Test cables at
26.5 GHz and above.

That sensitivity explains why cable movement belongs in the measurement budget. Calibration characterizes a particular system state. Movement, connector changes or temperature drift can change that state. Calibration corrects repeatable systematic errors; it does not eliminate subsequent drift or random noise. [2]

Focusimple’s FABTest addition. The selected 67 GHz FTSA assembly uses 1.85 mm interfaces and distinguishes installation from repeated-bending minima. Its typical mechanical stability figures have no published flex geometry, and the listed assembly loss has no typical / maximum qualifier. Keep those limits visible when requesting an equivalent-condition quotation. [10]

Read the flex test behind the stability number

Phase stability with flexure describes a change caused by movement; phase matching describes how closely separate assemblies’ electrical lengths agree. A closely matched pair still needs adequate stability in use. Amplitude stability similarly concerns a change in transmission or measured response during movement, rather than the assembly’s absolute insertion loss.

The selected Times and Gore specifications illustrate why the test geometry matters. Times describes a short-circuited assembly bent 90° around a 25.4 mm-radius mandrel. Gore describes a short-circuited, normalized assembly wrapped 360° around a 57 mm-radius mandrel and tested on both sides. Those are different mechanical excursions. The HUBER+SUHNER brochure does not specify an equivalent geometry for the cited stability figures. [3] [5] [6]

A short-circuit reflection method also involves a round-trip electrical path. Ask whether a reported phase excursion is a measured reflection change or an equivalent one-way transmission value, and whether “±” represents a limit about a reference or half a peak-to-peak excursion. Without those definitions, a smaller number may describe a different test.

Typical values and guaranteed limits answer different questions. Typical behavior can help identify candidates. An acceptance requirement needs an agreed limit, test method and frequency band. Missing information in a public datasheet is a question for the supplier, not evidence of inferior performance.

Compare loss at the same frequency and length

Start with the finished assembly’s insertion loss across the actual sweep. A cable-only attenuation value in dB/m does not automatically include connector transitions, and a value at 18 GHz cannot establish performance at 67 GHz. Keep typical values separate from maxima, and record the temperature basis when one is published.

For example, two matched test leads that each introduce 1.5 dB loss add 3 dB of uncorrected through-path attenuation, ignoring mismatch. Calibration can correct the stable response mathematically, but it cannot restore signal power lost before the receiver or remove the receiver’s random noise. Loss therefore remains relevant to usable measurement margin. [2]

The Gore and HUBER+SUHNER entries deliberately show the trap: both are approximately 36-inch assemblies, but the published loss figures use typical and maximum bases. Subtracting those figures would create a numerical difference without establishing a measured advantage.

Mechanical limits belong in the electrical decision

Separate a minimum routing radius from a qualified repeated-flex radius. The Times literature distinguishes a minimum from a larger radius recommended for flex life. HUBER+SUHNER explicitly labels the cited 570S radius static. A static or unlabeled minimum should not become a repeated-motion specification by assumption.

Armor can help protect a cable in a busy test area, but it also changes diameter and handling. Plan strain relief so the cable’s weight and movement are not concentrated at the connector. For a fixture that cycles continuously, qualify the complete motion path, including travel near the connector boots, rather than only a bend at the cable midpoint.

Connector care is equally practical. Inspect interfaces before mating, use the torque specified for the actual connector, and rotate the coupling nut rather than twisting the cable body. Check critical interfaces with the appropriate gauge and protect disconnected ends. Worn or contaminated connectors can undermine an otherwise careful cable selection. [7]

Qualify the cable in the motion it will see

A useful acceptance exercise is a controlled repeatability study using the intended instrument, sweep and routing. The following is an editorial qualification framework, not a standardized certification procedure:

1. Define the configuration. Record the assembly part number, length reference, connector types, adapters, sweep, source power, IF bandwidth and temperature.
2. Establish the reference planes. Stabilize the instrument and setup, then calibrate with the cables and connections in place using a method appropriate to the measurement. [8]
3. Use a stable check artifact. Record a baseline trace, then move the cable through a documented, non-damaging trajectory that represents normal handling while preserving the electrical connections.
4. Measure change across the band. Track amplitude and phase deviations relative to the baseline. Keep stationary repeat sweeps as a control so instrument drift and noise are visible.
5. Separate mechanisms. Evaluate disconnect/reconnect repeatability in a separate sequence. Recheck after controlled temperature changes and return to the baseline condition.
6. Set acceptance limits first. Derive them from the measurement uncertainty and decision margin, then document the results, handling conditions and a verification or replacement trigger.

A delay-only port extension is not a substitute for a full calibration at the desired planes. Instruments may offer additional loss or mismatch adjustments; use the instrument’s documented compensation model and verify its limitations. A cable’s changing response cannot be fixed permanently by entering one delay value. [9]

Choose for the workflow

A mostly stationary bench setup, a cable moved for every DUT, and a continuously cycling fixture present different requirements. Match the construction to the workflow, ask for the missing test conditions, and compare quoted assemblies on a common frequency, length and acceptance basis.

The useful purchasing question is: how much can this complete connection change during our measurement, and is that change small enough for the decision we need to make? Frequency capability gets a cable onto the shortlist. A documented, repeatable setup makes its performance meaningful.