PCB Antenna Integration: Keep-Out, Ground Plane and Tuning

The Antenna Is Part of the Board

An antenna printed on a circuit board is not a component that can be specified and forgotten. Its performance depends on the ground plane beside it, the plastic around it, the battery behind it and the hand holding the product. The same pattern of copper behaves differently in three different enclosures, which is why the antenna is usually the last thing to be tuned and the first thing to be blamed.

The advantage of the printed approach is cost and space. There is no part to buy, no soldering operation and no connector, and the antenna occupies area on a board that already exists. The disadvantage is that everything it needs has to be designed in from the start: the keep-out area, the ground plane reference, the feed impedance and the tuning.

The Common Types

• Inverted F antenna. A printed trace with a shorting stub, used at sub-gigahertz and at 2.4 gigahertz. It is compact, reasonably efficient and tolerant of the enclosure, and it is the default in most small wireless products.
• Meandered monopole. A folded trace that fits more electrical length into a small area, at the cost of bandwidth and efficiency. Used where the available space is very small.
• Chip antenna. A ceramic component a few millimetres long, soldered to the board like any other part. It needs a defined keep-out and a matching network, and it accepts the surrounding ground plane less well than a printed antenna.
• Patch antenna. A resonant copper area on a ground plane, used at microwave frequencies where a plane is needed underneath. Larger, directional and stable.
• Slot and notch antennas. Cut into a ground plane, used where the board edge is the only available space.

The Keep-Out Area

Every antenna defines an area around it that must contain no copper on any layer, no components, no battery and no metal. The size comes from the antenna design, and it is a hard constraint rather than a guideline.

Two mistakes are common. The first is placing the keep-out only on the top layer, when the ground plane on the inner layers is what actually detunes the antenna. The second is placing the battery or the metal shell of the product inside the keep-out volume in three dimensions, even though the board itself is clear.

The keep-out usually sits at a board edge, because an antenna works best with free space around it. Where the board is fully enclosed in metal, the antenna has to be moved outside the shield or brought to the surface through a connector, and that is a mechanical decision rather than a layout one.

The Ground Plane Is the Other Half

A small antenna on a handheld device is not really an antenna on its own; it is one half of a dipole whose other half is the ground plane and, at low frequencies, the user. That is why the ground plane size affects the bandwidth and the efficiency more than the antenna pattern does.

Three practical rules follow. Keep a solid ground plane under the electronics and as large as the product allows, because the current on it is part of the radiating structure. Do not split it into sections that are isolated from each other at radio frequencies. And keep the ground plane under the antenna genuinely clear, because a ground plane that approaches too close changes the resonance and the impedance.

The feed point is where the transmission line meets the antenna, and its impedance is usually not 50 ohms until it is tuned. A matching network of two or three components between the radio and the antenna brings the impedance to the value the radio requires, and it also provides a place to compensate for the effect of the enclosure during the tuning. Our notes on PCB design and layout cover the transmission line that feeds it.

Tuning in the Real Product

An antenna that measures perfectly on a bare board is almost never correct inside the finished product. The plastic of the housing, the battery, the display, the metal chassis and the user’s hand all shift the resonance, usually downwards, and the shift is often larger than the bandwidth of the antenna.

The practice is therefore to tune the antenna in the final mechanical assembly. A prototype housing is built, the antenna is measured with a network analyser through a temporary coax connection, and the matching network is adjusted until the return loss meets the requirement across the band. A second measurement is made with the product in the hand, or with a hand phantom, because the hand effect is part of the specification for a handheld device.

Where the shift cannot be corrected with the matching network, the antenna itself is changed: a longer meander, a wider trace or a different feed position. That is why a tuning iteration belongs in the schedule, and why the first prototype is not the production design.

Efficiency and Bandwidth

Two numbers describe the performance. The efficiency is the proportion of the power delivered to the antenna that is radiated rather than lost in the copper, the dielectric and the matching components; it is measured in a chamber or by a radiation pattern measurement. The bandwidth is the range of frequencies over which the antenna stays within the required return loss.

Both are limited by size. A small antenna has a narrow bandwidth and, in a product with little ground plane, a low efficiency; a larger antenna with more ground plane around it is better at both. Where the design needs a wide band, as a multi-band cellular product does, the antenna is either larger or a different structure is used.

Proximity loss matters as much as the antenna itself. A plastic housing with a high dielectric constant, or a conductive coating on the inside of a plastic cover, detunes and absorbs the signal. Where the housing must be coated for shielding, the coating is removed over the antenna area. Our notes on telecommunications PCBA describe the class of equipment these boards belong to.

Testing

The measurements that matter are the return loss across the band, the radiation pattern, the total radiated power and the receiver sensitivity. Return loss is the quickest and is used during tuning; the radiation measurements need a chamber and are used to confirm the design.

In production, a board level functional test with a conducted measurement through a temporary connector or a test pad is enough to confirm that the radio and the matching network are correct. A full radiated test on every unit is impractical, so it is performed on samples and on the first build of every revision. Our notes on PCB assembly describe the assembly, and our notes on internet of things PCBA cover the products this design serves.

Design Practice

• Place the antenna first. Decide the keep-out, the feed and the ground clearance before the rest of the layout exists.
• Keep the keep-out clear on every layer. Copper on an inner layer under the antenna is the most common cause of poor performance.
• Provide a matching network. Leave the pads for it even if the simulation says they are not needed, because the enclosure will change the answer.
• Test with a temporary feed. Bring the feed to a small pad or connector that allows a network analyser to be connected without altering the antenna.
• Measure in the enclosure, in the hand. The bench measurement without the product is not the specification.
• Allow for the plastic. Where the housing is coated or the plastic is thick, the antenna design has to account for it.

FAQ

How much board area does a PCB antenna need? A simple inverted F antenna at 2.4 gigahertz typically needs an area of roughly 15 by 25 millimetres plus its keep-out, and a sub-gigahertz antenna needs considerably more.

Can an antenna be placed inside a metal enclosure? Not usefully. The metal shields and detunes it, so the antenna is either placed outside the shield or brought to the surface through a connector.

Why does the antenna need tuning after the housing is fitted? Because the plastic, the battery and the user all shift its resonance, often by more than its bandwidth.

What is measured in production? A conducted test through a test pad on a sample basis, with full radiated measurements reserved for the first build of each revision.

Conclusion

A PCB antenna is a piece of the board that behaves as part of the whole product. Decide the keep-out and the feed first, keep the ground plane intact and large, provide a matching network and the pads to tune it, and measure in the finished enclosure with a hand present if the product is handheld. The antenna cost is a few square centimetres of copper, and the effort is spent on the mechanics and the enclosure rather than on the copper.