Microstrip Patch Antenna Design: Calculator to Copper

Why Patch Antennas Are Everywhere

Open up any WiFi router, GPS receiver, or 5G phone and you'll find at least one microstrip patch antenna. They're flat, cheap to manufacture, and integrate directly onto PCBs. The tradeoff? They're narrowband and require careful dimensioning to hit your target frequency.

The basic idea is simple: a rectangular copper patch sits above a ground plane, separated by a dielectric substrate. The patch acts as a resonant cavity, and the dimensions determine the operating frequency. Get the length wrong by a few percent and your antenna's resonance shifts enough to tank your return loss.

The Physics in Brief

A rectangular patch antenna radiates from its two "radiating edges" — the edges perpendicular to the feed direction. The patch length $L$ is approximately half a wavelength in the substrate material, accounting for fringing fields that make the effective length slightly longer than the physical length.

The resonant frequency relates to the patch length by:

where $L_{eff}$ is the effective length (physical length plus fringing extension), $\\varepsilon_{eff}$ is the effective dielectric constant, and $c$ is the speed of light.

The effective dielectric constant accounts for the fact that some field lines travel through air above the patch, not just through the substrate. For a microstrip structure:

The patch width $W$ affects the radiation efficiency and input impedance. A wider patch radiates more efficiently but also changes the effective dielectric constant. There's a practical optimum that balances efficiency against the impedance matching challenge.

Worked Example: 2.4 GHz WiFi on FR4

Let's design a patch antenna for 2.4 GHz WiFi using standard FR4 substrate. This is probably the most common patch antenna design scenario — cheap substrate, well-understood frequency.

Given parameters:

• Frequency: 2.4 GHz
• Substrate: FR4, $\\varepsilon_r = 4.4$
• Substrate height: 1.6 mm (standard thickness)

Step 1: Calculate patch width

The width for good radiation efficiency is:

Working through: $W = 0.0625 \times 0.609 = 38.1$ mm

Step 2: Calculate effective dielectric constant

With $W = 38.1$ mm and $h = 1.6$ mm:

Step 3: Calculate fringing extension

The fringing fields extend the effective length by $\\Delta L$ on each edge:

Plugging in: $\\Delta L = 0.412 \times 1.6 \times \frac{(4.09 + 0.3)(23.8 + 0.264)}{(4.09 - 0.258)(23.8 + 0.8)} = 0.74$ mm

Step 4: Calculate patch length

So our patch dimensions are roughly 38 mm × 29 mm. The edge-fed impedance will be around 200-300 Ω, which is why you need either an inset feed or a quarter-wave transformer to match to 50 Ω.

When you open the Microstrip Patch Antenna Calculator, you'll get these values instantly, plus the edge feed impedance and expected gain. The calculator also handles the feed offset calculation for inset feeding, which saves you from iterating through the matching network math.

Substrate Selection Matters More Than You Think

FR4 is cheap and available, but it's honestly not great for antennas above 3-4 GHz. The loss tangent (typically 0.02) eats into your efficiency, and the dielectric constant varies enough lot-to-lot that your resonant frequency can shift by 2-3%. For production WiFi products, most engineers accept this and tune during prototyping.

Rogers RO4003C ($\\varepsilon_r = 3.55$, loss tangent 0.0027) costs maybe 10× more but gives you consistent results and much better efficiency at 5.8 GHz and above. For 28 GHz 5G mmWave work, you basically have no choice — FR4 losses would kill your link budget.

The lower dielectric constant of Rogers also means larger patch dimensions for the same frequency, which can actually help with manufacturing tolerances. A 28 GHz patch on FR4 would be tiny and incredibly sensitive to etching variations.

Common Mistakes and Gotchas

Ignoring the ground plane size

The calculator assumes an infinite ground plane, which is a reasonable approximation if your ground extends at least λ/4 beyond the patch edges. Skimp on ground plane and your pattern distorts, your impedance shifts, and your back-lobe radiation increases. I've seen engineers try to squeeze a 2.4 GHz patch onto a 40 mm × 40 mm board and wonder why their return loss is 6 dB instead of 15 dB.

Using the wrong dielectric constant

Datasheet values for $\\varepsilon_r$ are typically specified at 1 MHz. At 2.4 GHz, FR4 is closer to 4.2-4.3 than 4.4. At 5.8 GHz, it might be 4.0. This frequency dependence isn't huge, but it's enough to shift your resonance by 50-100 MHz if you're not careful. For critical designs, measure your actual substrate or get frequency-specific data from the manufacturer.

Forgetting about the feed

The calculator gives you edge feed impedance, which is typically 150-300 Ω depending on your geometry. You can't just run a 50 Ω microstrip line to the patch edge and expect good matching. You need either:

• An inset feed (notch cut into the patch) to reach a 50 Ω point
• A quarter-wave transformer
• A probe feed from below

The feed offset parameter in the calculator helps you find the inset depth for 50 Ω matching. This is iterative in practice — the calculator gives you a starting point, but you'll likely need to tune in simulation.

Assuming the gain number is what you'll actually get

The nominal gain output (typically 5-7 dBi for a single patch) assumes perfect matching, no substrate losses, and an infinite ground plane. Real-world gain is usually 1-2 dB lower. For FR4 at 5.8 GHz, subtract another dB for dielectric losses. The number is useful for link budget estimates, but don't promise it to your customer.

Not accounting for solder mask

Solder mask over the patch adds another dielectric layer that shifts your resonant frequency downward. It's a small effect (maybe 1-2%), but at mmWave frequencies that's enough to matter. Either keep the patch area clear of solder mask or account for the effective dielectric change.

When Patch Antennas Are the Wrong Choice

Patch antennas are great for planar integration and hemispherical coverage, but they're narrowband — typically 2-5% bandwidth. If you need to cover both 2.4 GHz and 5.8 GHz WiFi bands with one antenna, a patch won't do it. You'd need a wideband design like a Vivaldi or a dual-band structure with separate radiating elements.

They're also inefficient when the substrate is electrically thin. The radiation resistance drops as substrate height decreases, which means more power lost to conductor and dielectric losses. For really thin substrates (< 0.01λ), you might get only 50% efficiency even with low-loss materials.

And if you need omnidirectional coverage, a patch is fundamentally wrong — they radiate broadside with a null along the ground plane. Use a monopole or dipole instead.

Try It

The math above is straightforward but tedious to repeat for every design iteration. Open the Microstrip Patch Antenna Calculator to get instant dimensions for your frequency and substrate combination. Plug in your parameters, get the patch width and length, check the edge impedance, and use the feed offset to plan your matching approach. Then take those numbers into your EM simulator for final optimization — the calculator gets you 90% of the way there, but nothing replaces a full-wave simulation for production designs.