
Why Antenna Design Mistakes Are Expensive
Most antenna design mistakes shift resonant frequency, degrade gain, or distort radiation patterns—problems that rarely surface until a prototype is measured in a chamber. Because the antenna is usually the last subsystem verified, a single error can force a full board respin costing four to eight weeks of schedule. The twelve mistakes below account for the majority of antenna-related respins across Wi-Fi, GNSS, UWB, cellular IoT, and radar bands. Each entry explains the root cause, the symptom you will see on a VNA or in a pattern measurement, and a concrete fix.
Mistake 1: Undersized or Missing Ground Plane

A PCB patch or monopole antenna uses the ground plane as part of its radiating structure. Shrinking the ground below roughly 0.25 λ on each side of the antenna element shifts the resonant frequency downward and tilts the pattern off-axis. For a 2.4 GHz patch, that means the ground should extend at least 31 mm (about 1.22 in) beyond the patch edge in every direction. If board size is constrained, simulate the actual ground dimensions and retune the element—do not rely on an ideal-ground datasheet pattern.
Mistake 2: Ground-Plane Slots Under the Antenna Feed
Routing digital or power traces on inner layers can create unintentional slots in the reference plane directly beneath the antenna feed line. A slot as narrow as 0.2 mm (8 mil) crossing a 50 Ω microstrip at 5.8 GHz introduces 2–4 dB of return-loss degradation and radiates common-mode noise. The fix is straightforward: keep a continuous, unbroken copper pour under every RF trace from the transceiver pad to the antenna element, and route other signals around this exclusion zone.
Mistake 3: Wrong Substrate Dk Assumption
Antenna dimensions scale with the effective dielectric constant (Dk). Using a nominal Dk of 4.4 for FR-4 when the actual value at 2.45 GHz is closer to 4.2 (or even 4.0 for some glass-weave styles) shifts the patch resonance by 50–100 MHz. For tighter designs, specify the laminate grade and request the manufacturer’s Dk at your operating frequency. High-frequency laminates such as Rogers RO4350B (Dk 3.48 ± 0.05 at 10 GHz) offer tighter tolerance, but only if you use the datasheet value at the correct frequency and account for the Rogers Dk-vs-frequency curve rather than the 1 MHz test value.
Mistake 4: Ignoring Feed-Line Impedance Transformation
A 50 Ω coaxial connector does not guarantee a 50 Ω interface at the antenna element. The microstrip trace width needed for 50 Ω depends on substrate thickness and Dk—on 0.8 mm FR-4 it is roughly 1.5 mm, but on 0.254 mm RO4350B it is about 0.54 mm. A mismatch of even 5 Ω at the feed point reflects enough power to raise VSWR above 2:1 and cut delivered power by 10 %. Always verify the feed-line characteristic impedance with a field solver, not a hand-sketch estimate.
Mistake 5: Poor Transition From Coax to PCB
The coaxial-to-microstrip launch is one of the most common sources of parasitic inductance. An SMA footprint with a signal via longer than 0.5 mm at 6 GHz adds enough inductance to shift S11 by several hundred MHz. Use a flush-mount edge-launch connector wherever possible, or add ground vias within 0.25 mm of the signal via to form a coaxial-like transition. Detailed antenna feed cable effects and common-mode leakage deserve their own analysis when measurement accuracy matters.
Mistake 6: Uncontrolled Keep-Out Zone
Components, vias, and copper fills placed inside the antenna’s near-field zone (typically within λ/4 of the radiating edge) detune the element and scatter the pattern. A 0402 decoupling capacitor sitting 3 mm from a 5.8 GHz patch edge can shift resonance by 80 MHz. Define a component-free and via-free keep-out zone in your layout constraints and enforce it with a DRC rule, not just a silkscreen outline.
Mistake 7: Neglecting Mutual Coupling in Arrays
Placing array elements closer than 0.5 λ center-to-center increases mutual coupling, which shifts each element’s input impedance and distorts the composite pattern. At 77 GHz (λ ≈ 3.9 mm), even 0.1 mm of placement error changes coupling by 1–2 dB. Full-wave simulation of the entire array—not just a single isolated element—is non-negotiable. For millimeter-wave radar arrays, a systematic radar design review checklist catches coupling issues before fabrication.
Mistake 8: Solder-Mask Over the Antenna Element
Solder mask has a Dk of roughly 3.3–3.8 and a thickness of 15–25 µm. Leaving it over a patch element adds a dielectric loading layer that shifts resonance downward by 20–60 MHz at 2.4 GHz. Either specify a solder-mask opening (sometimes called a “mask relief”) over the entire antenna copper, or include the mask layer in your EM simulation. This is especially critical for compact chip-antenna matching networks where every pF counts.
Mistake 9: Via Fencing Done Wrong

Via fences around a patch antenna or along a feed line are meant to suppress surface-wave modes and improve isolation. But spacing the vias farther apart than λ/10 at the highest frequency of interest lets energy leak between them, defeating the purpose. At 24 GHz (λ ≈ 12.5 mm), via pitch must stay below 1.25 mm. Conversely, placing vias too close to the radiating edge (inside the fringing-field region) detunes the patch. A safe starting rule: fence vias at λ/20 pitch, at least λ/10 from the nearest radiating edge.
Mistake 10: Forgetting Radome and Enclosure Effects
A plastic housing 2 mm above a 2.4 GHz patch (ABS, Dk ≈ 2.7) shifts resonance by 30–80 MHz and can reduce peak gain by 1–2 dB depending on thickness and spacing. The antenna must be designed and tuned with the radome in place—either by including it in simulation or by tuning on the final mechanical prototype. Designing in free space and hoping the enclosure “won’t matter” is one of the most common antenna design mistakes in consumer products.
Mistake 11: Incorrect Polarization Alignment
A linearly polarized patch oriented 90° off from the intended link polarization suffers a theoretical infinite cross-pol loss (practically 15–25 dB). In GNSS RTK receiver designs, using a linearly polarized antenna instead of right-hand circular polarization (RHCP) costs 3 dB of link margin and degrades multipath rejection. Always verify the polarization requirement from the system link budget before choosing the antenna topology.
Mistake 12: No Post-Layout EM Verification
Relying solely on schematic-level S-parameter models and analytical formulas without running a 3-D electromagnetic simulation of the final layout—including the actual stackup, ground shape, nearby components, and connectors—is a gamble. Tools such as Altium’s integrated EM solver or standalone solvers like HFSS and CST catch issues that lumped-element models cannot represent. Budget at least one full EM simulation cycle before releasing Gerbers. Verifying antenna efficiency through measurement after fabrication closes the loop.
Quantifying the Cost of a Respin
Each of the mistakes above can trigger a board respin. To decide whether investing time in a fix now (simulation, layout rule, better laminate) is worthwhile versus risking a respin later, estimate the return on investment of the design change. Enter your prototype cost, team hourly rate, and expected respin delay below to see the break-even point.
For a typical four-layer RF board with a 10-day prototype cycle and a three-person team, a single avoided respin saves roughly $3,000–$8,000 in direct costs and two to three weeks of calendar time. Even an afternoon spent on EM simulation or a $200 laminate upgrade pays for itself many times over if it prevents one iteration.
Antenna Design Checklist
Use the interactive checklist builder below to generate a project-specific verification list. Select the antenna type (patch, monopole, array, chip), frequency band, and substrate, and the tool compiles the relevant checks from the twelve mistakes discussed above—plus additional items for your combination.
Print or export the checklist and walk through it at two gates: once after schematic/simulation sign-off, and again after layout but before Gerber release. Teams that enforce a two-gate review catch 80–90 % of the errors in this article before they reach fabrication. For complex UWB positioning designs with tight antenna-delay requirements, add a third gate after initial VNA measurements.
Quick-Reference Summary Table
| # | Mistake | Typical Impact | Key Fix |
|---|---|---|---|
| 1 | Undersized ground plane | Frequency shift, pattern tilt | Extend ground ≥ 0.25 λ beyond element |
| 2 | Ground-plane slots under feed | 2–4 dB return-loss hit | Continuous pour under all RF traces |
| 3 | Wrong substrate Dk | 50–100 MHz resonance shift | Use Dk at operating frequency, not 1 MHz |
| 4 | Feed-line impedance error | VSWR > 2:1 | Field-solver verification of trace width |
| 5 | Poor coax-to-PCB transition | Parasitic inductance, S11 shift | Edge-launch connector, ground vias |
| 6 | Uncontrolled keep-out zone | 80 MHz detuning at 5.8 GHz | DRC-enforced exclusion zone |
| 7 | Array mutual coupling ignored | Pattern distortion, impedance shift | Full-array EM simulation |
| 8 | Solder mask over element | 20–60 MHz shift at 2.4 GHz | Mask relief or include mask in sim |
| 9 | Via fence pitch too wide | Surface-wave leakage | Pitch ≤ λ/10, offset ≥ λ/10 from edge |
| 10 | Radome/enclosure not modeled | 1–2 dB gain loss | Simulate and tune with housing in place |
| 11 | Polarization mismatch | 3–25 dB link loss | Match polarization to link budget |
| 12 | No post-layout EM verification | Undetected compound errors | Mandatory 3-D EM sim before Gerber release |
FAQ
What is the single most common antenna design mistake on PCBs?
An undersized or poorly shaped ground plane. It affects every antenna type that relies on an image plane—patches, monopoles, IFAs, and chip antennas—and its effects are hard to fix without a respin because the ground is tied to the board outline.
Can simulation completely replace antenna measurement?
No. Simulation catches geometry and material errors before fabrication, but manufacturing tolerances, solder joints, and enclosure fit introduce variables that only measurement can verify. Use simulation to get within 90 % of target, then confirm with a VNA and, ideally, a pattern measurement.
Does FR-4 work for PCB antennas above 3 GHz?
FR-4 can work up to about 6 GHz for cost-sensitive designs if you accept higher dielectric loss (Df ≈ 0.02 at 5 GHz) and wider Dk tolerance. Above 6 GHz, the loss and Dk spread usually make a low-loss laminate (PTFE-based or hydrocarbon ceramic) necessary for acceptable efficiency.
How do I know if my antenna keep-out zone is large enough?
Start with a λ/4 clearance from every radiating edge and feed point. Run an EM simulation with and without nearby components; if gain or S11 changes by more than 0.5 dB or 1 dB respectively, enlarge the zone or relocate the offending part.
How many EM simulation iterations should I budget?
For a single-element antenna on a known substrate, two to three iterations (initial, tuned, final with housing) are typical. For a phased array or multi-band design, plan for five or more. Each iteration takes minutes to hours depending on mesh size and solver, but the time is trivial compared to a physical respin.
Ready for a quotation?
Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.
