Analog RF over Fibre vs Coax: Move the Noise Boundary
Analog RF over Fibre vs Coax: Move the Noise Boundary
An optical span can remove a conductive RF path between an antenna site and the shack. It cannot make the converters, power supplies, antenna or receiver disappear from the engineering.
I like analog RF over fibre because it forces us to draw the receive boundary honestly. If unwanted current is travelling on the feedline exterior or through a station-to-antenna ground connection, a genuinely dielectric span can remove that route. If the noise already entered through the antenna, front end or local electric field, changing the transport medium will not recover the lost SNR.
The useful distinction: fibre can remove a conductive path. It does not provide zero noise, zero distortion, zero delay change or automatic array-grade phase tracking. Coax can carry unwanted exterior current, but a correctly terminated coaxial line is not intrinsically an exposed noise wire.
Start with the Coupling Path
A coaxial receive feedline supports the intended differential transmission-line mode between centre conductor and the inner surface of the shield. It can also support current on the shield exterior relative to the surroundings. Those are different modes with different return paths.
Noise may reach the receiver through the antenna pattern, direct electric- or magnetic-field coupling into the front end, common-mode current on the feedline exterior, a ground-potential difference, a control cable, a shared power supply or receiver overload. Only some of those paths are removed by substituting fibre for part of the coax.
That is why “coax picks up QRM” is not a sufficient diagnosis. Measure exterior current, disconnect or reroute candidate conductors, isolate control and power paths, and compare the receive noise with fixed bandwidth, gain and attenuation. The medium should be chosen after the unwanted path is identified.
What an Optical Span Changes
An analog radio-over-fibre link contains an electrical-to-optical converter, an optical path and an optical-to-electrical converter. A fully dielectric fibre with no metallic strength member cannot carry DC or RF common-mode current between those endpoints. It can therefore provide real galvanic separation for the signal transport.
The endpoints remain electronic RF devices. Their enclosures, connectors, power supplies, control interfaces and local grounds can collect or generate interference. Remote power delivered over copper, a metallic-armoured fibre, USB control or a shared shield can quietly recreate the path the optical span was meant to remove.
Fibre also does not transport DC power to an active antenna. A remote optical transmitter and any antenna preamplifier need local power, filtering, surge strategy and strong-signal headroom. Those practical details often decide whether the optical link improves the station.
Compare Complete Links, Not Metres of Cable
| Engineering question | Coaxial link | Analog optical link |
|---|---|---|
| Signal path | Passive transmission line when no active equalisation or preamplifier is added | Active E/O and O/E conversion plus the optical path |
| Matched attenuation | Depends on cable type, length, frequency, temperature, connectors and mismatch | Depends on converter gain, coupling, optical power, fibre loss, bends and connector condition |
| Added noise | Passive loss raises system noise factor according to its position in the cascade | Laser or LED noise, shot noise, detector and amplifier noise contribute to the link noise figure |
| Large-signal behaviour | Usually highly linear until connectors, corrosion, protection parts or following electronics dominate | Modulator, optical source, detector and RF amplifiers set compression and intermodulation limits |
| Conductive common mode | Shield exterior can be part of an unintended antenna or ground-current path | A genuinely dielectric span removes that path between isolated endpoints |
| Phase and delay | Electrical length changes with cable construction, temperature and routing | Group delay and phase change with converters, fibre, optical source, temperature and operating point |
| Remote power | Can share the cable through a Bias-T where the system is designed for it | Requires a separate isolated power solution at the remote end |
| Cost boundary | Cable, connectors, grounding, chokes and any remote preamplifier | Converters, power, enclosures, optical connectors, calibration and fibre—not fibre price alone |
The Link Budget Has More Than Loss
For weak-signal HF reception, start with cascaded gain and noise factor. If the antenna or remote preamplifier provides gain G1, the noise contribution of the following link is divided by that gain in the Friis relationship:
Ftotal = F1 + (F2 − 1) / G1 + …More front-end gain is not automatically better. Strong broadcast signals, local transmitters and impulsive noise can drive the antenna amplifier or optical transmitter into compression or intermodulation long before the wanted signal becomes receiver-noise limited.
An analog optical specification therefore needs frequency response, net RF gain, input-referred noise, noise figure, input and output compression, second- and third-order intercept behaviour, and spurious-free dynamic range in a declared bandwidth. A quoted SFDR normalized to 1 Hz cannot be copied directly into a receiver bandwidth; for third-order distortion the available range decreases with the two-thirds power of measurement bandwidth.
“Less than a tenth of a decibel” is not a system claim: the optical fibre may have very little loss over a short span while the E/O and O/E conversion has substantial net loss, gain, noise or distortion. Measure the RF input-to-output link at declared reference planes.
Phase Is Preserved Only Within a Specification
An analog optical link can reproduce the instantaneous RF waveform, including phase, but that does not make two links phase matched. Arrays need channel-to-channel amplitude, phase, group-delay and temperature tracking across the operating band. Optical source bias, converter filters, fibre length, component ageing and power-supply variation can all move the result.
A digital transport can also preserve amplitude and phase when sampling clocks, bit depth, anti-alias filtering, channel synchronization, latency and reconstruction are engineered for the job. The choice between analog and digital transport is therefore a system decision, not a claim that one medium uniquely preserves “real RF.”
For diversity selection, absolute phase may not matter. For coherent combining or beamforming, it does. State the reference planes, inject a common signal, calibrate the full channels and monitor drift rather than trusting identical labels on two modules.
Place the Conversion Where It Breaks the Real Path
Putting fibre in the last metres before the shack helps only when those metres, their shield connections or the station ground form the significant coupling route. If the coax exterior becomes an antenna along the full run, the optical break may belong nearer the antenna. If the dominant QRM is radiated directly into the antenna, better siting, pattern control, filtering or source suppression comes first.
A hybrid coax-and-fibre system can be sensible, but the transition point comes from the current map and link budget—not a universal distance. Coax may be the simpler and quieter choice for a short, well-routed link. Fibre may be decisive where galvanic separation removes a verified coupling path. Digitizing at the antenna may be preferable when calibrated multichannel sampling and network transport solve the actual requirement.
The Station-Level Decision
Choose coax when its passive loss, shielding, exterior-current control and grounding fit the site. Choose an optical span when breaking the conductive path is worth the converter noise, linearity, power and calibration work. Choose neither by comparing cable prices or quoting fibre attenuation without its electronics.
The most useful optical-link claim is modest and powerful: a dielectric span moves the electrical boundary. Everything after that—SNR, overload, phase tracking, array performance and QRM improvement—belongs to the measured complete system.
Bottom line: analog RF over fibre can solve a real common-mode or ground-coupling problem. It earns its place only when the RF link budget and the unwanted-current path are both better than the coaxial alternative.
Primary technical references
- ITU-T G Supplement 55 — Radio-over-Fibre Technologies and Applications
- ITU-R F.1332-1 — Radio-Frequency Signal Transport through Optical Fibres
- NIST SP 1024 — Optical-Fibre Measurements, Including RoF Dynamic Range
- NASA/JPL — Microwave Analog Fibre-Optic Link for the Deep Space Network
- NIST Technical Note 1095 — Shielding Effectiveness and Surface Transfer Impedance
- NASA GSFC-STD-7000B — EMC Application Guide and Cable Common Mode
Mini-FAQ
- Is fibre immune to RF interference? The dielectric span does not conduct RF common-mode current, but the optical converters, power supplies, enclosures and connected cables can still couple interference.
- Does fibre always improve HF SNR? No. It helps when it removes a significant coupling path without adding too much link noise or distortion. Noise already captured by the antenna remains.
- Is analog RF over fibre lossless over a short run? No. Fibre attenuation may be small, but complete RF gain or loss includes both converters, optical coupling, connectors and the detector.
- Can an analog optical link carry array phase? It carries waveform phase, but coherent arrays require measured channel-to-channel phase, amplitude and delay tracking over frequency and temperature.
- Should fibre replace only the last metres near the shack? Only if those metres form the unwanted path. Place the optical break where the diagnosed common-mode or ground connection is actually interrupted.
- Does fibre remove lightning and grounding requirements? A dielectric span removes one conductive route, but outdoor antennas, endpoint electronics, local power and other cables still require an appropriate protection and bonding design.