RO4350B Antenna PCB Design: Choosing a Validation Workflow
Validate the antenna as a complete installed RF structure. Start with a preliminary geometry, refine it with an electromagnetic model of the real construction, measure a prototype at defined reference planes, and repeat the relevant tests in the product enclosure. A good input match is one result; it does not establish radiation efficiency, gain, polarization or coverage.
This workflow extends the patch-antenna starting guide with review gates and a worked mismatch example. It is intended for engineers planning a prototype and its acceptance evidence. The examples are hypothetical, not measured performance of a RO4350B antenna.
1. Freeze the questions before optimizing dimensions
Write a short validation brief containing the operating band, polarization, available board area, ground outline, feed arrangement and mechanical environment. State the required radiation behavior in the useful directions, not only the frequency of the deepest S11 minimum. For an array, include scan states, element excitation and the phase/amplitude assumptions used in the model.
Separate the bare-board development goal from the installed-product goal. A connector-fed prototype may be convenient to measure but may not reproduce the final radio feed. Define which launches, matching parts, cables and transitions belong inside each result. Keep the reference planes visible on a drawing so simulation and measurement can be compared consistently.
Specify how close a measured result may be to a limit before uncertainty requires further investigation. An antenna that nominally touches the edge of a requirement has not demonstrated manufacturing margin. Select sample counts and environmental conditions according to the product's risk and validation plan; this article does not prescribe a universal qualification sample size.
2. Choose the construction and initial geometry together
The Rogers RO4000 datasheet distinguishes process Dk from design Dk, with different methods and frequency conditions. Use an appropriate material model rather than substituting one value for the other without justification. Document substrate thickness, foil, conductor thickness, surface finish and any mask or cover layer in the antenna region.
A half-wavelength estimate is a scale check, not a fabrication drawing. Fringing fields, feed loading and the finite ground can move the resonance. RO4350B is a candidate when its construction meets the relevant loss, dimensional and environmental requirements; it is not automatically needed for every low-frequency product or suitable for every millimeter-wave antenna.
3. Use staged evidence instead of one pass/fail plot
| Gate | Evidence to save | What remains unproven |
|---|---|---|
| Initial geometry | Band, assumptions, dimensions and construction revision | Final resonance, bandwidth and radiation pattern |
| EM model | Ports, boundary settings, mesh-convergence results and material basis | Actual fabricated dimensions and assembly effects |
| Prototype input test | Calibrated complex S11 with reference plane and setup photograph | Radiation efficiency and spatial coverage |
| Radiation test | Gain/pattern or efficiency results with method and uncertainty | Performance in an untested enclosure or orientation |
| Installed test | Mechanical configuration, operating states and repeated measurements | Unmeasured production variation or lifetime durability |
| Release | Approved drawing, tests, limits and change-control record | Compatibility of an unreviewed material substitution |
In the EM stage, include the feed, finite ground and nearby conductors that are expected to influence the antenna. Check that further mesh refinement does not materially change the decision-driving outputs. Save more than a favorable resonance plot: compare bandwidth and radiation behavior at the band edges as well as the center.
Run sensitivity cases for plausible dimensional and material variations agreed with the fabricator. A sweep is a model study, not evidence that the process achieves the swept limits. If a feature is critical, identify its finished dimension and how it will be inspected. For arrays, use the phase-matching guide to define electrical rather than merely physical length requirements.
4. Establish the input measurement reference plane
Rohde & Schwarz separates antenna impedance measurements from radiation measurements. Its VNA guidance describes a reflection measurement and calibration at the antenna connection, including the end of a test cable when one is used. Follow the instrument and calibration-kit procedures, and state whether a PCB launch remains inside the measured device or is removed through a validated fixture method.
Keep the sweep range, point count, source power and relevant analyzer settings with the result. Record cable routing and support fixtures, then check repeatability after disconnecting and reconnecting the prototype. Investigate changes caused by moving the cable or nearby objects before calling a small resonance shift a laminate effect.
Worked example: why a better match is not the whole answer
For a passive single-port antenna with a real reference impedance, let the reflection magnitude be |Γ|. Reflected power fraction is |Γ|² and accepted power fraction is 1 − |Γ|². With S11 expressed as 20 log10|Γ|, a value of −10 dB gives 10% reflected power and 90% accepted power. The corresponding positive return loss is 10 dB.
Now assume, solely for illustration, radiation efficiency of 60% relative to accepted power, excluding any separate upstream feed loss. Total efficiency relative to incident power is (1 − |Γ|²) × 0.60. Improving S11 to −20 dB changes that total from 54.0% to 59.4% if radiation efficiency stays unchanged. It does not turn the antenna into a 99%-efficient radiator.
| S11 | Reflected fraction | Accepted fraction | Total efficiency |
|---|---|---|---|
| −10 dB | 10% | 90% | 54.0% |
| −20 dB | 1% | 99% | 59.4% |
These equations describe power bookkeeping for the stated single-port case. They do not predict directivity, polarization or a particular link range. In a coupled array, excitation and active impedance complicate the interpretation; do not apply an isolated-element S11 result as a complete array efficiency assessment.
5. Close the loop in the actual product
Choose a radiation measurement method that can answer the requirements, and record its calibration, reference antenna where applicable, geometry, frequency range and uncertainty. Agree with the test facility on whether reported gain includes mismatch and which losses have been corrected. Comparing a simulated directivity with measured realized gain without reconciling definitions can conceal the real source of a difference.
Repeat the relevant measurements with the enclosure, mounting hardware, battery and intended cable arrangement. Define meaningful operating states and orientations. If installed performance changes, compare one controlled configuration at a time rather than simultaneously tuning the radiator, matching network and housing. Preserve each revision so improvement in one metric does not hide degradation in another.
Before release, connect each acceptance requirement to a drawing revision and a test result. Revisit the evidence after changes to dielectric spacing, foil, antenna features, mask, matching parts or mechanical boundaries. Manufacturing approval and antenna validation are related activities, but one cannot replace the other.
Primary sources and scope
- Rogers RO4000 laminate datasheet — material definitions and test conditions.
- Rohde & Schwarz: VNA antenna measurement — impedance versus radiation measurements and calibration location.
Sources reviewed 14 October 2026. The workflow table and calculation are original editorial work. No antenna measurements, certifications or manufacturer endorsement are claimed.