RF Interconnect Engineering

How to Specify an RF Coaxial Cable Assembly: Insertion Loss, VSWR, Power, Phase Stability and Bend Life

A field-ready method for specifying RF and microwave coaxial cable assemblies by topology, calibrated reference planes, insertion loss, return loss, power, phase and amplitude stability, delay, shielding, bend limits, connector control, environment and acceptance evidence.

Published: Jul 15, 2026
Reading time: 10 min

People-free two-port RF coaxial cable assembly verification bench with complete connected cable, controlled bend mandrel, semi-rigid sample, calibration standards and torque tool

Freeze the finished two-port assembly before comparing cable series

A defensible cable specification describes the complete path from one declared connector reference plane to the other. Cable construction, length, connectors, backshells, launch geometry, forming, routing and handling all contribute to the measured result. If any of them can change after approval, insertion loss, match, electrical length and repeatability can change with them.

Start with topology, installation and service state

State whether the path is a flexible jumper, semi-rigid formed assembly, test cable, low-loss feeder, internal harness or phase-matched set. Describe the equipment at both ends, the installed route, movement during service, replacement strategy and whether the cable remains part of the calibrated measurement path. These details determine whether lowest nominal attenuation, stable electrical length, repeated flexure, shielding or compact routing deserves priority.

Freeze the number of assemblies, channel pairing, branch symmetry and any matched-set requirement. A phase-matched antenna feed set is not accepted the same way as a single service jumper. A production test cable may need thousands of controlled movements and remates, while a semi-rigid internal line may be formed once and then judged by shape retention and connector alignment.

Define continuous frequency coverage and both reference planes

Use continuous operating and verification bands rather than one center frequency or the connector's headline rating. Name the reference impedance and both calibrated planes: normally the mating interfaces, but sometimes a fixture or launch plane when the assembly is integrated. State whether adapters, torque devices, right-angle transitions or bulkhead feedthroughs are inside the delivered boundary.

Reference-plane ambiguity creates apparently conflicting data. A connector pair, adapter or fixture can add loss, reflection and delay, and de-embedding can move the reported result. Preserve the calibration method, port extension or de-embedding file and exact assembly orientation so design, supplier and acceptance laboratories are comparing the same network.

Keep transmission loss and port match as separate limits

Insertion loss is derived from forward transmission, while return loss or VSWR describes reflection at each port. A low-loss cable can still have a poor connector transition, and a well-matched assembly can still dissipate too much power over a long route. Specify maximum insertion loss across the full band and separate input and output return-loss limits when the two ends or installation environments differ.

Avoid approving a smooth typical curve without production tolerance. Set frequency resolution, smoothing rules, temperature states and whether limits are absolute or relative to a baseline. Retain complex S-parameters where phase, time-domain gating, de-embedding or cascade analysis matters; scalar screenshots cannot support those operations.

Check power, voltage, heating, shielding and unwanted products

State average and peak power, waveform, duty cycle, crest factor, mismatch state and fault duration. Power capability can be limited by conductor heating, dielectric loss, connector contact resistance, peak voltage, altitude, contamination or a small local discontinuity. A connector series power statement does not automatically apply to a long cable assembly at the highest operating frequency and temperature.

Where receiver protection or multi-carrier purity matters, specify shielding effectiveness, leakage, passive intermodulation or connector cleanliness with a usable test method. PIM is an assembly and handling problem as much as a material property: contact condition, torque, plating, contamination and mechanical stress can dominate. Keep those requirements separate from ordinary S-parameter acceptance.

Specify phase, electrical length and delay stability by cause

Absolute phase, phase matching, group delay and time delay answer different system questions. A phased array may care about channel-to-channel phase over frequency and temperature; a timing path may care about delay; a laboratory cable may care about change after movement. State the baseline, frequency points, comparison channels and whether the limit is total variation, peak-to-peak change or repeatability after returning to a defined position.

Separate flexure-induced change from temperature-induced change. Define bend radius, bend angle, movement plane, cycle count, rate and dwell, then define thermal range, ramp, soak and measurement state. Without the test geometry, a phase-stability number cannot be reproduced or compared.

Choose flexible, conformable or semi-rigid construction by the failure boundary

Larger-diameter low-loss cable can reduce attenuation but raises mass, bend radius and connector load. Thin flexible cable routes easily but may lose more power and show greater phase change with movement. Semi-rigid cable offers repeatable shielding and formed geometry, yet requires controlled tooling, minimum bend radius and dimensional inspection. No construction is universally superior; the installed boundary chooses the tradeoff.

Current approved product records in this category include semi-rigid assembly evidence to 18 GHz and 20 GHz with configuration-specific dielectric-withstand and temperature data. Those examples demonstrate the need for record-level verification; they do not extend the same limits to a different length, connector, bend pattern or unapproved family name.

Control connector, bend, strain relief and routing details

Specify connector series, precision grade where relevant, gender, orientation, plating, interface standard, mating torque, coupling-nut access and allowable mating cycles. Freeze overall length and the reference points used to measure it. For angled connectors, define clocking. For semi-rigid assemblies, supply a controlled drawing with centerline dimensions, bend radii, straight lengths and tolerances.

Distinguish minimum static bend radius from dynamic bend radius and do not bend at the connector termination. Define strain relief, clamp spacing, unsupported mass, torsion, crush load and installation keep-outs. A cable that passes on a straight bench can fail after a tight route transfers bending moment into the connector or changes the dielectric geometry.

Translate the platform environment into a repeatable stress sequence

List operating and storage temperature, altitude or pressure, humidity, salt or chemical exposure, ingress, vibration, shock, radiation and flammability where applicable. Define whether RF measurements occur during exposure, at stabilized temperature or after recovery. Separate survival from retained performance: an assembly can remain electrically continuous while insertion loss, match or phase has moved outside the system budget.

For flex life, define the actual movement envelope rather than a generic cycle count. Include bend radius, travel, torsion, speed, dwell, connector restraint and inspection interval. Acceptance should compare the post-stress network against both absolute limits and its own serialized baseline so gradual degradation is visible.

Design calibration, repeatability and evidence before the RFQ

Use a VNA calibration appropriate to connector type, frequency and uncertainty. Stabilize and support test cables, inspect and clean interfaces, apply controlled torque and record adapter use. Run connector-remate and fixture-repeatability checks separately from unit repeatability; otherwise laboratory variation can be mistaken for cable instability or can hide a weak assembly.

Retain native complex data, calibration identity, frequency grid, power level, IF bandwidth where relevant, temperature, routing state, serial number, drawing revision, connector history and deviations. A PDF curve may support review, but it is not a substitute for machine-readable data when the assembly will be cascaded, matched, gated or compared over service life.

Decision boundary	Requirement to freeze	Do not accept
Assembly topology	Use, route, movement, matched-set role and both endpoints	A cable series is proposed without the installed configuration
Frequency and planes	Continuous bands, impedance and calibrated plane at each end	Only center frequency or connector rating is provided
Transmission and match	Full-band insertion loss plus S11 and S22 limits	One typical attenuation value substitutes for assembly data
Power and shielding	Waveform, average/peak power, mismatch, heating, leakage or PIM	A connector headline rating is applied to the whole path
Phase and delay	Metric, baseline, frequency, temperature and movement method	A stability number has no test geometry
Mechanical configuration	Connector, torque, length, clocking, bend radius and support	The delivered route can differ from the qualified drawing
Environment and life	Exposure sequence, flex geometry, cycle count and retained limits	Continuity after stress is treated as full RF qualification
Evidence	Calibration, uncertainty, native data, serial and revision traceability	Only a catalog curve or untraceable screenshot is available

Worked budget: convert the complete assembly loss into delivered power

Consider an illustrative 2.4 m assembly whose allowed cable attenuation is 0.30 dB/m at the band edge, with 0.12 dB allocated to each terminated connector transition. The assembly insertion-loss budget is 0.72 + 0.24 = 0.96 dB. A 0.96 dB loss corresponds to a power ratio of about 0.802, so 20 W entering a matched assembly would deliver about 16.0 W and dissipate roughly 4.0 W along the path. Do not subtract a separate VSWR penalty unless the system budget explicitly separates mismatch from the measured insertion metric; instead, verify S11, S22, power, temperature and uncertainty at the declared planes. The numbers illustrate the calculation method, not a product rating.

Text-free engineering sequence showing a two-port RF cable assembly at calibrated reference planes, straight baseline, controlled mandrel flexure, thermal exposure and connector remate repeatability — Qualify one serialized assembly through a controlled sequence: baseline network, known flexure, defined thermal state and repeatable connector remate, always at the same reference planes.

Build acceptance from baseline network data to post-stress repeatability

Approve topology, endpoints, continuous frequency bands, reference impedance and calibrated planes.
Freeze connector, gender, orientation, torque, length, bend geometry, supports and drawing revision.
Inspect interfaces and route the assembly in the declared baseline state without torsion or connector loading.
Calibrate the VNA and verify residual directivity, source/load match, tracking and cable stability appropriate to the limits.
Measure complex S11, S21, S12 and S22 with the required frequency resolution and temperature state.
Apply controlled flexure using the declared bend radius, movement plane, cycle count, speed and dwell.
Apply thermal and environmental exposure with stated ramp, soak, measurement and recovery conditions.
Run controlled connector remates to separate interface repeatability from cable-body change.
Repeat network and, where required, power, shielding or PIM tests against absolute limits and serialized baseline.
Retain native data, calibration, uncertainty, photographs of routing without people, serial, revision and deviation record.

Cable-assembly failures hidden by a catalog-only comparison

Using bulk cable attenuation as the finished assembly insertion loss
Specifying one VSWR value without port, frequency or reference plane
Applying the connector frequency or power rating to every assembled configuration
Leaving overall length, connector clocking or bend geometry uncontrolled
Quoting phase stability without movement and temperature test methods
Bending too close to the termination or allowing connector side load
Mixing connector-remate variation with cable-body phase change
Treating continuity after vibration or flexure as proof of retained RF performance
Accepting a smooth typical plot without tolerance, uncertainty or native data
Assuming the breadth of a category page proves that every family is stocked

Information required for an RF coaxial cable assembly RFQ

System use, equipment at both ends and whether the assembly is calibrated, fixed or repeatedly moved
Quantity, matched-set or channel-pairing requirement and replacement strategy
Continuous operating and verification bands plus characteristic impedance
Connector series, precision grade, gender, orientation, plating, torque and mating-cycle requirement
Overall length, tolerance, reference points, route drawing, clocking and keep-out envelope
Static and dynamic bend radius, movement plane, travel, torsion, cycle count, speed and support
Maximum insertion loss, input/output return loss, phase, delay or matching limits by frequency
Average and peak power, waveform, duty, mismatch, fault duration and thermal condition
Shielding, leakage, PIM or environmental-sealing limits and test method where required
Operating/storage temperature, pressure, humidity, vibration, shock and exposure sequence
Calibration plane, VNA method, uncertainty, remate protocol and pass/fail limits
First-article data, native Touchstone files, serial traceability, drawing revision and change control

Continue the interconnect engineering decision

Related FAQ

How should insertion loss and return loss be specified for a complete RF cable assembly?

Specify the finished two-port assembly, not only bulk cable attenuation. State continuous frequency bands, overall length, both connector interfaces and calibrated reference planes. Set maximum insertion loss from S21 and separate input and output return-loss or VSWR limits from S11 and S22.

Name temperature, routing and bend state, frequency resolution and production tolerance. Include adapters or fixtures only when they are part of the delivered boundary; otherwise calibrate or de-embed to the mating interfaces. Preserve complex data when phase, delay, gating or cascade analysis is required.

Acceptance should document calibration, connector inspection and torque, cable support, remate repeatability, uncertainty, serial number and drawing revision. A typical cable attenuation curve or screen capture cannot qualify the complete assembly.

How should microwave cable phase stability be tested under flexure and temperature?

Choose the system metric first: absolute phase, phase matching between channels, electrical length, group delay or time delay. State the frequency points, baseline condition and whether the limit is total change, peak-to-peak movement or repeatability after the cable returns to one position.

For flexure, define bend radius, bend angle, movement plane, travel, torsion, speed, cycle count, dwell and connector restraint. For temperature, define range, ramp, soak, measurement state and recovery. Test the mechanisms separately before applying a combined mission sequence.

Retain complex S-parameters and the serialized baseline, then compare post-stress data at identical reference planes. Connector remate variation should be measured independently so it is not misreported as cable-body phase instability.

What bend-radius and connector details belong in an RF cable assembly RFQ?

State connector series, precision grade, gender, orientation or clocking, plating, mating torque, access and mating-cycle requirement at both ends. Define overall length, tolerance and the exact reference points used for measurement; attach a centerline drawing for formed or semi-rigid assemblies.

Separate minimum static and dynamic bend radius. Define movement plane, travel, torsion, cycle count, speed, clamp locations, strain relief, unsupported mass and keep-outs, and prohibit bending at the termination unless the supplier has qualified that geometry.

Ask for assembly-level S-parameter evidence in the delivered route and for dimensional inspection where shape is controlled. Any connector, length, bend or support change should trigger configuration review rather than inheriting the previous approval automatically.

What belongs in RF cable assembly power, VSWR, environmental and acceptance testing?

Baseline the complete assembly with calibrated complex S-parameters at declared reference planes. Record routing, bend state, connector torque, temperature and uncertainty. Where power matters, state average and peak waveform, duty cycle, mismatch, fault duration and the thermal locations to monitor.

Apply only the relevant flexure, temperature, pressure, humidity, vibration, shock or sealing sequence with defined ramp, dwell and recovery. Measure during exposure when the requirement is operational; otherwise state the stabilized or post-recovery condition. Survival and continuity are not substitutes for retained RF limits.

Repeat insertion loss, return loss and any phase, delay, shielding or PIM checks after stress and controlled remates. Retain native data, calibration identity, serial number, drawing revision, connector history and deviations so production and replacement units can be compared.

Engineering inquiry

Share your RF requirement

Share the product, operating requirements and project context. Our engineering team will route your request to the right specialist.