Internal antennas win on cost, size and ruggedness; external antennas win on efficiency, bandwidth and placement freedom. The decision usually turns on three questions: is the enclosure metal, how much ground plane do you have, and can the device be oriented freely.
| Factor | Internal (chip / PCB / FPC) | External (whip / magnetic / fiberglass) |
|---|---|---|
| Efficiency | 40–70 % typical at 2.4 GHz on small boards; 60–80 % for well-laid-out FPC/PCB | 80–95 % typical at sub-GHz |
| Bandwidth | Narrower; miniaturisation costs bandwidth | Wider; easier to cover 863–928 MHz or 698–2690 MHz |
| Size / integration | No protrusion, no assembly step, no connector | Needs a connector, a bulkhead and routing |
| Unit cost | Lowest at volume (a chip antenna is cents; FPC a little more) | Higher: antenna + connector + cable + assembly labour |
| Engineering cost | Higher: layout, keep-out, matching, chamber time | Lower: pick a 50 Ω part, verify VSWR |
| Ruggedness / vandal resistance | Excellent — nothing to break off | Vulnerable; needs a robust mount or a short stubby |
| Placement freedom | None — the antenna is where the board is | High — magnetic base, mast, outside the cabinet |
| Detuning risk | High — battery, display, hand, housing all detune it | Low — the antenna is outside the noise |
| Waterproofing | Trivial (already inside) | Requires a sealed bulkhead or IP-rated assembly |
| Regulatory EIRP headroom | Lower gain helps stay under EIRP limits | Higher gain may push you over the limit — check |
| Best for | Wearables, handhelds, consumer sensors, high-volume SKUs | Meters, gateways, industrial cabinets, global SKUs |
If yes, an internal antenna will not radiate. RF cannot escape a Faraday cage. Your options are an external antenna, or a deliberate aperture — a plastic window in the housing with the antenna directly behind it, which is a mechanical design project in its own right.
Partial metal matters too: a metal battery can, a shield can, or a display frame 5 mm from the element will detune and absorb. Keep a keep-out zone free of copper and metal on all layers under and around the antenna, sized as the antenna vendor specifies.
For most chip and PCB antennas, the ground plane is half the antenna. A 2.4 GHz chip antenna designed for a 40 × 15 mm ground plane will shift up in frequency and lose efficiency on a 20 × 10 mm board. In practice:
| Ground plane (at 2.4 GHz) | Expectation |
|---|---|
| Small compared with λ/4 (≈30 mm) | Significant detune; efficiency may drop below 40 % |
| Around λ/4 | Vendor's specified performance, subject to matching |
| Large (≥ λ/2) | Closest to datasheet; the clearest case for internal |
At sub-GHz the wavelengths are long (λ/4 ≈ 8 cm at 868 MHz), so a compact device almost never provides enough ground. This is why sub-GHz products default to external antennas unless the whole product is large.
An internal antenna inherits the device's orientation. If the device is wall-mounted in a random position, or worn on a body, or buried in a pit, the pattern points wherever the device points. An external antenna — especially on a magnetic base or a short cable — can be placed where the RF is good, independent of where the electronics need to be.
``` Is the enclosure metal or mostly metal? └─ Yes → EXTERNAL (or a designed plastic window + internal) └─ No ↓
Is the ground plane >= about a quarter wavelength on the longest side? └─ No → EXTERNAL, or accept materially reduced performance └─ Yes ↓
Is the device installed in a pit, cabinet, basement or against a wall? └─ Yes → EXTERNAL (move the antenna out of the lossy environment) └─ No ↓
Do you ship to more than one RF region (868 and 915, or 470 and 915)? └─ Yes → EXTERNAL wideband, or region-specific internal variants └─ No ↓
Is the link budget comfortable (>=10 dB of margin)? └─ Yes → INTERNAL └─ No → EXTERNAL, or re-run the budget with real efficiency numbers ```
| Product | Recommendation | Reason |
|---|---|---|
| Wearable, BLE only | Internal chip or FPC | No protrusion; body loss dominates anyway |
| Handheld scanner, Wi-Fi + BLE | Internal FPC, dual-band | Ruggedness and form factor rule |
| Smart water meter in a pit | External, short lead-out or wall-mounted | The pit is the problem, not the antenna |
| LoRa sensor, 868 MHz, plastic box | Internal spring or PCB if the box is large; otherwise external stubby | λ/4 is ~8 cm — measure your box against it |
| Industrial gateway | External, N-type, mast or roof | Placement and low-loss feeder matter most |
| Global tracker (LTE + GNSS) | External for LTE; GNSS active external or a large internal patch | Band coverage plus placement |
| Vending machine / kiosk (metal) | External, magnetic mount on top | Metal enclosure |
1. Quoting the datasheet gain without the ground plane. Peak gain measured on a 100 mm reference plane is not your product's number. 2. Routing the feed cable across the antenna. It couples, skews the pattern and detunes. 3. Using U.FL as a service port. It is a ~30-cycle factory connector; field service needs SMA. 4. Forgetting the human body. A wearable measured on a bench and on a wrist are two antennas. 5. Sealing an external antenna at the connector only. Water enters at the bulkhead, the cable jacket and the antenna base — three separate paths. See the outdoor IP-rating guide.
Contact: Mr.Li
Phone: 0755-33169935
Tel: +86-755-33169935
Email: cs#antennahome.com (change # to @ )
Add: Build-2 201#, ShangXue Industry Park, Bantian, Longgang District, Shenzhen, Guangdong Province, China