SWR and Coax Faults: Find the Fault, Fix the Cause
SWR and Coax Faults: Find the Fault, Fix the Cause
A complete route from a changing SWR reading to a sound feedline: isolate the fault, measure reflection and loss, interpret TDR correctly, and repair water damage at its source.
RF.Guru working definition: Common-mode current is the non-cancelling phasor-sum current in a specified set of conductors, evaluated at a defined cross-section and using a declared current-direction convention. In the intended differential transmission-line mode, the outgoing and return currents are equal and opposite, so their phasor sum is zero. When they do not cancel, the remaining current must close through another reference or return path—such as the outside of a coax shield, a mast, equipment chassis, station wiring, nearby structures, earth, the operator, or distributed coupling through the environment.
This broader working definition is especially useful in practical antenna systems. On transmit, non-cancelling current on the outside of the coax can make the feedline and connected structures part of the radiating antenna system unless that path is intentional, clearly defined and properly controlled—for example by providing the required return path and placing a suitable common-mode choke at the correct boundary.
When SWR changes, the first thing I want to know is what changed in the complete RF path. The antenna may have moved. A connector may be loose. The feedline may be wet, crushed or partly open. The instrument may be reporting a different reference plane. Turning the tuner until the number becomes comfortable does not distinguish those cases.
My approach is to separate four questions: is the load mismatched, is the feedline losing too much power, where is the abnormality, and what physical repair will keep it from returning? SWR, insertion loss, time-domain reflectometry and inspection answer different parts of that investigation. Used together, they are much more useful than any one green indicator.
SWR tells you about a reflection at a particular place
Standing-wave ratio, usually written SWR or VSWR, describes the relationship between forward and reflected waves. For a real reference impedance Z0, normally 50 Ω in an amateur station, the load reflection coefficient is:
Γ = (ZL − Z0)/(ZL + Z0)
SWR = (1 + |Γ|)/(1 − |Γ|)
Return loss = −20 log10|Γ| dB
At the same reference plane, reflected power divided by forward power is |Γ|². A 3:1 SWR therefore means |Γ| = 0.5, 25% reflected power and about 6.02 dB return loss. It does not mean that 25% of transmitter power is necessarily dissipated in the cable. Dissipation, antenna acceptance and subsequent reflections require the actual system model.
Higher positive return loss means a smaller reflection. A VNA often displays S11 log magnitude with the opposite sign: −20 dB S11 corresponds to 20 dB return loss. Check the axis before deciding which trace is better.
A station meter measures at its own location. A meter before a tuner sees the tuner's input, not the antenna feedpoint. Meter directivity, frequency coverage, calibration and minimum useful forward power also matter. First prove the measuring arrangement with a known load and a known-good short jumper.
A lossy cable can make the SWR look better
For a uniform line whose characteristic impedance matches the measurement reference, the reflection travels to the load and back:
Γin = ΓLe−2γl, with γ = α + jβ.
If the one-way matched attenuation is A dB:
|Γin| = |ΓL| × 10−A/10.
The reflected voltage is attenuated on both journeys. Here is independent arithmetic for a fixed 3:1 load, using that simple line model:
| One-way attenuation | Reflection magnitude at the input | Input SWR |
|---|---|---|
| 0 dB | 0.500 | 3.00:1 |
| 3 dB | 0.251 | 1.67:1 |
| 6 dB | 0.126 | 1.29:1 |
Even a completely open end can appear as approximately 1.22:1 through an ideal uniform line with 10 dB one-way attenuation. That is a calculation, not a suggested operating condition. A wet or damaged real cable may also introduce distributed mismatches, so its response can be more complicated.
This is why I pair return-loss testing with a loss measurement. A long cable terminated in a good 50 Ω load can remain well matched while wasting considerable power. My earlier SWR and coax-loss article develops the power and heating consequences.
In an ideal lossless uniform line, changing cable length rotates reflection phase without changing its magnitude. In real installations it can also change loss, tuner loading and exterior cable current. None of those effects repairs a damaged connector. See why cutting coax is not an SWR repair.
Recognize the symptom without declaring the cause
| Observation | Plausible causes | Useful next check |
|---|---|---|
| SWR changes after rain | Wet antenna surroundings, an exposed feedpoint, connector contamination or water inside the cable | Substitute a verified load at the antenna end and compare with the saved dry baseline. |
| SWR improves while received signals and transmitted reports deteriorate | Extra feedline loss; also check equipment gain and propagation before diagnosing the cable | Measure cable insertion loss and compare at identical frequencies. |
| Moving a jumper changes the reading | Loose contact, broken braid or center conductor, a bad test lead, or exterior current changing the antenna system | Repeat with an isolated, terminated cable and controlled low-level excitation. |
| Low-power tests pass; trouble appears with power or time | Arcing, contact heating, dielectric damage, component compression or thermal movement | Stop transmitting; inspect and isolate before a controlled return to service. |
| Regular ripple appears across frequency | Interference between reflections at separated discontinuities | Inspect complex S11 and a calibrated distance trace; do not count ripple peaks as faults. |
| DC continuity passes but RF performance is poor | Incorrect connector geometry, partial shield contact, loss or an intermittent/nonlinear joint | Measure RF return loss and transmission as well as DC resistance. |
An SWR curve alone cannot identify water. Rain can change an antenna's capacitance, ground conditions or nearby vegetation without entering the coax. Conversely, a cable can be wet without producing a dramatic new SWR peak.
Start by dividing the station into testable sections
Draw the actual chain: radio, jumper, meter, tuner, switch, surge protector, main feeder, choke, matching network and antenna. Include adapters and bias tees. Mark where DC power is injected. A missing item in that sketch often becomes the item that escapes testing.
- Make the disconnected test safe. Disable transmit and remove RF and DC sources from the cable under test. Protect analyzer ports from nearby transmitters, stored charge and accidental bias power. Use the instrument's connection limits and procedure. Never reconnect a VNA while a transmitter can key into it.
- Verify the station end. Test the transmitter/meter arrangement into a suitable 50 Ω dummy load through a known-good jumper, at a power and duration the load supports. A tiny VNA calibration load is not a transmitter dummy load.
- Save the symptom. Record frequency, complex S11 if available, SWR, power, weather, temperature, cable routing and tuner state. Save files rather than only the best-looking marker.
- Replace the antenna with a verified load at the far end. If the reflection changes substantially, investigate the antenna/feedpoint branch. If the terminated feeder remains abnormal, investigate the line and everything still in that path. A good match alone does not clear the feeder's attenuation.
- Bypass one removable component at a time. Use known-good substitutions, restore the baseline between comparisons and record each configuration. Then test the suspect part separately.
- Locate, inspect and repair. Use reflection timing to narrow the physical search. Confirm the suspect connector, bend or section before cutting.
- Repeat the same measurements after repair. Establish a new baseline before sealing, then check again after sealing and final routing.
If the antenna is inaccessible, an abnormal trace still helps prioritize inspection, but it does not justify claiming that the antenna and feeder have been separated experimentally. State which termination was actually present.
What a multimeter can establish
On an isolated bare coax assembly, check center-to-center and shield-to-shield continuity, then center-to-shield insulation with the far end open. Account for test-lead resistance and the conductor resistance expected for the length. Where practical, four-wire resistance measurement improves small-resistance comparisons.
With access at one end only, temporarily shorting the far-end center to shield allows a loop-resistance check. That reads both conductors and both terminations in series; it cannot tell which conductor contains the problem. Remove the temporary short afterward.
“50 Ω cable” does not mean a multimeter should read 50 Ω across an open cable. Characteristic impedance is a traveling-wave property. A connected 50 Ω termination is what produces an approximately 50 Ω DC reading, plus conductor resistance. A transformer-fed or DC-grounded antenna can legitimately read as a short. Remove attached devices before interpreting a cable-only test.
A continuity beep is not an RF certificate. One surviving strand can pass that test while leaving an unreliable RF shield termination. An ordinary meter can also miss insulation breakdown that appears only at higher voltage. Insulation-resistance or withstand testing belongs only on an isolated assembly with a cable/connector-approved voltage and procedure; do not apply a megohmmeter through radios, analyzers, amplifiers, matching electronics or surge devices.
Measure loss as well as match
When both ends can reach a calibrated two-port VNA arrangement, measure S21 with the test leads and adapters accounted for at the reference planes. For equal matched reference impedances, −20 log10|S21| is the insertion loss in dB. If the assembly is mismatched, that result includes mismatch effects; inspect S11 and S22 rather than naming all missing transmission “dielectric loss.”
For an installed line accessible at only one end, a known open or short at the far end provides a useful approximate round-trip loss test. With a reasonably uniform, well-matched line and |Γend| near unity, measured positive return loss is approximately twice the one-way attenuation. A −12 dB S11 magnitude at the input therefore corresponds to about 6 dB one-way loss under those assumptions. Anritsu's cable-analysis guidance describes this reflected-signal method.
Disable distance-dependent loss compensation when extracting raw attenuation this way. A badly mismatched connector can dominate S11 before the signal reaches the far end. Strong ripple, multiple reflections, an uncertain termination or a reflection near the instrument floor defeats a simple divide-by-two interpretation. Compare open and short cases and use calibrated two-port testing or section substitution when the assumptions fail.
Keep frequency, length and temperature consistent with the manufacturer's attenuation data. A newly measured cable should be compared with its own installation baseline when possible. There is no single acceptable loss number for every cable, length and band.
TDR locates changes by their round-trip delay
A time-domain reflectometer launches a fast electrical step or pulse and observes returning reflections. A change in impedance sends some energy back. Its delay locates the event; its shape gives clues about the discontinuity.
d = c × VF × Δt / 2
Here d is one-way distance from the reference plane, c is the speed of light, VF is velocity factor and Δt is the reflected signal's round-trip delay. The factor of two is essential. With VF = 0.82, a reflection at 180 ns corresponds to approximately 22.1 m along the cable, including its routing and slack.
Use the value for the exact cable construction, or characterize a known length. If the real VF is 0.66 but 0.82 is entered, the displayed distance is about 24% too large. A mixed cable route has several propagation velocities; one global setting gives an electrical-distance estimate, not a reliable tape-measure location for every section. Wet regions can introduce additional delay uncertainty.
For a conventional signed step response referenced to the incident step, the basic interpretation is:
| Step-response feature | Ideal interpretation | Practical caution |
|---|---|---|
| Positive step toward Γ = +1 | Open circuit | Loss and bandwidth may prevent the displayed value reaching +1. |
| Negative step toward Γ = −1 | Short circuit | The first severe fault may hide everything beyond it. |
| Smaller upward change | Higher impedance | Could be geometry, contact resistance or a reactive feature; not uniquely a broken conductor. |
| Smaller downward change | Lower impedance | Could be geometry, excess capacitance or leakage; not uniquely water. |
| Brief bump or dip followed by recovery | A short discontinuity or reactive feature | Its apparent width is convolved with the instrument response. |
| Extended disturbed region | A distributed change | Loss, dispersion and overlapping reflections can also create slopes or broad features. |
The open/short sign convention is explained in Tektronix's TDR reflection guidance. Do not transfer these signs to a magnitude-only DTF graph: taking the magnitude discards polarity.
For a single ideal impedance transition, Z = Z0(1 + Γ)/(1 − Γ). Converting every point on a complicated lossy trace with that formula does not automatically recover a true local impedance profile. Earlier reflections alter the excitation reaching later features; multiple reflections may need a suitable reconstruction model.
A VNA can calculate a time trace from a frequency sweep
Many cable analyzers and VNAs, including suitable NanoVNA configurations, measure complex S11 versus frequency and transform it into a time or distance response. This is frequency-domain reflectometry, often presented as DTF or a TDR function. A log-magnitude-only SWR sweep lacks the phase needed for that transform. Anritsu describes the swept-frequency DTF approach.
A low-pass step transform is useful for impedance interpretation when the frequency sampling and extrapolation toward DC are appropriate. A bandpass impulse transform can examine a selected RF range and paths containing filters or DC blocks, but its magnitude response does not directly supply the familiar open-up/short-down step plot. Follow the transform requirements of the actual analyzer or software. Keysight distinguishes the transform modes; its DC extrapolation documentation explains why an incorrect DC estimate can produce a false slope.
Choose the sweep and transform first, then calibrate open/short/load at the end of the measurement jumper, where the feeder will attach. That point is distance zero. Include the adapters you actually use. Check the result with a known load and a known-length cable. Moving an imperfect jumper or adding adapters after calibration can reintroduce artifacts.
A phase-only port extension moves the apparent reference plane; it does not generally remove a jumper's mismatch, loss and dispersion. Do not use it to hide a troublesome first connector and then declare that connector good.
Resolution, range and accuracy are different
For a step TDR, an approximate spatial resolution scale is c × VF × tr/2, where tr is the effective rise time at the feature. A 2 ns transition and VF = 0.82 give about 0.25 m. Loss and dispersion slow the edge as it travels. The fastest rise time printed on the instrument does not apply unchanged at the far end of a long feeder. Tektronix discusses these resolution and cable-loss effects.
For an evenly sampled VNA sweep with span B, N points and frequency increment Δf = B/(N − 1), an approximate reflection alias-free distance is:
dmax ≈ c × VF /(2Δf).
Responses repeat at a delay of 1/Δf. Actual usable range may be shorter because of attenuation, receiver noise and the analyzer's selected time window. Increasing the number of frequency points at fixed span extends the unambiguous range. It does not provide the same resolution improvement as increasing bandwidth. Zooming or zero-padding only produces more display samples.
As a concrete calculation, take a 300 MHz span, 401 points and VF = 0.82. The frequency increment is 750 kHz and the reflection alias-free distance is about 164 m. Using Keysight's normal-window bandpass impulse coefficient, the response width is approximately 1.95/B: about 0.80 m expressed as one-way reflection distance. A 30 MHz span gives about 8.0 m with that same window. Those are illustrative settings, not guarantees for every analyzer or two unequal-sized faults. The window coefficients and range relationships are documented in Keysight's time-domain guide.
This explains why a narrow HF-only sweep may identify a distant cable end yet fail to separate two nearby connectors. Broaden the sweep only where the cable, calibration and intervening components support it. Use an in-band sweep separately to establish operating performance. A broadband locator trace and an operating-band acceptance test have different purposes.
Make the distance trace answer a physical question
- Use a known termination. A matched far-end load helps expose internal discontinuities. An intentional open or short establishes the cable end and checks length. These are disconnected analyzer tests, not transmitter operating tests.
- Keep raw data. Save S-parameters, span, point count, window, reference plane, VF and termination. Overlay future tests using the same settings.
- Control compensation. Cable-loss compensation estimates the reflection strength back at its location. A guessed loss profile can exaggerate distant features and noise. Keep the uncompensated trace too.
- Test from the opposite end when possible. A real feature should map to the corresponding place along the route, allowing for reference offsets and varying VF. Reversal also reduces the distance to a fault hidden behind a lossy section.
- Check suspicious late peaks. Re-reflections and aliases can resemble extra connectors. Change frequency spacing with valid calibration and compare positions. A genuine fixed feature should remain at approximately the same delay, although its shape can change with bandwidth.
- Respect masking. A severe early reflection or loss may leave too little energy to reveal downstream damage. A quiet trace beyond a break proves nothing about that section.
Time gating can help separate responses for analysis, but removing a peak from the display does not repair the physical component or restore energy it absorbed. Keep an ungated acceptance trace of the complete repaired assembly.
What water actually changes
Water may enter through a damaged jacket, an incorrectly assembled connector, an exposed cable end or a failed sealing interface. It can migrate along braid and other internal paths. Construction matters: a bonded foam dielectric, solid dielectric and air-spaced cable do not admit or retain water in identical ways. Belden's moisture guidance describes braid migration and corrosion extending beyond an exposed end; visible clean metal alone does not prove the remaining cable is dry.
The distributed circuit model makes the electrical consequences clear:
Zc = √[(R′ + jωL′)/(G′ + jωC′)]
γ = √[(R′ + jωL′)(G′ + jωC′)]
R′, L′, G′ and C′ are resistance, inductance, leakage conductance and capacitance per unit length. Water or ionic contamination in an electric-field region can change capacitance and dielectric loss. Corrosion can alter conductor and contact resistance. Damage need not occur uniformly, so both attenuation and reflections may change.
For a simple low-loss coax model, increased effective permittivity increases capacitance, reduces propagation velocity and lowers characteristic impedance when geometry and inductance stay approximately constant. A finite affected section can therefore produce an entry reflection and another at its exit. Real ingress may instead be gradual, confined near the shield, or mixed with corrosion: there is no universal “water-shaped” TDR trace.
Temperature and drying can change the symptom without restoring the assembly. A reading that improves in sunshine is evidence of environmental dependence, not proof that salts, damaged plating or corroded braid have recovered. Nonlinear contaminated contacts can also produce intermodulation under RF; a small-signal VNA sweep does not qualify that behavior.
Repair the cause, not just the displayed mismatch
| Confirmed condition | Repair decision | Verification |
|---|---|---|
| Loose or incorrectly fitted connector; cable remains sound | Reassemble with the correct cable-specific connector, preparation dimensions, tools and torque. Replace damaged contacts or parts. | Continuity, insulation, return loss, loss and repeatability after normal handling. |
| Moisture confined to a serviceable external interface | Inspect both mating halves and seals; clean/dry only by the manufacturer's permitted procedure. Replace corroded or compromised parts. | Show that the cable itself remains sound before resealing. |
| Water or corrosion has entered ordinary flexible coax | Replace the affected run when the extent cannot be established. A local cutback is defensible only when the remaining section is demonstrably sound and passes the required tests. | Test retained cable from both ends where practical, including attenuation. No fixed trim length guarantees recovery. |
| Crushed dielectric, kink, broken shield or burned/carbonized section | Replace the damaged section or run. Use a rated, correctly installed splice only when the application permits it. | Check the new joint and the full assembly; new splices add interfaces requiring support and sealing. |
| Suspected arcing or failure only at operating power | Remove the suspect assembly from service and inspect its voltage, connector and thermal limits. | Small-signal results alone are insufficient; return to power under controlled conditions after repair. |
I do not treat heating a wet feeder or blowing air through it as a general repair method. Drying can move moisture rather than establish its removal, and it cannot reverse corrosion. Belden warns that attempted drying can push moisture farther into a cable. Purpose-designed pressurized feeder systems have their own manufacturer maintenance procedures; that is a different construction from ordinary braided flexible coax.
After a surge, isolate the protector as another component in the chain. Passing DC continuity or a low-power RF test does not establish remaining protection capability. My surge-protector inspection article explains that separate decision.
Keep the repaired connection dry and mechanically stable
Weatherproof only a correctly assembled, clean, dry and electrically tested joint. Select an external sealing system for the actual connector geometry, cable diameters, jacket and environment. Follow its instructions for surface preparation, overlap, stretch and any protective overwrap. Cold-shrink sleeves and self-amalgamating systems each have specific fitting requirements; see my outdoor connector guide.
Provide separate strain relief, respect the cable's bend radius and arrange the route to shed water. A drip loop helps surface runoff but does not stop water already inside the cable. Keep required antenna vents and drains clear. Cap disconnected ends during storage and installation. Choose cable explicitly suitable for the exposure; an outdoor jacket does not automatically establish direct-burial or water-blocking capability.
If movement changes the complete antenna system but the isolated cable passes its tests, examine current on the coax exterior and the intended antenna return path. Common-mode current can make cable position affect feedpoint behavior. A choke may address that current path; it cannot restore a broken braid or remove water from the dielectric.
A repair is complete when the evidence agrees
My acceptance record would contain the cable identity and route length, connector/component list, calibration planes, termination, operating-band S11, a loss measurement, the saved time trace and photographs of the finished routing and seal. Record the actual conditions, not just “SWR good.” Compare against the assembly specification and the original baseline; there is no universal pass threshold for every station.
After disconnected tests pass and the measurement equipment is removed, return the system to operation within the ratings of every component. Increase power in controlled stages appropriate to the equipment and check stability over the intended duty cycle. Stop for arcing, odor, unusual heating or intermittent readings. Never handle a suspect connector while transmitting.
The practical conclusion is straightforward: use SWR to detect a mismatch, measure loss to expose hidden attenuation, use TDR or DTF to locate a change, and use inspection to decide the repair. Then correct the installation condition that caused the failure. A lower number at the radio is useful only when the feeder is also electrically sound and dependable.
Mini-FAQ
- Can damaged coax have a low SWR? Yes. Attenuation reduces the returning reflection and can hide a poor load match. Test cable loss as well as SWR; a good dummy-load match alone does not clear the feeder.
- Does a TDR trace prove there is water in the cable? No. It locates electrical changes. Moisture can change loss, impedance and delay, but similar features can arise from geometry, contacts and overlapping reflections. Correlate the trace with loss tests and physical inspection.
- Why is TDR distance divided by two? The measured reflection delay includes travel to the discontinuity and back. One-way distance is the speed of light multiplied by cable velocity factor and round-trip delay, divided by two.
- Is a VNA TDR function the same as a step reflectometer? A VNA normally transforms complex frequency measurements into a time response. A step reflectometer launches a fast edge directly. Both can locate discontinuities, but transform mode, bandwidth, calibration and display format determine how the trace should be interpreted.
- Should 50-ohm coax read 50 ohms on a multimeter? Not with its far end open. Characteristic impedance is a traveling-wave property. A connected 50-ohm termination produces approximately that DC resistance plus conductor resistance.
- Can I dry wet coax or cut off a fixed length? Neither is a general repair rule. Drying does not reverse corrosion or prove all moisture is removed. Retain a section only when its sound condition and electrical performance can be established; otherwise replace the affected run.
- What should pass after a repair? Check continuity and insulation as appropriate, operating-band match, attenuation, repeatable time-domain behavior and mechanical/sealing condition. Then verify operation within all component ratings at the intended duty cycle.