Introduction to Flexible and Conformal Antennas

Flexible and conformal antennas have transitioned from laboratory curiosities into essential components of next-generation wireless systems. By using substrates made of polymers, textiles, paper, or thin flexible ceramics, these antennas can bend, fold, twist, or wrap around curved surfaces without significant damage. Applications span wearable health monitors, soft robotics, aircraft skin antennas, in-vehicle communication systems, and 5G/6G millimeter-wave arrays that must adapt to non-planar platforms. As the Internet of Things (IoT) expands into every corner of industrial and consumer environments, the need for antennas that conform to irregular shapes—rather than force those shapes to accommodate rigid radiators—has become a fundamental design requirement.

The defining advantage—mechanical adaptability—directly influences the electrical behavior that designers quantify through scattering parameters (S-parameters). S-parameters describe how a signal reflects, transmits, or couples between ports of a multi-port device. For a single-port antenna, S11 (return loss) indicates how well the antenna matches its feed transmission line, while for multi-element arrays, S21, S31, etc. reveal mutual coupling. Accurate S-parameter data is the foundation for deriving impedance, bandwidth, efficiency, and radiation pattern estimates. When the antenna deforms, its current distribution shifts, altering the very S-parameters that the design simulated on a flat model. Understanding and overcoming the obstacles in measuring these parameters is therefore central to producing reliable, real-world flexible antenna systems.

Why Accurate S-Parameter Measurement Matters

At first glance, measuring S-parameters seems straightforward with a vector network analyzer (VNA). Yet for conformal antennas, small inaccuracies in S11 or coupling terms can cascade into large errors when computing link budgets, total radiated power, or specific absorption rate (SAR) compliance. An antenna that appears well-matched at 2.45 GHz on a flat bench may detune drastically when curved around a wearer's arm, causing reflection losses that degrade communication range and battery life. For medical implants or aerospace systems, such performance shifts can translate into critical mission failures or patient safety risks.

Furthermore, many flexible antennas are narrowband by design, so tiny frequency shifts of a few megahertz—often induced by mechanical stress—can push the operating band out of specification. In phased arrays or MIMO systems, mutual coupling measured via S21 and S31 affects beamforming accuracy and channel capacity. Without reliable measurement protocols that replicate operational conditions, designers risk overstating performance or missing failure modes. Thus, the ability to characterize S-parameters under bending, stretching, or mounting on real surfaces is what separates a laboratory prototype from a field-deployable product. The economic implications are significant: a poorly characterized flexible antenna can lead to costly redesigns, field failures, and lost revenue in competitive markets like consumer electronics or defense.

Unique Measurement Challenges

Mechanical Deformation and Shape Sensitivity

The most immediate hurdle is maintaining a known and repeatable deformation state during a VNA sweep. Unlike rigid antennas, a flexible element can subtly change curvature if the test fixture applies non-uniform pressure or if the antenna material creeps over time. Even a few degrees of bending radius variation can shift the resonant frequency by several percent. For stretchable antennas fabricated with liquid metal or serpentine meshes, elongation changes not only the physical dimensions of the radiator but also the dielectric properties of the elastomeric substrate—both altering S-parameters in nonlinear ways. Capturing data across a full set of bend radii and stretch ratios requires mechanical fixturing that can apply precise displacement without introducing metallic structures that couple to the antenna. Motorized stages with closed-loop position control and soft, RF-transparent contact surfaces are becoming standard, but they add cost and complexity to the measurement setup. Advanced systems now incorporate capacitive or optical sensors to verify the actual curvature during the sweep, creating a feedback loop that compensates for any drift.

Connectorization and Feeding

Many flexible antennas use microstrip or coplanar waveguide feeds on thin, soft substrates. Attaching a standard coaxial connector can stiffen a region locally, creating a stress concentration that modifies the antenna's boundary conditions. Soldered connectors may crack under repeated bending, while conductive epoxy or mechanical pressure contacts can introduce unpredictable parasitic reactance if not kept under constant force. In wearable devices, the antenna feed often transitions to a flexible transmission line, and the junction between the VNA cable and that line becomes a variable impedance mismatch if the cable routing changes between calibration and measurement. Non-contact probing methods such as near-field probes or optical waveguide feeds can circumvent some of these problems but demand advanced calibration techniques to move the reference plane to the antenna input. A common workaround is to include a short flexible extension cable that remains fixed during calibration and measurement, but that cable must itself be characterized under the same bend state. Emerging solutions use inkjet-printed coaxial lines that are intrinsically flexible, allowing the connector to be placed far from the radiating element, thereby reducing stress on the antenna itself.

Test Fixture Design

Conventional antenna test fixtures rely on rigid mounting brackets, dielectric spacers, and positioning stages that assume planar geometries. For conformal antennas, a fixture must replicate the intended mounting surface—whether it is a cylindrical phantom limb, an aircraft fuselage curvature, or a curved windshield. The materials used in the fixture should be RF-transparent or at least well-characterized so their influence can be de-embedded. Foam blocks, 3D-printed low-permittivity plastics, and customized vacuum chucks have been employed, but each introduces subtle loading. Achieving intra-lab repeatability across multiple test setups remains an open challenge, particularly when comparing measurements taken on different days or by different operators. Fixture resonance can also introduce spurious responses that interfere with the antenna's S-parameters, requiring careful design to keep fixture self-resonances outside the band of interest. One effective strategy is to incorporate absorbing material into the fixture structure or to use a frequency-sweeping pre-measurement to identify and avoid resonant peaks. Future fixtures may incorporate active cancellation techniques that null out fixture reflections in real time.

Environmental Sensitivity

Flexible antennas often employ hygroscopic substrates such as cotton, polyester fabrics, or hydrogels that absorb moisture. Humidity swings can change the substrate's effective permittivity and loss tangent, directly affecting S11 and transmission coefficients. Temperature drift similarly alters the dielectric constant of elastomers and the resistance of conductive traces. When testing on-body antennas, the proximity of high-permittivity human tissue (with εr ~ 40–60 at microwave frequencies) drastically detunes the antenna, yet standard VNA calibration uses an air environment. Replicating the complex dielectric environment in a controlled lab setting—through tissue-simulating liquids or phantoms—is essential but introduces additional variables that must be factored into measurement uncertainty budgets. Even the residual moisture from a user's skin can significantly change results if not allowed to equilibrate. Modern measurement protocols recommend stabilizing samples in a controlled environment (e.g., 23°C, 50% RH) for at least 24 hours prior to testing, and logging environmental data alongside each sweep to enable post-hoc corrections.

Material Nonlinearities and Aging

Flexible antenna substrates are often polymers that exhibit viscoelastic behavior. When subjected to cyclic deformation, the material may not return to its exact initial dimensions, introducing hysteresis in S-parameter measurements. Over repeated bends, microcracks can develop in conductive traces, gradually increasing resistance and altering impedance matching. This aging effect means that S-parameters measured on a fresh prototype may not represent the antenna's performance after hundreds of flex cycles. Accelerated life testing combined with periodic S-parameter sweeps is necessary to characterize these degradation trends, but the test protocol must ensure that the measurement itself does not further damage the delicate structure. Techniques such as low-power VNA sweeps (e.g., -10 dBm) and short measurement dwell times help minimize stress-induced artifacts. Additionally, using a reference antenna made of the same materials but stored undeformed allows operators to separate aging effects from measurement drift.

Radiation and Coupling Artifacts

S-parameter measurements assume that the device under test (DUT) is a linear, time-invariant network. When a conformal antenna is mounted on a metallic or composite structure, the radiated fields interact with nearby scatterers, and the antenna may receive energy back through its own radiation pattern. This happens even in an anechoic chamber if the fixture or the operator's body is in the near field. The result is a ripple in S11 or S21 that can be mistaken for internal impedance anomalies. Correctly gating the time-domain response or performing the measurement in a fully absorber-lined enclosure minimizes these effects, yet doing so while the antenna is bent on a curved fixture complicates positioning. Time-domain gating is particularly powerful because it can isolate the antenna's own reflections from those caused by the environment, provided the gate width and position are set based on the expected electrical length of the DUT. Some modern VNAs offer automated gate optimization algorithms that use the impulse response to set the gate boundaries without user intervention.

Reference Plane Ambiguity

In rigid antenna measurements, the calibration reference plane is typically established at a coaxial connector interface using short-open-load-thru (SOLT) or thru-reflect-line (TRL) standards. For flexible antennas, the actual radiating element is often a printed dipole or patch that begins at the edge of a flexible transmission line. Any intervening length of line—itself deformed by bending—adds phase delay and mismatch that must be de-embedded. De-embedding algorithms assume a uniform transmission-line model, but the characteristic impedance of a microstrip on a curved, thin substrate may differ from the flat-case value. Without accurate knowledge of the line's electrical length under curvature, the de-embedded S-parameters at the antenna terminals carry considerable uncertainty. A practical solution is to fabricate custom calibration standards on the same flexible substrate and bend them together with the antenna, ensuring that the reference plane moves with the deformation. Advanced techniques use on-wafer-like TRL structures printed directly on the flexible film, allowing calibration to be performed at the exact location of the antenna feed.

Emerging Measurement Methodologies

Customized Flexible Test Fixtures

To bring repeatability to deformation states, several laboratories have developed motorized bending jigs that can set a precise radius of curvature and hold it during the sweep. One approach uses a pair of rotating drums wrapped with a thin, low-dielectric-constant film that presses the antenna against a curved surface without metallic contact. Another employs vacuum-assisted shaping where the antenna substrate is drawn against a thermoplastic mold of known permittivity. In all cases, the fixture's S-parameters are characterized first as an empty "fixture adapter," then de-embedded from the total measurement, a process made more robust by adaptive de-embedding algorithms that account for frequency-dependent substrate compression. Fixtures designed for stretchable antennas incorporate linear actuators with strain gauges to precisely control elongation while monitoring the applied force. The adoption of 3D-printed, low-loss materials (e.g., polyethylene or PTFE-based filaments) allows rapid prototyping of custom-curved supports that match the exact geometry of the final application.

In-Situ and On-Body Measurements

For wearables, the most meaningful S-parameter data often comes from measurements performed directly on the human body or on an anatomical phantom. Portable VNAs such as the Keysight FieldFox handheld analyzers enable one-port reflection measurements while a subject wears the antenna. To reduce movement artifacts, a wireless synchronization trigger or a lightweight coaxial cable routed along the body can be used. Similarly, tissue-equivalent phantoms filled with liquids that match the dielectric properties of muscle or fat at the target frequency allow repeatable benchtop tests. These phantoms can be shaped as cylinders or spheres to simulate limb curvature, providing a standardized environment that improves cross-study comparisons. For critical applications, liquid phantoms are preferred over solid ones because they more accurately mimic the frequency-dispersive behavior of human tissue. Recent advances include the use of semi-liquid gels that combine shape retention with tissue-like permittivity, offering a compromise between ease of handling and accuracy.

Over-the-Air (OTA) Near-Field Scanning

When physical connectors are undesirable, one can infer S-parameters indirectly from radiated near-field scans. A probe antenna scans the near-field region of the flexible DUT, and back-projection algorithms reconstruct the antenna's reflection coefficient. While not a direct S-parameter measurement in the traditional two-port sense, combining near-field amplitude and phase data with a numerical model of the feed network yields equivalent S11 and mutual coupling terms. This non-contact method eliminates connector stress entirely, and it can be performed while the antenna is mounted on any surface. The main limitation is the need for high-resolution phase-accurate scanners and careful calibration of the probe's own scattering characteristics, a topic well-covered in resources like the IEEE 149-2021 Standard Test Procedures for Antennas. Advanced implementations use synthetic aperture techniques to reduce scan time without sacrificing accuracy. Some systems now integrate machine learning to reconstruct S-parameters from partial near-field scans, reducing measurement time by up to 70%.

Reconfigurable Measurement Systems with Robotic Manipulators

To capture S-parameters across an entire range of curvatures or stretch conditions, some facilities are integrating six-axis robotic arms that can gently bend the antenna while a VNA sweeps continuously. The robot can apply a predefined sequence of deformations, recording S-parameters at each step. This allows the generation of three-dimensional S-parameter maps as a function of bend radius and azimuthal angle. When coupled with machine learning models trained on simulated data, such maps can be used to predict performance for any arbitrary shape, dramatically reducing the number of physical measurements needed. The robot's end effector must be designed with compliant, RF-transparent materials to avoid perturbing the antenna's fields, and the entire system is often enclosed in a shielded anechoic chamber to isolate external interference. Newer robotic systems use prehensile grippers with force feedback to gently hold flexible antennas without introducing creases or stress concentrations.

Time-Domain Gating for Reflection Artifacts

Time-domain reflectometry (TDR) built into modern VNAs can be leveraged to separate the antenna's own response from spurious reflections due to fixtures or nearby objects. By transforming the frequency-domain S-parameters to the time domain, an operator can apply a gate that passes only the portion of the response corresponding to the antenna itself. This technique is especially useful for conformal antennas mounted on structures where absorber placement is impractical. The gate must be carefully chosen to exclude fixture reflections without truncating the antenna's own resonant behavior. Automatic gate detection algorithms based on the derivative of the time-domain response can set optimal gate boundaries, reducing operator variability. Many VNAs now offer built-in time-domain analysis that allows users to view the impulse response directly and select a gate with a simple cursor, making the technique accessible even to non-specialists.

Advanced Calibration and Error Correction

Correcting for the non-idealities of a flexible test setup often requires going beyond standard SOLT calibration. TRL (Thru-Reflect-Line) calibration can be implemented on the flexible substrate itself, using printed standards that flex together with the antenna. When the same substrate and manufacturing process are used for the calibration structures, many systematic errors cancel out. A comprehensive overview of VNA calibration methods is provided by Understanding VNA Calibration guides, which describe how to choose the right technique for challenging fixture environments. Another emerging approach is the use of adaptive error models where the VNA compares the measured response of a known reference tag placed adjacent to the DUT and mathematically compensates for residual fixture drift in real time. Machine learning models that are trained on a large dataset of measurements under various deformation states can also predict and correct systematic errors without requiring physical recalibration. These models can even interpolate S-parameters for bend states that were never measured, effectively creating a digital twin of the antenna's electrical behavior.

Case Studies and Application Examples

Wearable Health-Monitoring Antenna

A textile-based dual-band antenna designed for off-body communication in the 2.45 GHz ISM band was measured first on a flat foam substrate, showing an S11 below −15 dB. When the same antenna was wrapped around a forearm phantom with a 50 mm radius, the resonance shifted upward by 120 MHz and the S11 degraded to −8 dB due to tissue coupling and curvature. By using a portable VNA with a lightweight semi-rigid cable taped along the arm, the team captured on-body S-parameters that guided a design iteration: the patch was lengthened slightly to compensate for the effective permittivity increase, bringing the bent-state return loss back to better than −15 dB. The study, documented in a review of wearable antenna measurement setups, highlights the critical role of realistic test conditions. Further tests at different humidity levels (30%, 60%, and 90% relative humidity) revealed that moisture absorption in the cotton substrate shifted the resonance an additional 30 MHz, emphasizing the need for environmental control during qualification. The final design included a moisture-barrier coating that stabilized the dielectric constant, reducing the humidity-induced shift to under 10 MHz.

Conformal Aircraft Blade Antenna

An ultra-wideband blade antenna integrated into a composite wing leading edge needed to maintain a VSWR below 2:1 from 500 MHz to 3 GHz. Traditional planar fixture measurements predicted acceptable performance, but when the antenna was bonded onto a representative curved wing section and tested in an anechoic chamber, S11 exhibited a deep null at 1.8 GHz. Subsequent time-domain gating revealed a reflection from the joint between the wing mock-up and the test support, which was absent in the flat scenario. The issue was resolved by adding absorptive material around the wing edges, demonstrating that S-parameter measurements of conformal antennas must account for the entire structural environment. Additional sweeps with the antenna at different positions along the curved surface showed that the impedance match varied by up to 5 dB depending on curvature radius, requiring a database of S11 maps across the antenna's deployment envelope. The final qualification process involved measuring the antenna in three orthogonal orientations to ensure multi-axis robustness.

Flexible RFID Tag on Curved Packaging

For supply-chain applications, UHF RFID tags need to read reliably when affixed to bottles or curved packages. The tag IC's impedance is conjugate-matched to the antenna's S11 at 915 MHz on a flat surface. A battery of S-parameter measurements performed with the tag mounted on cylinders of varying diameters showed that the impedance match degraded rapidly at radii below 30 mm. By analyzing the S11 data and recalculating the power transfer coefficient, engineers determined the maximum read-range reduction threshold and adjusted the antenna geometry to be more tolerant of curvature. This iterative measurement-driven design cycle prevented expensive tooling corrections later. The final antenna design used a meandered radiator that maintained better than −10 dB return loss down to a 20 mm bend radius, validated by over 100 repeated measurements across three different test labs. To ensure consistency, each lab followed a detailed measurement protocol that specified the phantom material, temperature, and humidity conditions.

Flexible 5G mmWave Array

A 28 GHz 1x4 patch array printed on a liquid crystal polymer substrate was designed for a curved smartphone display. Planar measurements showed mutual coupling between adjacent patches below −20 dB, but when the array was conformed to a radius of 40 mm (typical for a phone edge), the coupling increased to −14 dB. This change altered the array's phase response, affecting beam steering accuracy. By using a robotic arm to systematically vary the curvature while measuring all S-parameters with a 4-port VNA, the team developed a compensation algorithm that adjusted the phase shifters based on real-time curvature sensor data. The success of this approach hinged on the repeatability of the robotic fixture, which maintained curvature to within ±0.5 mm during each sweep. The measurement campaign also revealed that the array's mutual coupling was sensitive to the direction of bending (convex vs. concave), prompting a redesign of the feed network to balance the coupling in both states.

Advancements in Test Equipment and Future Directions

The landscape of S-parameter measurement tools is evolving to better serve flexible and conformal antennas. Modular, software-defined VNAs now combine compact form factors with frequency extension modules that cover mmWave bands where many future flexible arrays will operate. Built-in "fixture de-embedding" wizards guide users through the calibration of non-standard fixtures, lowering the barrier to accurate measurements. USB-connected VNAs with open-source control scripts allow automated sweep sequences synchronized with a motion controller, enabling hands-free deformation studies. Real-time temperature and humidity sensors integrated into the test chamber can now log environmental conditions alongside S-parameter data, enabling robust uncertainty analysis. One notable trend is the use of multi-port VNAs with up to 24 ports, which allow simultaneous measurement of all coupling terms in large flexible arrays without reconnections.

Standards organizations are actively updating test methods. The IEEE Antennas and Propagation Society is expanding recommendations to include measurement techniques for reconfigurable and conformal antennas, recognizing that the static, far-field paradigm does not translate directly to structures that change shape during operation. A good starting point for understanding the wider context of S-parameters is the classic resource on scattering parameters from Microwaves101. Additionally, ongoing work on digital twins—where a high-fidelity simulation model is updated with sparse measured data—could reduce the measurement burden by interpolating S-parameter surfaces from a handful of deformation states. The European Telecommunications Standards Institute (ETSI) is also developing specific test procedures for flexible antennas used in wearable medical devices, which will mandate standardized phantoms and measurement protocols.

In parallel, additive manufacturing and printed electronics are enabling the co-fabrication of antennas alongside dedicated calibration structures directly on the flexible substrate. This will allow the TRL standards to experience the same mechanical loads as the antenna, eliminating many de-embedding ambiguities. The integration of optical fiber interrogation to monitor strain in real time during a VNA sweep may soon provide an independent check on the antenna's mechanical state, correlating S-parameter changes directly with local strain fields. Future networked measurement systems that combine multiple robots, multi-port VNAs, and machine learning engines promise to fully automate the characterization of flexible antennas under any conceivable deformation scenario. Cloud-based data analysis platforms are emerging that allow design teams to share S-parameter maps and compare results across different laboratories, accelerating the development cycle.

Conclusion

The measurement of S-parameters for flexible and conformal antennas demands a multidisciplinary approach that blends RF engineering, materials science, and precision mechanics. Shape sensitivity, fixture design, environmental loading, and reference plane definition interact in ways that are not present in rigid-antenna testing. By embracing custom flexible fixtures, portable measurement instruments, OTA near-field scanning, and adaptive calibration routines, engineers can obtain dependable S-parameter data under realistic operational conditions. These datasets feed back into simulation models, verify manufacturing consistency, and ultimately give designers the confidence to deploy conformal antennas in safety-critical and high-performance systems. As test equipment becomes more agile and standards evolve, the gap between bench-level convenience and in-field reality will continue to narrow, ensuring that flexible antennas deliver the robust performance their applications demand. The investment in careful measurement practice today will pay dividends in reduced field failures, faster time-to-market, and the ability to push the boundaries of what conformal antennas can achieve.