A cave is one of the most hostile radio environments an engineer can imagine, not because it is electrically noisy in the ordinary urban sense, but because it is almost brutally indifferent to the kind of radio systems modern people instinctively trust. Take a handheld VHF radio underground and it quickly becomes a talisman rather than a communications device. Cellular networks vanish almost immediately. Wi-Fi, Bluetooth, LoRa, and most public-safety radios behave as if the mountain has swallowed the electromagnetic spectrum whole. The problem is not simply distance. It is geology, conductivity, wavelength, antenna size, absorption, moisture, geometry, and the inconvenient fact that most practical radios were designed for air, towers, rooftops, vehicles, satellites, or at least open space—not for a rescuer wedged in a limestone passage with a hundred metres of rock between their antenna and the surface.
That is why a recent Canadian cave-communication test involving QRP Labs QDX-M transceivers, the JS8Call digital mode, and the 2200 metre amateur band is so technically interesting. On paper, the ingredients sound almost modest: roughly five watts of transmitter power, a compact digital-mode transceiver, laptop or mobile computing support, and a low-speed text mode derived from the weak-signal world of amateur radio. Yet the reported result—two-way text communication through more than 100 metres of solid rock, with signal-to-noise ratios around +9 dB in successful tests, and practical use in the Rats Nest Cave rescue environment—touches a much larger story. It is a story about how radio stops being “radio” in the everyday sense and becomes a carefully exploited electromagnetic coupling problem. It is also a reminder that, in extreme communications, raw bandwidth is often far less valuable than getting one short, reliable sentence through when nothing else works.
The headline version is tempting: five watts talks through a mountain. The real engineering is subtler and more impressive. A conventional 5-watt handheld on VHF can travel astonishing distances when both antennas can see the same horizon, yet become useless after a few bends in a cave passage. A 5-watt transmitter near 137 kHz, by contrast, is not winning by blasting a high-frequency wave through stone like a laser through glass. It is operating in a regime where the wavelength is so enormous, the antennas so electrically small, and the distances so deep inside the near field that familiar assumptions about propagation begin to fail. The system is not trying to create a normal far-field radio link. It is coupling energy through the ground by means of low-frequency magnetic and conductive fields, then using a digital mode that can dig intelligible messages out of signals that would make voice communication impractical or impossible.
For cave rescue, that distinction matters. A trapped person may not need a video stream, a phone call, or a broadband data pipe. They may need to say where they are, whether they are injured, whether water is rising, whether they can move, whether they can hear rescuers, or whether a rope system has reached the right chamber. Rescuers may need to coordinate medical information, rigging instructions, team movement, and evacuation timing without forcing runners to spend hours moving between underground and surface control. In that environment, a slow text channel is not a compromise in the consumer sense. It can be the difference between guessing and knowing.
The mountain as a radio component
Most people learn radio through the mental image of waves traveling outward from an antenna. The transmitter launches electromagnetic energy, the receiver intercepts a small fraction of it, and the link budget accounts for path loss, antenna gain, noise, modulation, receiver sensitivity, and margins. That model works beautifully for many ordinary systems, from FM broadcast and airband to Wi-Fi and satellite links. But caves punish ordinary link budgets because rock, soil, and water are not empty space. They are lossy, irregular dielectric and conductive media. Their electrical properties change with mineral content, fractures, moisture, dissolved salts, clay layers, ore bodies, and even seasonal hydrology. A radio wave entering that environment does not merely spread out; it is absorbed, scattered, refracted, and attenuated.
At VHF and UHF, the situation is especially unforgiving. Wavelengths are short enough to be convenient for portable antennas, but short wavelengths interact strongly with cave geometry and conductive losses. A handheld radio at 144 MHz has a wavelength of about two metres; at 440 MHz, less than a metre. Those waves can sometimes sneak along tunnels, reflect around corners, or travel through air-filled passages in a waveguide-like fashion, especially in large, dry, relatively straight sections. But when the task is to communicate through the ceiling of a cave to the surface, or through a bulk of rock rather than along an open passage, attenuation is severe. The signal is not politely delayed. It is converted into heat and lost.
Low frequency changes the problem. The 2200 metre amateur band, centred around the 135.7–137.8 kHz region in many jurisdictions, has a wavelength on the order of two kilometres. That wavelength is absurdly large compared with a person, a rescue pack, a cave chamber, or a loop antenna that can be carried underground. In free space, an efficient quarter-wave antenna for that band would be hundreds of metres long. Nobody is dragging that through a crawlway. But through-the-earth communication does not depend on building a textbook efficient radiator. Instead, it often uses small loop antennas or earth-electrode systems that are extremely inefficient as far-field radiators but useful as local field sources. In a cave rescue, inefficiency can be acceptable if the path is short enough, the frequency is low enough, the receiver is sensitive enough, and the modulation is patient enough.
This is where near-field thinking becomes essential. Close to an antenna, the electric and magnetic fields are not arranged in the clean, self-sustaining wavefront familiar from far-field radio. The relationship between the E-field and H-field is different, the energy storage around the antenna is significant, and coupling can occur by mechanisms that look more like induction or conductive current flow than ordinary radiation. A loop antenna driven at low frequency produces a magnetic field. Another loop can detect the changing magnetic flux. Earth electrodes can inject current into the ground, allowing a receiver to detect potential differences caused by conductive propagation. The exact balance of mechanisms depends on antenna type, orientation, ground conductivity, depth, geology, and frequency. Cavers have been exploiting these effects for decades, often with equipment that looks strange to radio amateurs accustomed to resonant dipoles and coax-fed antennas.
A hundred metres of rock is not a trivial barrier, but it is also not beyond the history of cave radio. Older systems such as the Molefone, HeyPhone, Nicola system, and various military or mining through-the-earth radios showed long ago that low-frequency communication through rock is possible. Some used single-sideband voice; others used beaconing, radiolocation, or earth-current techniques. What makes the QDX-M and JS8Call combination intriguing is not that it is the first system to penetrate rock. It is that it brings a modern, low-cost, compact, software-assisted digital architecture to a problem traditionally served by specialized gear, volunteer-built equipment, or heavy rescue-specific systems.
That shift mirrors a broader pattern in radio engineering. Many once-exotic capabilities become practical when a difficult analog problem is partly converted into a digital signal-processing problem. Weak-signal amateur modes are a perfect example. FT8, WSPR, JT65, and related modes did not repeal physics; they changed the trade space. Instead of requiring a human ear to copy Morse or voice in real time, they use structured transmissions, narrow bandwidth, time synchronization, forward error tolerance, and coherent decoding to recover information at very low signal-to-noise ratios. JS8Call inherits that lineage but adds a messaging layer that feels more like slow tactical chat than a contest exchange. In a cave, where the channel is narrow, slow, and hostile, that is exactly the kind of bargain that begins to make sense.
The strange part is that the system’s weakness is also its strength. It cannot offer ordinary speech quality. It cannot carry arbitrary high-speed data. It may require setup discipline, frequency coordination, batteries, antenna deployment, and operator knowledge. But the same narrowness that makes it slow also makes it resilient. A voice channel spreads human speech over enough bandwidth that the receiver must preserve a great deal of information moment by moment. JS8Call only needs to move symbols slowly and reliably. A few tens of characters, delivered with confidence, may be enough to transform a rescue operation.
From cave telephones to weak-signal text
Cave communication has always been a field of compromises. The simplest system is still the human runner: a person physically carries messages between teams. Runners are robust in the sense that they do not depend on electronics, but they consume time, energy, and personnel, and they can become a safety risk themselves. Wired cave telephones solve some of that by laying cable through passages, but cable is heavy, vulnerable to abrasion and water, difficult to deploy in complex terrain, and not always available where the emergency happens. In long systems with tight crawls, vertical pitches, sumps, mud, breakdown chambers, and multiple branches, the act of installing communications infrastructure can become an operation inside the operation.
Radio promised freedom from wire, but ordinary radio did not solve the underground problem cleanly. In some mines and transport tunnels, leaky feeder systems work well: a coaxial cable with controlled leakage acts like a distributed antenna along the tunnel. But that is infrastructure, not an emergency throw-in solution for wild caves. Mesh radios can function underground when enough nodes are placed to maintain line-of-sight or passage-following links, but they require deployment through the route and may fail across rock barriers. VHF cave links can work opportunistically along passages, and low-power devices can be excellent for local team coordination. But surface-to-cave communication demands something else.
That “something else” emerged historically from low-frequency induction and earth-current systems. Early trench communications in wartime used ground conduction because wires were easily destroyed. Mining communication systems explored ultra-low and very-low-frequency techniques because higher frequencies would not penetrate. Cave-radio experimenters adapted those ideas using loop antennas, audio or low-RF frequencies, single-sideband modulation, and portable receivers. The goal was never to win a spectral-efficiency contest. It was to make contact through a lossy medium with antennas that human beings could carry.
The classic loop-based cave radio is almost a paradox. A loop one metre across at tens or hundreds of kilohertz is electrically tiny. As a radiator, it is terrible. Radiation resistance is minuscule, loss resistance dominates, and most transmitter power becomes heat rather than useful far-field signal. But in near-field magnetic coupling, the loop does not need to behave like a broadcast antenna. It needs to create a magnetic field strong enough that another loop, some distance away and separated by rock, can detect a changing flux above its noise floor. Orientation matters greatly. Two loops couple best when their magnetic axes are aligned; rotate one loop poorly and the signal can drop dramatically. Rock geometry matters too, because the field is not traveling through a uniform laboratory slab but through a messy geological volume.
Earth-current systems take a different approach. Instead of relying primarily on magnetic induction between loops, they use electrodes placed in the ground to drive currents through the earth. A receiving station uses its own electrode pair to detect voltage differences. In some geological conditions, this can outperform loops; in others, electrode contact resistance, dryness, layout constraints, and conductive inhomogeneity become problems. Rescue teams often care less about theoretical elegance than deployability. Can the antenna be carried by tired people? Can it be installed in mud or on rock? Does it require long wires across a passage where rescuers will trip over them? Can a wet, cold operator make it work under stress?
The QDX-M approach sits in a useful middle ground because it borrows from amateur digital radio rather than requiring a wholly bespoke cave-radio architecture. The QDX-M is a monoband digital transceiver with an embedded software-defined receiver, built-in USB audio interface, CAT control, stable frequency synthesis, and support for single-tone FSK digital modes. For normal amateur use, that makes it a compact appliance for modes such as FT8 and JS8Call. For cave work, the significance is different: it packages the RF generation, receiver, sound-card interface, and computer control into a small, inexpensive unit that can be configured for the 2200 metre band.
That matters because practical rescue equipment lives or dies by logistics. A beautiful laboratory prototype that needs delicate alignment, bench instruments, mains power, and a PhD student is not a rescue tool. A small box that can be bought or built affordably, powered from field batteries, connected by USB, and driven by widely available software is much closer to something volunteer teams can experiment with, train on, and adapt. It also benefits from the amateur-radio ecosystem: people already understand digital-mode operation, timing, signal reports, weak-signal decoding, antennas, and improvised field deployment. Cave rescue is specialized, but it should not have to reinvent every part of the communications stack.
JS8Call is a particularly apt partner because it was designed around message passing under weak-signal conditions. FT8 is extraordinarily good at making brief structured contacts, but it is not a conversational emergency messaging system. JS8Call retains the weak-signal DNA while adding directed messages, acknowledgements, stored messages, relays, heartbeat-style awareness, and free-text conversation. It is not fast, especially in its slower and more robust modes, but speed is not the only metric. In a surface-to-cave path, where the channel may be stable but weak, narrowband text becomes attractive because it converts a bad communications path into a tolerable human workflow: type, send, wait, decode, acknowledge.
The reported +9 dB signal-to-noise ratio through more than 100 metres of rock is therefore more than a bragging number. In weak-signal digital modes, SNR is measured in a narrow reference bandwidth and interpreted in the context of the decoder, not like the full-channel audio SNR of a voice circuit. A +9 dB report suggests a comfortable margin for that test configuration, especially compared with modes that routinely decode below the noise in human-audible terms. But it should not be misread as a universal guarantee. Move the antennas, rotate a loop, change the rock conductivity, add electrical noise, alter the depth, or operate in a different cave system, and the margin may change dramatically. The impressive part is not that 100 metres is a magical threshold. It is that a cheap five-watt digital station produced enough margin to support usable two-way messaging in a real cave-rescue context.
Why 2200 metres behaves differently
The 2200 metre band sounds archaic to anyone raised on microwave networks and gigahertz processors. Its frequency is lower than long-wave broadcast, far below medium-wave AM, and almost unimaginably below the VHF and UHF bands used by handheld radios. Yet the reasons it looks old-fashioned are the same reasons it becomes useful underground. At very low radio frequencies, attenuation in conductive material is governed by skin depth, a concept that describes how deeply electromagnetic fields penetrate into a conductor before decaying significantly. Rock is not copper, but neither is it free space. Its conductivity is low compared with metals and highly variable, yet sufficient to make high-frequency penetration poor. Lowering frequency increases skin depth, allowing fields to extend farther into lossy media.
The simplified skin-depth equation says penetration improves as frequency decreases and as conductivity and magnetic permeability decrease. Real caves complicate this because the “rock” is not a uniform conductor. Limestone, dolomite, sandstone, shale, granite, clay seams, mineralized zones, groundwater, and fractured voids all behave differently. Water content can help or hurt depending on whether the mechanism is magnetic induction or conductive coupling. Salty water increases conductivity, which may make earth-current coupling possible but can also increase losses. Dry, resistive rock may attenuate less but make electrode coupling harder. A cave system is not a clean RF channel; it is a three-dimensional analog circuit drawn by geology.
At 137 kHz, the free-space wavelength is roughly 2.19 kilometres. A 100 metre rock path is only a small fraction of a wavelength, which reinforces the near-field character of the link. In that regime, the field strength does not follow the same simple inverse-distance behavior as far-field radiation, and antenna orientation can dominate performance. A loop antenna may produce a magnetic dipole field whose strength falls rapidly with distance in the near field, but because the receiver is close in wavelength terms, the system can still work over rescue-relevant distances if losses and noise are manageable. The phrase “through solid rock” is accurate, but it can encourage the wrong picture. The signal is not a narrow beam drilling through a wall. It is a low-frequency field distribution coupling between two stations through a conductive, lossy, irregular volume.
Antenna design becomes the central practical problem. On 2200 metres, no portable cave antenna is electrically efficient in the normal amateur sense. A small loop must balance inductance, resistance, tuning capacitance, voltage, current, mechanical ruggedness, and portability. High circulating currents can produce useful magnetic fields, but copper losses rise, connectors matter, and tuning becomes sharp. The loop may need to be laid on the ground, propped against a wall, or oriented to maximize coupling with the surface station. An earth-electrode antenna may require wire runs and good contact points, which may be easy in wet soil and miserable on dry rock. In rescue conditions, a theoretically better antenna that takes thirty minutes longer to deploy may be operationally worse than a lower-performing one that can be thrown down and tuned quickly.
The transmitter also faces unusual stresses. Digital weak-signal modes often operate at high duty cycle: unlike speech, where average power is much lower than peak envelope power, modes such as FT8 and JS8Call can ask the transmitter to deliver continuous RF for the duration of each transmission. A five-watt rating is therefore not the same as a five-watt voice peak rating. The power amplifier, filters, switching devices, and thermal path must tolerate sustained operation.
Top comments (0)