When Does CAT5 Start to Radiate?
When Does CAT5 Start to Radiate?
Twisted pair earns its quiet reputation through balance, not magic. When unwanted common-mode current reaches the cable, the complete Ethernet installation can become part of the transmitting—and receiving—structure.
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.
Ethernet is not silent because the jacket says CAT5e or CAT6. It is quiet when the transmitter, magnetics, connector, twisted pairs, terminations, power system and chassis preserve the intended differential path. Break that symmetry—or capacitively inject switching current into the cable as a whole—and the outside world sees a current distribution that can radiate.
This distinction matters around an SDR or low-noise HF station. Pulling one cable and watching the waterfall fall is a valuable clue, but it does not yet say whether the noise came from the Ethernet PHY, a PoE converter, a plug-in supply, the remote device, the cable’s common-mode current or an overloaded receiver.
Twisting Reduces Radiation; It Does Not Repeal Maxwell
BASE-T Ethernet uses balanced differential signalling. At a given cross-section, the wanted signal currents in a pair are approximately equal and opposite. Tight, regular twisting keeps the conductors close and continually exchanges their position relative to external fields. That reduces the loop area and helps the fields cancel at distance.
The cancellation is never mathematically perfect. Pair geometry, connector transitions, unequal source and load impedances, PCB routing, magnetics, termination components and nearby metal all affect longitudinal balance. Part of the differential signal or internal equipment noise can therefore convert to common mode. Conversely, an external common-mode disturbance can convert back into a differential voltage at the receiver.
There is no single cable length at which CAT5 suddenly becomes an antenna. Radiation depends on common-mode current amplitude and phase, the distributed return path, electrical length, routing, height, nearby conductors and frequency. Peaks can move when a cable is rerouted or when another lead is connected because the entire common-mode network changed.
PoE Does Not Automatically Destroy the Balance
Power over Ethernet deliberately applies power in common mode with respect to the conductors of each data pair. The power extraction and injection networks separate that common-mode power from the differential data. A standards-compliant implementation is designed to do this without turning the wanted data pair into an unbalanced RF radiator.
That does not make every assembled PoE link quiet. The PSE, powered device and their converters can produce fast switching edges. Parasitic capacitance across an isolated converter, intentional EMI capacitors, unequal pair resistance, imperfect magnetics, connector geometry and chassis coupling can drive common-mode current onto the cable. A non-PoE Ethernet device can do the same; PoE simply adds more circuitry and more possible coupling paths.
| Observation | What it supports | What it does not prove |
|---|---|---|
| Noise appears when the Ethernet cable is connected | The assembled link changed the source, path or receiver condition | That the cable itself generated the noise |
| Noise follows PoE power but not data activity | The power conversion or its coupling path deserves priority | That every PoE implementation is noisy |
| A comb changes with link speed or traffic | A clocked digital process is involved | That comb spacing identifies common mode or a specific PHY clock |
| A whole-cable clamp shows current at the same frequencies | Net/common-mode cable current is present at that position | That it is the only radiating path or the original generator |
| Input attenuation or preselection collapses many spurs | Receiver overload or mixing may contribute | That the external source has disappeared |
The Twelve-Metre Cable Was the Clue, Not the Verdict
One RF.Guru station investigation began with a twelve-metre CAT5e run between a PoE switch and a remote SDR. With that metallic link assembled, the HF waterfall filled with noise. Replacing the long copper segment with fibre and treating the terminal equipment changed the observation dramatically.
The important lesson was not “all PoE is bad” or “fibre cures everything.” The change implicated the assembled copper-and-power path. The fibre removed one galvanic conductor, while the different terminal arrangement also changed supplies, chassis currents and cable geometry. That is why a convincing investigation separates those variables instead of assigning the result to the jacket label.
A Noise Comb Is a Clock Clue, Not a Mode Meter
Regularly spaced lines suggest a periodic process, its harmonics, modulation sidebands or nonlinear mixing. Ethernet electronics, switch-mode converters, display clocks and microcontrollers can all create such patterns. The spacing seen at one station cannot be promoted into a universal Ethernet signature.
Common mode describes how current flows on the conductors; it does not describe why the spectrum is evenly spaced. First find the fundamental or repetition mechanism. Then use current and field measurements to find how that energy reaches the antenna or receiver.
Shielding and Bonding Need a Complete Design
Shielded twisted pair can reduce coupling when connectors, equipment bonding and the installation form a continuous high-frequency shield system. A floating pigtail or an unbonded connector can make the screen ineffective at the frequencies of interest. Poor equipotential bonding can also allow unwanted low-frequency current or hazardous potential differences.
The popular instruction to bond every data-cable shield at one end is not a universal HF-EMC rule. High-frequency installations commonly bond shields at both equipment ends so the shield remains effective over the channel; that assumes the equipment and building bonding system are designed for it. Use the connector and equipment manufacturer’s bonding scheme and the applicable cabling and electrical rules. Never disconnect protective earth to improve a noise trace.
Chokes and Fibre Are Tools, Not Incantations
A common-mode choke adds frequency-dependent complex impedance to the unwanted external-current path. Its effect depends on material, turns, winding capacitance, cable bundle, current, position and the impedance already present. Blindly specifying three or six turns says little without an impedance sweep and an installed current comparison. Do not place ferrite around individual conductors of a data pair or modify certified PoE wiring.
Fibre removes the metallic data path across the fibre span, which is often extremely useful. The media converters and their power supplies still contain oscillators and conductors, so they can radiate locally or couple through power wiring. Place and power them as part of the same source–path–victim investigation.
Make the Cable Confess with Repeatable Tests
- Freeze the receiver state. Record frequency, bandwidth, detector, averaging, attenuation, gain, AGC, preselection and antenna selection. Compare wanted-signal SNR as well as the displayed floor.
- Change one link state. Compare cable disconnected, cable connected without PoE, PoE powered with data idle, and representative traffic where the equipment permits safe testing.
- Separate the supplies. If the equipment supports it, compare known suitable supplies without defeating protective earth, shielding or mandatory bonding.
- Map the whole-cable current. Clamp around the complete Ethernet cable at several positions and correlate its spectrum with the receiver. A single point can sit near a current minimum.
- Localise the generator. Use a near-field probe around the switch, injector, powered device, DC/DC section, connector and attached leads at low disturbance level.
- Check receiver linearity. Insert known attenuation or suitable preselection. If many lines collapse disproportionately, the receiver is contributing products.
- Test one remedy at a time. Route, choke, bond, substitute or convert to fibre, then restore the baseline and repeat. Confirm that the wanted signal was not reduced along with the noise.
Practical Conclusion
CAT5 does not begin to radiate because PoE adds DC, because a cable reaches a magic fraction of a wavelength or because the label says unshielded. It becomes relevant when the complete installation supports unwanted current and provides a return path that lets that current form an effective radiator.
So keep the blunt version: it is not the cable name; it is the current on the cable. Then do the engineering work needed to identify who generated that current, how it reached the link, where it returned and whether the receiver was still linear.
Primary Standards and Engineering Guidance
- IEEE 802.3 Working Group archive — Ethernet and Power over Ethernet standards lineage
- Ethernet Alliance — PoE and PoDL standards explained
- Texas Instruments SNLA107A — Sources of EMI in Ethernet Applications
- Texas Instruments SLUA454 — Practical Guidelines to Designing an EMI-Compliant PoE Powered Device
- Cisco CPwE Physical Infrastructure — shield bonding and network-cabling EMC
- CISPR 32 — multimedia-equipment emission requirements
Mini-FAQ
- Does every Ethernet cable radiate? Every time-varying current produces a field, but a well-balanced twisted pair keeps external radiation very small. Troublesome radiation usually needs mode conversion, common-mode current or coupling through the attached equipment.
- Does PoE automatically unbalance Ethernet? No. PoE intentionally applies power in common mode while preserving differential data. Converter noise, parasitic coupling, resistance imbalance, magnetics and chassis paths can still drive unwanted cable current.
- Do evenly spaced birdies prove Ethernet common mode? No. A comb supports a periodic or nonlinear mechanism. Current probing, source-state tests and receiver-linearity checks are needed to identify the coupling mode.
- Should a shielded Ethernet cable be bonded at only one end? Not as a universal rule. High-frequency shield performance often uses bonds at both equipment ends within a proper equipotential system. Follow the equipment, cabling and electrical design.
- Will ferrite fix a noisy PoE link? It may reduce a measured common-mode current if its impedance and position suit the installed path. It cannot correct every converter, chassis, radiated-field or receiver-overload problem.
- Does fibre guarantee a quiet receiver? Fibre removes one metallic data path, but its converters and supplies can still generate or radiate noise. Verify the complete replacement with fixed-setting SNR and restored-baseline tests.