Measure a waveguide antenna’s radiation pattern by recording its received field or power versus angle at a stated frequency and polarization. Use a validated far-field range, or collect suitable near-field data and apply the corresponding transformation. A normalized pattern shows beam shape; absolute gain needs a separate amplitude reference and the appropriate corrections.
Choose the measurement geometry
For a direct far-field scan, keep a fixed range antenna and rotate the antenna under test (AUT) through the required angles. A VNA can record complex S21 between the antennas, while a suitable source and receiver can measure power-only patterns. The full RF path, alignment and measurement range must remain controlled during rotation.
Near-field scanning is a different acquisition process. Planar, cylindrical and spherical scans use different geometries and transformations; the scan extent, sampling and probe correction affect the recovered far-field result. Applying an arbitrary Fourier transform to a few azimuth or elevation cuts does not create a valid 3D pattern.[1]
Use the range’s validated distance and quiet-zone requirements. The common estimate R ≥ 2D²/λ is a starting point, not a complete uncertainty assessment. For D = 0.15 m at 10 GHz, λ is approximately 0.03 m and the estimate is 1.5 m. A particular range may need more distance; 5 m or 5.8 m is not a universal requirement for that aperture.
Define coordinates and polarization before rotating
- Axes: identify the antenna’s mechanical boresight, positive rotation direction, azimuth/elevation convention and mounting orientation.
- Rotation center: align it with the reference point required by the range procedure. Record any offset and compensation.
- Principal planes: for a linearly polarized antenna, identify the E- and H-plane cuts from the field orientation, not simply from “horizontal” and “vertical” labels in the room.
- Polarization basis: state how co-polar and cross-polar components are defined, especially away from boresight or for circular polarization.
Keep the feed and moving cable from pulling on the antenna. Follow the flange assembly drawing for fasteners and torque; waveguide size alone does not specify a universal tightening value. A changed connection can alter both amplitude and phase.
Separate feed calibration from range calibration
VNA calibration corrects the electrical test path at stated reference planes. It does not by itself remove chamber reflections, positioning error or the range antenna’s response. Use the waveguide calibration workflow for the feed interfaces, then apply the antenna range’s calibration procedure.
For absolute gain, use a characterized reference antenna with data at the test frequencies. When replacing it with the AUT, account for differences in position, feed loss and mismatch. A standard gain horn must have suitable reference data; its catalog designation alone does not establish measurement uncertainty.
Check the source-on and source-off receiver levels and the range’s reflection limits. The available dynamic range must cover the lowest sidelobe or cross-polar signal that the test is intended to resolve.
Choose angular and frequency sampling that resolves the beam
Start with the expected beamwidth and required angular accuracy. A fixed 5° step can miss a narrow peak and cannot resolve a beam only a few degrees wide. Refine the scan until reducing the step no longer changes the reported metrics beyond the required tolerance.
For a planning example, a 4° beam sampled every 0.25° has 16 intervals across that width. This can be a useful initial grid, but the needed spacing still depends on sidelobes, nulls and the range procedure. Do not claim 0.1° beamwidth accuracy from sparse data merely because the plot interpolates between points.
Measure the angular sector required by the specification. A scan around the main beam cannot establish a front-to-back ratio unless the back direction is also measured. Two principal-plane cuts are useful for inspection but do not measure the complete radiation sphere.
Repeat scans at the band edges and other required frequencies. For broadband antennas, add frequencies where the beam direction, gain or match changes. The broadband antenna test guide combines these scans with input-match and gain measurements.
Normalize the trace without confusing dB and dBi
For a relative power pattern, calculate:
Pnorm(θ) = 10 log10[P(θ)/Ppeak].
If the receiver already reports power in dBm, subtract the peak dBm value from each reading. If using linear S21 magnitude, use 20 log10(|S21(θ)|/|S21|peak). Subtracting the peak from an S21 trace already expressed in dB gives the same normalized result.
Label the peak “0 dB relative,” not “0 dBi.” Normalization discards the absolute level. Retain the raw calibrated data separately so absolute gain and frequency-to-frequency variation can still be evaluated.
For cross-polar discrimination, compare co- and cross-polar levels at the same frequency and direction using the calibrated channel relationship. Do not independently normalize both traces to their own peaks; that hides their level difference.
Extract beam metrics with explicit definitions
| Metric | How to extract it | What to state |
|---|---|---|
| Peak direction | Angle of the main-beam maximum | Coordinate convention and whether interpolation was used |
| Half-power beamwidth | Separation between the main-beam half-power crossings, approximately −3 dB | Plane, frequency and sampling resolution |
| Sidelobe level | Sidelobe maximum relative to the main-beam peak | First sidelobe or maximum over a defined angular sector |
| Front-to-back ratio | Forward level minus the defined rear level | Exactly 180° or maximum in a stated rear sector |
| Cross-polar discrimination | Co-polar minus cross-polar level | Polarization basis, direction and measurement floor |
As a geometric example, half-power crossings at −2.2° and +2.4° give a 4.6° beamwidth. The midpoint is +0.1°, but it need not equal the actual peak direction of an asymmetric beam. Report the peak from the measured trace.
Add absolute gain only when the reference supports it
In a gain-substitution measurement with equivalent geometry and corrected paths, the AUT gain equals the reference gain plus the measured level difference. For an illustrative calculation, a 20.0 dBi reference with S21 = −45.2 dB and an AUT with S21 = −44.0 dB gives 21.2 dBi before any remaining correction terms. The number is valid only if the reference quantity, mismatch treatment and path conditions are consistent.
NIST’s antenna measurement work distinguishes measured pattern levels from gain corrected for mismatch; that distinction is why the report must state whether the output is gain or realized gain.[2]
Check repeatability and deliver usable data
Repeat a critical cut after returning the positioner to zero. A disagreement can expose cable movement, backlash, drift or alignment changes. Investigate unexpected ripple by controlled changes to the range geometry; reflections can imitate antenna features.
Do not smooth a failing sidelobe away or assign an exact value to a null below the noise floor. Record any averaging, time gating, interpolation or filtering used in the result. NIST’s uncertainty guidance includes analytical estimates, simulations and controlled changes to the measurement system as ways to assess contributions.[3]
Deliver frequency, angle, polarization, complex response where measured, calibrated gain where available, and the normalization reference. Include the antenna serial number, mounting diagram, range geometry, reference antenna, correction files and uncertainty statement.
Send Dolph the required band, polarization, scan planes and beam limits when requesting an antenna quotation, and specify whether delivery needs normalized cuts, absolute gain data or a complete 3D measurement.
