Once an RF signal has been generated, conditioned, and amplified, it must be routed — distributed to multiple antennas, switched between transmit and receive paths, combined for higher power, or modulated with information. The signal distribution and switching subsystem is the RF plumbing that connects sources to destinations with minimal loss and maximum isolation. This article covers the seven core functions — RF Distribution Networks, Power Combiners, Splitters, Switch Matrices, Antenna Switches, Modulators, and Demodulators — with detailed PCB design analysis for each.

1. RF Distribution Network PCB Design

RF distribution networks route a common signal to multiple destinations — most prominently in phased array feed networks, where one transmitter drives 16, 64, or even 1024 antenna elements. The distribution network must maintain equal amplitude and phase at every output port while presenting a well-matched input impedance.

1.1 Corporate Feed Networks

The corporate (parallel) feed network uses a binary tree of two-way power dividers to split one input into N outputs. For N = 64, this requires 63 two-way dividers arranged in six levels. The advantages are equal path lengths to all outputs (by construction) and broad bandwidth. The PCB challenge is fitting this tree within the available area — for a 64-element array at 28 GHz, the total network occupies approximately 50 mm × 50 mm on a single RF layer. Careful layout avoids crossovers: all dividers and connecting transmission lines must exist on the same layer, which forces a serpentine or H-tree topology. When crossovers are unavoidable, they are implemented using air-bridges (wire bonds) or zero-ohm resistors on the opposite side, each adding 0.1–0.3 dB of insertion loss.

1.2 Serial (Traveling-Wave) Feed Networks

In a serial feed, a single transmission line passes each antenna element, with directional couplers tapping off a fraction of the power at each element. This topology is simpler to lay out but suffers from progressive amplitude taper — the last element receives less power because preceding elements have extracted their share. The taper is compensated by designing each coupler with a different coupling coefficient, increasing from ~3 dB at the first tap to ~20 dB at the last. The transmission line between couplers must maintain precise phase length, as any error compounds across the array. At 60 GHz, the λ/2 spacing between elements is only 2.5 mm, demanding extraordinary PCB precision.

2. Power Combiner and Splitter PCB Design

Power combiners and splitters are reciprocal — a Wilkinson divider used as a splitter can equally function as a combiner. However, the design priorities differ: splitters emphasize output port isolation, while combiners emphasize power handling and thermal management.

2.1 Wilkinson Power Divider/Combiner

The Wilkinson divider is the most widely used RF splitter/combiner due to its simplicity, good isolation (>20 dB), and perfect output matching when all ports are terminated in the system impedance. It consists of two quarter-wavelength (λ/4) transmission lines with characteristic impedance Z₀√2 (70.7 Ω for a 50 Ω system) and a 2Z₀ (100 Ω) isolation resistor connecting the output ports. The PCB implementation of the isolation resistor is critical: it must be placed exactly at the junction of the two λ/4 sections with the shortest possible lead length. For a Wilkinson at 10 GHz, the λ/4 section is approximately 4 mm on Rogers 4350B, and the isolation resistor should be a flip-chip or 0201 package placed within 0.5 mm of the junction. Any additional trace length adds series inductance that degrades isolation at the upper band edge.

2.2 N-Way Radial Combiner

For combining more than two signals, the N-way radial (Wilkinson) combiner uses N quarter-wave transformers from each input converging at a common output node, with isolation resistors connected in a star or ring configuration. The PCB challenge is geometric: placing N transmission lines symmetrically around a central point. For N = 8, the lines radiate at 45° intervals, and the isolation resistors must be electrically connected to a common floating node. This node is implemented as a small copper island at the geometric center, with the star-connected resistors attached around its perimeter — a layout pattern well-suited to Superb Tech's precision etching capabilities.

2.3 High-Power Combining: Thermal and Breakdown Considerations

When combining multiple 100 W PA outputs into a single 800 W feed, the combiner's PCB must handle both the RF currents and the dissipated heat. The combiner traces must be wide enough to carry the peak RF current without excessive ohmic heating; for 800 W average power at 1 GHz, the RMS current is 4 A (50 Ω), and the trace width should be at least 1.5 mm on 1 oz copper to limit temperature rise to<20°c. the="" isolation="" resistors="" must="" dissipate="" power="" when="" input="" signals="" are="" not="" perfectly="" amplitude-and-phase="" balanced="">

3. RF Switch Matrix PCB Design

RF switch matrices route any of M inputs to any of N outputs, forming the backbone of automated test systems, satellite payloads, and reconfigurable communication systems. A 16×16 switch matrix at 18 GHz presents one of the most challenging PCB design problems in RF engineering.

3.1 Switch Topologies: Blocking vs. Non-Blocking

A fully non-blocking matrix allows any input to connect to any unused output. This is typically implemented using crossbar (M × N crosspoints, each a SPST switch) or Clos architectures. The PCB must route M input traces and N output traces in a grid pattern, with each crosspoint implemented as a series-shunt switch (typically a PIN diode or GaAs FET). The trace grid in a 16×16 matrix has 32 RF traces crossing each other, requiring careful layer assignment: input traces on layer 1, output traces on layer 3, with a continuous ground plane on layer 2 providing isolation. Each crosspoint connects from an input trace (layer 1) through a via to the switch component (mounted on layer 1) and back through a via to the output trace (layer 3). The via transitions add approximately 0.15 dB each at 18 GHz, totaling 0.6 dB for a complete through-path — significant enough to require accounting in the system loss budget.

3.2 Switch Driver and Control Routing

Each switch element requires DC bias — for PIN diode switches, 10–20 mA forward current; for FET switches, 0/-5 V or 0/+3.3 V control voltages. A 16×16 matrix has 256 control lines that must be routed without crossing or coupling onto the RF paths. The solution uses inner-layer routing: all DC control traces on layers 4–5, with only the RF traces on the outer layers. Bias tees or RF chokes (typically 10–100 nH inductors with >1 kΩ impedance at the operating frequency) are placed at each switch node to inject the DC control while isolating the RF path. Superb Tech's multilayer capability (up to 30 layers) enables the layer partitioning that cleanly separates RF, DC, and digital signals.

4. Antenna Switch PCB Design

The antenna switch — increasingly referred to as the Antenna Switch Module (ASM) or Antenna Tuning Switch — selects which antenna port connects to which transceiver path. In modern smartphones, the antenna switch may route 10+ frequency bands across 4–6 antennas, with insertion loss<1.5 db="" and="">25 dB isolation between inactive paths.

4.1 SPnT Switch Integration

A Single-Pole N-Throw (SPnT) switch uses one common RF port and N selectable RF ports. The PCB must provide a low-inductance ground for each throw's shunt switching element. For a high-isolation SP4T at 6 GHz, the ground vias for the shunt FETs must be within 0.3 mm of the switch IC pad to keep the ground inductance below 0.2 nH; a 0.5 nH ground inductance creates 19 Ω of impedance at 6 GHz, degrading the OFF-state isolation by 6–10 dB. Multiple parallel ground vias and via-in-pad construction are standard practice.

4.2 Antenna Switch with Integrated Filtering

Increasingly, antenna switches incorporate bandpass filtering (multiplexing) on each throw port to provide both switching and frequency selectivity. An antenna switch plus triplexer for WiFi 7 selects among 2.4, 5, and 6 GHz bands with >30 dB inter-band isolation. The filter sections for each band must be physically separated on the PCB, with grounded via fences or shield walls between them, to prevent coupling that would degrade the filter's out-of-band rejection.

5. RF Modulator PCB Design

Modulators imprint information onto an RF carrier through amplitude, phase, frequency, or combined (vector) modulation. The PCB's role is to provide a clean, well-matched environment for the modulator IC and to maintain the quality of both the baseband (modulating) and RF (carrier) signals.

5.1 I/Q Vector Modulator PCB

The I/Q modulator (or quadrature modulator) is the universal modulator for digital communications, producing any combination of amplitude and phase by varying the I (in-phase) and Q (quadrature) baseband inputs. The PCB requires four critical interfaces: the LO input (50 Ω, typically +0 to +5 dBm), the I and Q baseband inputs (differential, 100 Ω, DC to 500 MHz bandwidth for 5G 100 MHz carriers), and the modulated RF output (50 Ω). The baseband traces must have bandwidth sufficient for the modulation; for a 400 MHz 5G NR carrier, the I/Q traces must maintain flat group delay and<0.1 200="" db="" amplitude="" variation="" from="" dc="" to="" at="" least="" mhz.="" this="" is="" typically="" achieved="" using="" coplanar="" waveguide="" structures="" with="" ground="" on="" adjacent="" layers="">

5.2 Direct Digital Synthesis (DDS) and Digital Modulation

Direct digital synthesizers generate modulated RF directly from digital samples, eliminating the analog modulator. The DDS PCB challenge is the DAC output interface: a high-speed current-steering DAC (sampling at 2–10 GSPS) produces output tones from DC to Nyquist, and the PCB must provide a broadband balun and reconstruction filter with<1 db="" flatness="" across="" the="" entire="" output="" band.="" dac="" clock="">

6. RF Demodulator PCB Design

Demodulators recover the baseband information from the modulated RF carrier. The most common topology in modern systems is the direct-conversion (zero-IF) I/Q demodulator, which uses the same hardware architecture as the modulator operating in reverse.

6.1 I/Q Demodulator: Image Rejection and DC Offsets

The I/Q demodulator down-converts the RF signal to baseband using an LO at the carrier frequency. The primary degradation mechanisms — I/Q imbalance and DC offset — are PCB-sensitive. I/Q amplitude imbalance of 0.2 dB and phase imbalance of 2° degrades the Error Vector Magnitude (EVM) by approximately 2–3 dB, which can mean the difference between 256-QAM and 64-QAM operation. The PCB must ensure symmetric routing for the I and Q LO distribution paths, identical loading at the I and Q mixer outputs, and matched transmission line impedances for the in-phase and quadrature signal paths.

6.2 Demodulator LO Leakage and Self-Mixing

LO leakage from the demodulator's LO port to its RF input causes self-mixing, producing a DC offset that can saturate the baseband amplifiers. At the PCB level, LO-to-RF isolation is achieved through: physical separation (>5 mm between LO and RF routing), grounded coplanar waveguide for all LO traces, and shielding cans over the demodulator IC. The self-mixing DC offset can be 10–100 mV; the baseband PCB traces must include AC-coupling capacitors to block this offset from downstream stages.

7. System Integration: The Complete Signal Routing Subsystem

A complete RF routing subsystem for a modern communication or radar platform integrates distribution, switching, and modulation/demodulation onto a single PCB or interconnected set of PCBs. The key integration challenges are: maintaining >60 dB isolation between the high-power transmit distribution network and the low-noise receive switching network, preventing LO leakage from the modulator/demodulator into the distribution network, and keeping the total end-to-end insertion loss within the system budget.