Designing a PCB Antenna: Trace Sizing, Prototyping, and First-Spin Testing

A PCB antenna etches the radiating element—typically a meander line or inverted-F structure—directly into the FR4 copper layer. This approach eliminates the $1 to $3 per-unit cost of surface-mount chip antennas or SMA connectors, but it demands strict 50-ohm impedance control, precise trace geometry, and a well-managed ground plane. If you are transitioning an RF design from a breadboard to a custom board, the physical realities of high-frequency electromagnetics will immediately expose any prototyping shortcuts.

The Physics of PCB Antennas and Trace Sizing

The most common topology for hobbyist and commercial IoT boards (like those using the ESP32 or nRF52 series) is the Inverted-F Antenna (IFA). The IFA relies on a shorted stub to ground and a specific trace length to resonate at your target frequency (e.g., 2.44 GHz for Wi-Fi/Bluetooth). The feedline connecting your RF transceiver to the antenna must be routed as a 50-ohm microstrip or coplanar waveguide. On standard 62-mil (1.6mm) FR4 with 1 oz copper, a 50-ohm microstrip requires a trace width of roughly 110 mils (2.8mm).

However, while the RF signal trace is sized for impedance, the DC power traces feeding the RF Power Amplifier (PA) must be sized for current. A Wi-Fi PA can draw peak currents of 350mA to 500mA during transmission bursts. Undersizing these power traces introduces voltage drop and parasitic inductance, which causes the PA to brown out or compress its output power.

To determine what trace width a specific current needs, we rely on the IPC-2221 standard. Below is a reference table for internal and external traces targeting a conservative 10°C temperature rise above ambient.

Trace Width vs. Current Capacity (10°C Temp Rise)

Current (A)	1 oz Copper (External)	2 oz Copper (External)	1 oz Copper (Internal)
0.5 A	10 mils (0.25 mm)	5 mils (0.13 mm)	20 mils (0.50 mm)
1.0 A	20 mils (0.50 mm)	10 mils (0.25 mm)	45 mils (1.14 mm)
2.0 A	50 mils (1.27 mm)	25 mils (0.64 mm)	120 mils (3.05 mm)
3.0 A	85 mils (2.16 mm)	40 mils (1.02 mm)	200 mils (5.08 mm)

Note: Internal layers dissipate heat poorly compared to external layers, requiring roughly double the trace width for the same current.

Breadboard-to-PCB Migration: What Survives and What Fails

When moving an RF circuit from a solderless breadboard to a PCB, you must understand which mistakes survive the transition. Breadboards introduce 2pF to 5pF of parasitic capacitance between adjacent rows. At audio frequencies, this is invisible. At 2.4 GHz, the capacitive reactance ($X_c = \frac{1}{2\pi f C}$) drops to roughly 13 ohms, effectively shorting your RF signal to ground.

Mistakes that survive the migration include floating grounds, inadequate high-frequency decoupling, and routing digital clock lines parallel to RF paths. To ensure a clean migration, follow this checklist before finalizing your Gerber files:

• Consolidate the Ground Plane: Never use a fragmented ground plane for RF. The return currents for your 50-ohm microstrip need a continuous, unbroken copper pour directly beneath the trace. Slotted grounds create inductive bottlenecks.
• Decoupling Placement: Place a 100nF (0402 or 0201 package) ceramic capacitor as close to the VDD pin of the RF chip as physically possible, followed by a 10µF bulk capacitor. The 100nF cap must have a high self-resonant frequency (SRF) to shunt 2.4 GHz noise.
• Keep Digital Away from RF: SPI and I2C clock lines generate sharp harmonic edges. Route them on the opposite side of the board from the PCB antenna, separated by a solid ground plane.
• Component Footprints: Use 0402 (1005 metric) footprints for your RF matching network. Larger 0805 pads introduce parasitic pad capacitance that will detune your antenna.

Workshop Safety and Soldering Protocols for RF Boards

Soldering 0402 matching components and fine-pitch RF ICs requires strict thermal control and environmental safety. Modern RF prototyping often utilizes lead-free SAC305 (Sn96.5/Ag3.0/Cu0.5) alloy to comply with RoHS, though hobbyists may still prefer eutectic Sn63/Pb37.

Temperature and Alloy Specifics:

• SAC305 (Lead-Free): Requires a tip temperature of 350°C to 360°C. Use a chisel tip (e.g., Hakko T18-D12) to maximize thermal transfer to the ground pad. Dwell time should not exceed 3 seconds to prevent pad lift-off on thin FR4.
• Sn63/Pb37 (Leaded): Requires a tip temperature of 315°C to 320°C. The wetting action is faster, allowing for 1.5 to 2-second dwell times.
• Flux Selection: Use a no-clean, low-solids flux pen (like Kester 186 or MG Chemicals 8341) for RF matching networks. Liquid flux can wick under 0402 capacitors and alter the dielectric constant of the local area, slightly shifting your resonant frequency.

Testing Your First-Spin RF Board

Never power on a first-spin RF board and immediately attempt to pair it over Bluetooth. You must verify the physical layer first. Here is how to test a first-spin board methodically:

1. Visual and Continuity Check: Inspect the PCB antenna trace under a 10x loupe. Look for over-etching (which narrows the trace and raises impedance) or copper slivers. Use a digital multimeter (DMM) in continuity mode to verify there is no dead short between the RF feedline and the ground plane.
2. Vector Network Analyzer (VNA) Sweep: Connect a VNA to the antenna feed point. For hobbyists, a calibrated NanoVNA-F V2 or LibreVNA ($60–$150) is sufficient for sub-3GHz work. Professionals will use a Keysight or Rohde & Schwarz benchtop unit.
3. Analyze S11 (Return Loss): Sweep from 1 GHz to 3 GHz. You are looking for the S11 dip (the point where the antenna absorbs power rather than reflecting it). A well-tuned PCB antenna will show an S11 of < -10 dB (which equates to a VSWR of < 2:1) at your target center frequency (e.g., 2.44 GHz).
4. Smith Chart Verification: Switch the VNA display to a Smith Chart. The trace should pass through or very near the center 50-ohm point at resonance. If the loop is inductive (shifted right), you need to add series capacitance or shorten the antenna stub.

PCB Antenna FAQ

What size ground plane do I need for a PCB antenna?

The ground plane acts as the counterpoise (the "other half" of the dipole) for a PCB antenna. For a 2.4 GHz inverted-F antenna, the ground plane should ideally extend at least a quarter-wavelength (approx. 31mm or 1.2 inches) in all directions from the antenna feed point. If your PCB is smaller than this, the antenna's resonant frequency will shift upward, and the radiation pattern will become distorted and asymmetrical. Always tune the antenna on the final, fully-populated board, as nearby components and batteries alter the ground plane's effective electrical size.

Should I remove the ground pour under a PCB antenna?

Yes, absolutely. You must keep the ground plane completely clear (voided) directly beneath the radiating element of the PCB antenna. If you place a solid ground pour under the meandering trace, you create a parasitic parallel-plate capacitor. This capacitance couples the RF energy directly into the ground plane, severely reducing radiation efficiency and shifting the resonant frequency downward. The 50-ohm feedline requires a ground plane beneath it, but the moment the trace transitions into the radiating antenna structure, the ground layer below it must be removed.

How do I tune a PCB antenna matching network if my S11 is off-center?

If your VNA shows the S11 dip is at 2.5 GHz but you need 2.4 GHz, your antenna is too short (electrically). You can fix this physically by lengthening the open stub of the inverted-F trace. However, it is much easier to fix it using a Pi-network matching circuit between the RF chip and the antenna. If the impedance is inductive, add a shunt capacitor (e.g., 1.2pF) to ground. If it is capacitive, add a series inductor (e.g., 2.2nH). Use high-Q, tight-tolerance (±0.1pF) NP0/C0G ceramic capacitors; standard X7R dielectrics will drift with temperature and RF voltage, causing your tuning to fail in the field. For deeper matching theory, refer to the Analog Devices RF layout guidelines.