A VNA Trace on the Coax Shield Is Not the Whole Choke Test
A VNA Trace on the Coax Shield Is Not the Whole Choke Test
Clip ferrite around coax, measure a fall in S21 and declare victory? The trace may contain useful information, but only after we define the circuit, the reference planes and the current mode we actually measured.
I understand why the braid-only test is attractive. Common-mode current is associated with the exterior-current path of the feedline, so using the coax shield as the test conductor feels direct. The mistake is not touching the shield. The mistake is assuming that any two-port attenuation trace automatically equals the installed choke's common-mode suppression.
A useful small-signal impedance measurement and proof of an installed choke are different things. A defined fixture can drive all conductors through the ferrite in the same direction and measure the resulting series impedance. It still does not reproduce every antenna return path, shunt capacitance, cable route, drive level or thermal condition found at the station.
Begin With the Circuit the VNA Actually Sees
A two-port VNA launches and receives waves through two nominally terminated ports. When the centre pins are connected to opposite ends of one conductor and the port shells provide the return, the instrument sees a complete fixture: launch transitions, the conductor under test, the choke, the return path between the port shells and every electric or magnetic coupling path around the device.
That arrangement can be a legitimate series-through impedance fixture. It is not invalid simply because the conductor happens to be a coax shield. But the result belongs to that fixture. Leaving the centre conductor floating, bonding it to the shield, terminating it or routing it near the return structure changes the parasitic network. Moving the cables changes the loop area and mutual coupling. A trace without the fixture drawing does not tell another builder what was measured.
For a coaxial common-mode choke, the intended test current passes through all enclosed conductors in the same direction with respect to an external return. The wanted differential current inside the coax has equal and opposite conductor currents and ideally contributes little net core flux. A fixture should make that modal distinction explicit rather than relying on the word shield.
S21 Is a Measurement, Not Yet the Measurand
S21 is the complex forward transmission coefficient of the complete two-port network at the calibrated reference planes. Its magnitude alone is not choke impedance. To infer a series impedance from S21, we must use a declared circuit model and complex data. For an ideal series element between equal reference impedances Z0, the series-through conversion is:
ZDUT = 2 Z0 (1 − S21) / S21
This relation assumes the fixture behaves like the stated series model. It does not remove unmodelled shunt admittance, direct port-to-port coupling or radiation.
A full S-to-Y conversion uses all four complex S-parameters. Under a suitable pi-network interpretation, the transfer admittance Y21 can help separate the series branch from shunt branches, and the branch impedance is related to −1/Y21 under the stated sign convention. That is why Y21 can be powerful. It is not magic and it is not the only measurement that can be valid. The model, calibration and fixture still decide what the extracted quantity means.
What Can Corrupt a Braid-Path Trace?
- Reference-plane error: adapters, exposed braid, clips and lead length add delay, inductance and loss unless calibration or de-embedding moves the planes to the DUT.
- Shunt capacitance: winding-to-winding, winding-to-core, cable-to-fixture and port-to-port capacitance can bypass a high series impedance.
- Direct coupling: the two test leads can exchange energy electrically or magnetically without following the intended conductor path.
- Return-path ambiguity: the port shells, bench, instrument chassis, bonding lead and nearby metal form the external return. Their geometry is part of the result.
- Dynamic range: deep transmission notches can approach the analyser and fixture noise floor. A smooth-looking high-attenuation trace may then be a measurement limit.
- Distributed behaviour: the winding becomes a network of line sections and parasitics as frequency rises. One lumped series impedance may stop describing it.
Skin effect changes conductor resistance with frequency, but it is rarely a sufficient explanation for a dramatic choke trace by itself. Likewise, a resonance is not automatically proof of useful suppression or proof of failure. It is a feature of the complete complex network and must be interpreted with its resistive and reactive parts.
A Repeatable Bench Measurement
Define
Draw the DUT and return path. State whether centre and shield are joined, floating or terminated, and identify the intended common-mode conductor.
Calibrate
Move the reference planes to the fixture terminals where possible. Characterise or de-embed adapters and lead-ins; save complex S-parameters.
Challenge
Measure an empty fixture, a through, known reference impedances and repeated cable positions. A useful method must distinguish DUT response from fixture coupling.
Select the measurement method for the expected impedance range. Reflection, series-through, shunt-through, full two-port Y-parameter extraction and an impedance analyser each have useful and difficult regions. Agreement between independent methods across their overlapping range is far stronger evidence than loyalty to one conversion.
Record complex impedance, not only attenuation. A choke whose impedance is mainly resistive can damp a current-path resonance differently from one whose magnitude is similar but mostly reactive. Neither R nor X alone is a universal score: the installed source impedance, return structure and target bandwidth matter.
The Station Adds the Part the Bench Cannot Promise
The bench result describes a small-signal DUT in a stated fixture. On the antenna, common-mode current is set by the complete network: antenna imbalance, deliberate counterpoise or radials, transformer, feedline exterior, mast, bonding, earth coupling, equipment and the choke's position along that path. The same choke can encounter different source impedances at different locations and frequencies.
Verify the installation with repeatable exterior-current measurements on the same cable route, before and after the choke, while keeping accepted transmitter power and measurement geometry controlled. Also compare wanted signal, received noise and station behaviour where relevant. A lower current at one probe location does not by itself prove radiation efficiency, field pattern, EMC compliance or RF-exposure compliance.
Small-signal sweeps do not establish a power rating. At transmit power, ferrite permeability and loss vary with frequency, field and temperature. Coax voltage, current, dielectric loss, bend radius, connector heating, duty cycle and enclosure temperature also matter. Powered qualification needs declared mismatch, waveform, duty cycle, duration and stop limits.
What the Shield Test Can Honestly Tell Us
A carefully designed braid-path or all-conductors-together fixture can provide useful complex common-mode impedance data. It can compare winding counts, core stacks, cable choices and frequency behaviour under controlled small-signal conditions. It can reveal resonances and help choose candidates for an installation.
What it cannot do alone is turn an S21 dip into a universal number of decibels removed from every station. That final claim requires the missing context: calibrated fixture model, impedance extraction, installation current path and powered behaviour. Follow the current path, but draw the return path too.
Primary and authoritative technical references
- Keysight E5061B — impedance-measurement methods and conversion equations
- Keysight E5061B — full S-parameter to Y-parameter conversion
- Keysight — fixture embedding, de-embedding and mixed-mode conversion
- Rohde & Schwarz — accurate fixture characterisation and de-embedding
- Fair-Rite — specifying ferrite from complex permeability and frequency
Mini-FAQ
- Is connecting a VNA through the coax shield always invalid? No. A defined series-through fixture can measure useful small-signal impedance. The circuit, return path, calibration and model must be stated.
- Does a drop in S21 equal the installed choke suppression? No. S21 describes the complete two-port fixture. Installation suppression also depends on the antenna's common-mode source, return path, choke position and parasitic coupling.
- Why keep complex impedance instead of only attenuation? Resistance and reactance influence current-path resonances differently. Magnitude alone hides that distinction.
- Is Y21 the only valid way to measure a choke? No. It is a useful branch extraction under a suitable network model. Reflection, series-through, shunt-through and impedance-analyser methods answer related questions over different ranges.
- How do I check whether the fixture is dominating? Measure empty-fixture and through baselines, known reference impedances, repeated cable positions and, where possible, an independent method over an overlapping range.
- Does a VNA sweep prove the transmit-power rating? No. Powered electrical and thermal tests must cover the intended load, mismatch, waveform, duty cycle, duration and temperature limits.