A Bluetooth module‘s radio is only as good as the antenna bolted to it. You can buy the most sensitive SoC on the market, but if the antenna is mismatched, detuned, or buried next to a battery, your real-world range collapses and your certification fails. This article walks through antenna integration for a Bluetooth module from first principles: return loss, matching networks, ground planes, and the mistakes that quietly eat 10 dB of link budget.
Three ways to attach an antenna
| Option | Footprint | Peak gain | Radiation efficiency | Tunable | BOM cost | Design effort |
|---|---|---|---|---|---|---|
| Chip antenna | 3.2 x 1.6 mm | 0 to -2 dBi | 40-60 % | via matching | Low | Medium |
| PCB trace antenna | 15-25 mm | 1-3 dBi | 50-70 % | via matching | None | High |
| U.FL / IPEX + whip | cable-dependent | 2-5 dBi | 70-90 % | Fixed | High | Low |
A chip antenna is the path of least resistance: drop it on the board, add a matching network, tune with a VNA. A PCB trace antenna costs nothing in BOM but costs you board space and a tuning cycle. An external whip wins on efficiency but needs a connector, a cable, and a mechanical mount.
Return loss, S11, and VSWR
The antenna is a 50 ohm load only at one frequency. Off that point it reflects power. We quantify the reflection with S11 (return loss, dB) and the voltage standing wave ratio VSWR:
| S11 (dB) | VSWR | Reflected power |
|---|---|---|
| -6 | 3.0 : 1 | 25 % |
| -10 | 1.92 : 1 | 10 % |
| -15 | 1.43 : 1 | 3.2 % |
| -20 | 1.22 : 1 | 1.0 % |
For 2.4 GHz you usually target S11 <= -10 dB (<=10 % reflected). A Bluetooth module broadcasting at +8 dBm with a -3 dB mismatch (S11 ~= -3 dB, ~50 % reflected) is effectively radiating +5 dBm – you paid for half the range and threw it away.
Matching network: moving ZL to 50 ohm
Most chip and PCB antennas present an impedance far from 50 ohm at 2.44 GHz – commonly 20 + j35 ohm to 40 – j20 ohm. A simple L or pi network of one series capacitor and one shunt inductor (or vice versa) rotates that point onto the Smith chart center. Example values for a 2.44 GHz load of 22 + j35 ohm:
<h1>Goal: move ZL to 50 ohm at 2.44 GHz</h1>
ZL = 22 + 35j # ohms, measured on VNA
<h1>Series C cancels the +j35 reactance</h1>
Xc = -35 # need -35 ohm series reactance
C_series = 1 / (2*pi*2.44e9 * 35) # ~1.87 pF -> use 1.8 pF
<h1>Shunt L brings 22 ohm up to 50 ohm</h1>
R_ratio = 50 / 22 # ~2.27
L_shunt = 3.9e-9 # 3.9 nH typical starting point
<h1>Sweep with VNA until S11 <= -10 dB</h1>
while s11(measure()) > -10:
adjust(C_series, L_shunt)Use 0201/0402 NP0/C0G parts – their tolerance and temperature stability dominate the result. A 10 % capacitor drift here shifts S11 by several dB.
Ground plane and keep-out
The reference ground is part of the antenna. Rules that hold across every Bluetooth module design:
- Keep a clearance of 5-15 mm around and beneath the antenna – no copper, no traces, no components, no silkscreen ground.
- The ground plane should be at least 15 x 15 mm; smaller planes raise the impedance and pull S11 toward -5 dB.
- Feed the antenna with a 50 ohm controlled-impedance microstrip; length matters less than consistency, but keep it short.
- Orient the antenna away from the enclosure wall and any battery. A Li-Po cell two millimeters behind the radiator detunes it by 50-150 MHz and adds a 5-10 dB shield.
Radiation pattern and orientation
A chip antenna is fairly omnidirectional in its board plane but has deep nulls broadside (perpendicular to the board). A PCB trace antenna can be more directional. For a beacon stuck to a ceiling, the null points at the floor where people are – mount it flat, radiator down, or pick an antenna whose pattern covers the volume you care about.
Human body and SAR
2.4 GHz is absorbed by water. Keep the radiator >=5 mm from the body; closer and you both lose efficiency and approach SAR limits (FCC 1.6 W/kg averaged over 1 g, EU 2.0 W/kg over 10 g). Wearable tags should use an antenna tuned *with* the body present, not in free space.
Metal enclosures detune everything
Put a Bluetooth module inside a steel enclosure and the resonance moves. The fix is to tune the matching network with the final enclosure and battery in place – never in free space on the bench. A sealed metal box with no antenna window is a Faraday cage; you need an external feed or an approved RF window.
How to measure
- A VNA gives you S11 and the Smith chart – the only way to tune a matching network properly.
- Radiated vs conducted efficiency in an anechoic chamber tells you the truth: conducted power minus radiated power = losses.
- A quick field test (phone RSSI at 10 m) catches gross errors but not subtle detuning.
Common mistakes
| Mistake | Effect |
|---|---|
| Ground plane < 15 x 15 mm | Impedance shift, S11 ~ -5 dB |
| Copper inside keep-out | Detuning, radiation nulls |
| No matching network | S11 ~ -3 dB, ~50 % reflected |
| Antenna against battery | -10 dB detune, shielding |
| Wrong feed impedance | Standing waves, ripples |
| Tuned only in free space | Fails inside enclosure |
Range math you can defend
In free space, every +6 dB of link budget doubles range. A +3 dBi antenna at both ends adds +6 dB -> 2x range. But a 50 % efficient antenna costs -3 dB ERP -> ~0.7x range. Net: a well-matched, efficiently radiating Bluetooth module easily beats a “higher power” module with a bad antenna by 2-3x in distance.
OEM checklist
1. Get the vendor’s reference matching schematic and layout – copy it exactly.
2. Budget a VNA tuning pass on the first article.
3. Define the keep-out in the mechanical drawings.
4. Certify with the final enclosure and battery.
5. Specify antenna gain and efficiency, not just TX power, in your datasheet claims.
A Bluetooth module is a system, not a chip. Get the antenna right and the rest of the link budget takes care of itself.