PCB should include thermal vias under any power-handling trace, and the laminate’s Tg should exceed the maximum ambient plus the expected self-heating (typically Tg ≥ 170 °C for outdoor radio units). Aluminum-backed or metal-core substrates are sometimes used when the filter doubles as a heat spreader for an adjacent RF power amplifier board.

Tuning Access

Each cavity has a tuning screw that penetrates through the lid into the resonant volume. The PCB must provide clearance holes with adequate annular ring to avoid cracking during screw adjustment. For filters that are tuned once at the factory, a locking compound is applied; for field-tunable designs, the hole must survive hundreds of adjustment cycles without copper fatigue.

Filter Response Prototyping with an RC Calculator

Before committing to a full electromagnetic simulation of the cavity, engineers often prototype the coupling network on the PCB as a lumped-element equivalent to verify bandwidth and roll-off targets. The RC time constant of the feed network’s coupling capacitor and load resistance sets the 3 dB point of each section. Enter your estimated coupling capacitance and load impedance below to check the resulting cutoff frequency and debounce time constant.

For example, a 1.5 pF coupling capacitor into a 50 Ω load gives a 3 dB corner near 2.12 GHz—useful as a quick sanity check that the probe geometry is in the right ballpark before you run a 3-D EM solve. If the corner frequency lands more than 20 % away from your target, resize the probe pad area (which scales roughly linearly with capacitance) and re-check.

Assembly and Integration Considerations

• Solder seal vs. conductive gasket – Soldering the PCB to the cavity floor gives the lowest contact resistance but makes rework difficult. Conductive elastomer gaskets (silver-filled silicone, < 5 mΩ per cm) allow disassembly but add a small amount of loss.
• Connector interface – SMA or N-type connectors are typically mounted through the cavity wall with the center pin soldered to the PCB feed trace. Specify the connector footprint with a controlled-impedance transition; a step from 50 Ω coax to an uncontrolled pad can introduce a VSWR bump of 1.3:1 or worse.
• ESD and lightning protection – Tower-mounted filters need gas-discharge tubes or quarter-wave stubs on the PCB to shunt transient energy. These protection elements must not load the passband; place them at a high-impedance point on the feed line.

Testing and Tuning Workflow

1. Bare-board verification – Measure impedance of each probe pad with a VNA calibrated to the connector reference plane. Target |S11| < –25 dB at center frequency before the cavity lid is installed.
2. Single-cavity sweep – Install one cavity at a time, sweep S21, and adjust the tuning screw until the resonant peak aligns with the design frequency within ±100 kHz.
3. Full-filter assembly – Bolt all cavities together, sweep the complete S-parameter set, and iteratively adjust coupling screws and tuning screws. A typical 6-pole duplexer requires 15–30 minutes of manual tuning by an experienced technician.
4. Temperature soak – Cycle from –40 °C to +55 °C and verify the passband stays within spec. Frequency drift should be < ±200 kHz for a 1.8 GHz BTS duplexer across this range.

Cavity Filter PCB vs. Planar Filter PCB

Choose a cavity filter PCB when insertion loss, selectivity and power handling are paramount and volume/weight is acceptable. Choose a planar filter when size, weight and cost per unit dominate, and a few dB of extra loss is tolerable.

Design Checklist

• Select laminate Df < 0.003 for the RF feed section; FR-4 only for DC/control.
• Specify immersion silver or electrolytic silver finish on RF pads to avoid nickel loss.
• Ground-via pitch ≤ λ/20 around the cavity perimeter.
• Board outline tolerance ±50 µm; etch tolerance ±25 µm.
• Clearance holes for tuning screws with ≥ 0.25 mm annular ring.
• Thermal vias under high-power traces; Tg ≥ 170 °C laminate.
• Controlled-impedance transition at every connector footprint.
• Include ESD/lightning protection at a feed-line high-impedance node.

Frequently Asked Questions

Can I use FR-4 for a cavity filter PCB?

Only for the DC bias, control or digital sections of the assembly. FR-4’s high loss tangent (Df ≈ 0.020) and Dk variation make it unsuitable for the RF probe or coupling network, where it would degrade cavity Q by introducing unnecessary dielectric loss at the feed point.

What plating gives the lowest loss on a cavity filter PCB?

Immersion silver or electrolytic silver applied directly over copper, with no nickel barrier, gives the lowest RF surface resistance. ENIG is acceptable where loss budget permits, but the nickel underlayer sits within the skin depth at UHF and adds measurable resistive loss.

How many poles does a typical base-station duplexer cavity filter have?

Most macro-cell duplexers use 4 to 8 poles per path (Tx and Rx), with 6 poles being the most common compromise between isolation (> 80 dB), insertion loss (< 0.8 dB) and physical size.

What causes frequency drift in a cavity filter?

Thermal expansion of the aluminum cavity body is the primary source. A CTE mismatch between the cavity walls and the PCB substrate can shift the probe coupling and change the loaded Q. Using Invar tuning rods and a low-CTE laminate (z-axis CTE < 40 ppm/°C) keeps drift below ±1 ppm/°C. Additional resources on resonator theory are available from the IEEE Microwave Theory and Techniques Society.

How is a cavity filter PCB different from an RF attenuator board?

A cavity filter PCB is optimized for minimal insertion loss and maximum Q in a resonant structure, while an RF attenuator PCB is designed to dissipate a precise amount of signal power across a broad bandwidth. The material, thermal and tolerance requirements diverge significantly.