How to Design a Microstrip Patch Antenna: A Step-by-Step Engineering Guide
Microstrip patch antennas are among the most widely used antennas in wireless, radar, navigation, sensing, and automotive electronics. Their appeal is straightforward: they are low-profile, lightweight, inexpensive to manufacture, and easy to integrate with printed circuit boards or dielectric carriers. Their limitations are equally important: relatively narrow bandwidth, moderate gain, sensitivity to substrate properties, and performance that can change significantly with the feed, enclosure, ground plane, and nearby materials.
A successful patch antenna design therefore requires more than calculating the length and width of a copper rectangle. The calculation provides a starting point. The complete design process combines electromagnetic theory, transmission-line or cavity-model equations, feed selection, full-wave simulation, fabrication controls, and measurement.
1. Define the antenna requirements
Begin with the requirements, not the geometry. Establish the target resonant frequency, operating bandwidth, polarization, gain, efficiency, radiation direction, impedance, allowable size, and environmental conditions.
For a single rectangular patch, the most common starting point is the dominant TM10 mode. The antenna is typically linearly polarized and radiates broadside, meaning the strongest radiation is directed approximately normal to the patch plane. If circular polarization, dual polarization, wide bandwidth, beam steering, or multiple frequency bands are required, a conventional single patch may not be sufficient. Slots, stacked patches, parasitic elements, aperture coupling, or a different antenna architecture may be needed.
Also define the actual operating environment. A patch that performs well in free space may detune when installed above a vehicle roof, inside a plastic housing, near a battery, beside a display, or over a finite and irregular ground plane. Include the expected radome, bracket, adhesive, cable, connector, and nearby metal in the system-level requirements.
2. Select the substrate and conductor stack-up
The substrate strongly affects frequency, bandwidth, efficiency, mechanical stability, and manufacturability. Record the relative permittivity εr, dielectric loss tangent tanδ, thickness h, and expected tolerance over temperature and frequency.
A low-permittivity substrate generally produces a larger antenna with stronger fringing fields and often better bandwidth and efficiency. A high-permittivity substrate reduces the physical footprint but confines more energy in the dielectric and can increase loss and manufacturing sensitivity. Thickness is also a trade-off. Increasing h can improve bandwidth, but it may excite surface waves, increase spurious radiation, complicate fabrication, or make the antenna less compatible with the mechanical package.
Use the effective, frequency-relevant dielectric properties rather than relying only on a nominal catalog value. Automotive plastics, adhesives, foams, painted surfaces, and molded materials can have anisotropic or frequency-dependent properties. When accuracy matters, characterize the material or use verified supplier data.
Choose the conductor thickness and surface finish as well. At lower microwave frequencies, ordinary copper thickness may be adequate. At higher frequencies, conductor roughness and plating can affect loss. The ground plane should be treated as part of the antenna, not as an ideal infinite reference unless the design environment truly supports that assumption.
3. Calculate the initial patch width
For a rectangular patch designed for the dominant mode, calculate the initial width using
W = c / (2f0) √(2 / (εr + 1))
where W is the patch width, c is the speed of light, f0 is the target resonant frequency, and εr is the substrate relative permittivity.
The width is not the primary dimension controlling the half-wave resonance, but it affects radiation conductance, impedance, bandwidth, efficiency, and the overall aspect ratio. A wider patch often supports stronger radiation and lower quality factor, but it also occupies more area and can support unwanted modes if made excessively large.
This equation assumes a conventional rectangular patch and provides an initial value. It is not a final manufacturing dimension. The eventual width may need adjustment because of finite ground, housing materials, feed structure, and the exact electromagnetic model.
4. Calculate effective permittivity
The electric field is not completely confined to the substrate. It extends partly through the dielectric and partly into the surrounding air. The patch therefore behaves as though it sees an effective relative permittivity between 1 and εr. A common approximation is
εeff = (εr + 1)/2 + (εr − 1)/2 [1 + 12h/W]−1/2
Use εeff to account for the fringing-field effect on the guided wavelength. Because the field distribution changes near the patch edges, εeff is generally lower than εr.
This step is important because using εr directly in the resonance equation usually predicts a patch that is too short. The effective permittivity also changes with geometry, frequency, substrate thickness, and nearby materials, so it should be treated as an initial design parameter rather than an exact constant.
5. Calculate fringing extension and patch length
The patch resonates when its effective electrical length is approximately one-half of a guided wavelength. The physical patch is shorter than this electrical length because the electric field fringes beyond the physical edges.
First estimate the length extension at each radiating edge:
ΔL/h = 0.412 [(εeff + 0.3)(W/h + 0.264)] / [(εeff − 0.258)(W/h + 0.8)]
Then calculate the effective resonant length:
Leff = c / (2f0√εeff)
Finally, calculate the physical patch length:
L = Leff − 2ΔL
For the dominant TM10 mode, the two edges separated by L are the principal radiating slots. Their fringing electric fields are approximately in phase in the broadside direction, so their radiation combines constructively. The two side edges separated by W are often modeled as non-radiating magnetic walls in the simplified cavity model. In a real antenna, those edges can radiate weakly, especially when the geometry is asymmetric or higher-order modes are present, but they are not the primary radiating slots in the standard TM10 design.
6. Size the ground plane and surrounding clearance
A patch antenna is influenced by the ground-plane dimensions. Infinite-ground equations are useful for first-pass calculations, but a practical antenna uses a finite ground plane. As a starting rule, make the ground plane extend at least several substrate heights beyond the patch on every side. A commonly used first estimate is
Lg ≥ L + 6h
Wg ≥ W + 6h
where Lg and Wg are the ground-plane dimensions. More clearance may be required for stable patterns, high efficiency, or a specific installation.
The ground plane can be the PCB ground, a metal vehicle panel, a chassis, or a dedicated conductor. If the antenna is mounted over a large conductive surface, the effective design may be closer to the infinite-ground assumption. If it is mounted over a small or discontinuous ground, resonance, impedance, polarization purity, and pattern can all change.
7. Choose and design the feed
The feed determines how efficiently power is transferred into the patch. Common options include a microstrip edge feed, inset feed, coaxial probe, aperture-coupled feed, and proximity-coupled feed.
An edge-fed patch is simple and planar, but the input impedance at the radiating edge may be higher than the desired system impedance. An inset feed moves the feed point toward the center, where the input resistance is lower. A coaxial probe can provide flexible placement and is useful for prototypes or three-dimensional assemblies, but the probe adds inductance and can complicate manufacturing.
For an inset feed, use the feed depth as an optimization variable. A simplified relationship often used for the dominant mode is
Rin(y0) ≈ Redge cos²(πy0/L)
where Redge is the approximate resistance at the radiating edge and y0 is the inset distance measured from that edge. This relationship is only an approximation because the actual input impedance depends on the feed width, notch geometry, substrate, ground plane, solder geometry, and nearby materials.
Design the feed line for the required characteristic impedance, commonly 50 ohms. Include the actual line width, bend, launch, via transitions, connector, and reference plane in the model. If the feed enters through a connector, model the connector geometry or de-embed the measurement plane carefully.
8. Build the first electromagnetic model
Create a parameterized model with variables for L, W, h, εr, feed position, feed width, ground size, and any housing dimensions. A parameterized model is much more valuable than a fixed drawing because it lets you sweep the design and optimize it systematically.
Use a full-wave three-dimensional solver when possible. Define the metal conductivity, dielectric loss, radiation boundaries, ports, and symmetry conditions correctly. The air region must be large enough to prevent the boundary from artificially influencing the antenna. For a finite antenna, use an open or absorbing boundary with sufficient spacing from the radiating structure.
Start with a reasonable mesh, then refine it around the patch edges, feed transition, slot regions, vias, and dielectric interfaces. Fringing fields and narrow feed gaps can require much finer resolution than the broad patch surface. Always check mesh convergence by comparing resonant frequency, input impedance, and S-parameters as the mesh is refined.
9. Verify resonance, impedance, and radiation behavior
Run a frequency sweep around the target frequency. Inspect the input reflection coefficient, resonant frequency, impedance, current distribution, radiation pattern, gain, directivity, efficiency, and polarization.
The simulated resonance should be close to the target, but the location of the S11 minimum alone is not enough. A narrow and deep S11 notch can occur even when the radiation efficiency is poor. Confirm that accepted power is being radiated rather than dissipated in the dielectric, conductor, or numerical boundaries.
Review the surface current and electric-field animations. For a correctly excited TM10 mode, the surface current has a half-wave variation along the patch length and is relatively uniform along the width. Strong electric-field fringing should appear at the two radiating edges. The field orientation at those edges also indicates the polarization direction.
Check the radiation pattern for broadside behavior, sidelobes, back radiation, cross-polarization, and pattern distortion. In an installed system, repeat the analysis with the surrounding housing and nearby components present.
10. Tune the design in a controlled order
Avoid changing every dimension at once. A disciplined tuning sequence makes the design easier to understand and reduces the risk of compensating one error with another.
First tune L to move the resonant frequency. Increasing L generally lowers the resonant frequency; decreasing L generally raises it. Next tune the feed position or inset depth to improve impedance matching. Then adjust W to influence bandwidth, radiation conductance, and pattern. After that, refine the ground-plane size, feed-line dimensions, substrate representation, and enclosure details.
Use parameter sweeps before automated optimization. Sweeps reveal sensitivity and identify dimensions that have little effect. After the response is understood, use an optimizer with realistic bounds and a meaningful objective function. An objective might combine resonant-frequency error, minimum S11, bandwidth, efficiency, gain, and polarization constraints.
11. Fabricate a controlled prototype
The prototype should represent the final design as closely as possible. Control the substrate thickness, dielectric material, copper thickness, etching accuracy, feed dimensions, connector placement, solder volume, and mechanical stack-up.
Record actual dimensions rather than assuming nominal CAD values. A small error in patch length, inset depth, or substrate thickness can be significant when the antenna is electrically small or narrowband. Keep the first prototype simple enough to isolate the main antenna behavior. Add enclosure and vehicle integration in later builds, or test them in a controlled sequence.
12. Measure and correlate the results
Measure S11 or VSWR with a calibrated vector network analyzer. Calibrate to the appropriate connector or reference plane, and use a fixture that does not significantly alter the antenna. For a coax-fed design, account for cable movement and common-mode currents. For a PCB-fed design, define the calibration plane at the intended launch.
Compare measured and simulated resonance, impedance, bandwidth, and loss. If the measured resonance is lower than predicted, the antenna is often electrically longer than modeled because of excess length, higher effective permittivity, thicker dielectric, or nearby materials. If it is higher, the effective length may be too short or the dielectric loading may be lower than expected.
Next measure the radiation pattern, gain, efficiency, and polarization in an appropriate chamber or test range. Correlation is most useful when the same coordinate system, reference plane, mounting fixture, and operating configuration are used in simulation and measurement.
13. Document the final design
A complete design record should include the target requirements, material data, equations, nominal dimensions, tolerances, feed details, simulation setup, mesh-convergence evidence, S-parameter plots, radiation results, prototype photos, measurement setup, and correlation conclusions.
Document sensitivity to substrate thickness, dielectric constant, feed placement, patch length, ground-plane size, enclosure clearance, temperature, and assembly variation. This information supports design-for-manufacture decisions and helps distinguish a robust design from one that works only at nominal conditions.
Conclusion
Designing a patch antenna is a controlled progression from requirements to first-pass equations, feed selection, full-wave analysis, optimization, fabrication, and measurement. The analytical model provides an efficient starting point: calculate W, estimate εeff, account for ΔL, and determine L for the desired TM10 resonance. The electromagnetic model then captures the effects that simplified equations cannot fully represent, including finite ground planes, feed inductance, enclosure materials, surface waves, loss, and installation geometry.
The most reliable workflow is iterative but disciplined. Change one major variable at a time, inspect the fields rather than relying only on S11, validate material properties, and correlate simulation with a carefully controlled prototype. With that process, a simple copper patch becomes a repeatable engineering design rather than a trial-and-error exercise.
Keywords
patch antenna, microstrip antenna, rectangular patch antenna, antenna design, TM10 mode, radiating slot, non-radiating slot, fringing fields, effective permittivity, effective length, patch width, patch length, resonant frequency, guided wavelength, dielectric substrate, relative permittivity, substrate thickness, ground plane, finite ground plane, inset feed, coaxial probe feed, microstrip feed, antenna impedance, input impedance, S11, VSWR, return loss, antenna bandwidth, radiation pattern, broadside radiation, linear polarization, surface current, cavity model, transmission line model, electromagnetic simulation, full-wave simulation, HFSS, CST Studio Suite, antenna optimization, mesh convergence, radiation efficiency, antenna gain, directivity, cross polarization, prototype fabrication, vector network analyzer, VNA calibration, chamber measurement, antenna correlation, feed-point design, wireless antenna, RF engineering, automotive antenna
Comments