Login Start free Dashboard Logout

Leaky Feeder Design

3D Leaky Feeder Coverage

Predict radiating cable (leaky feeder) radio coverage in a full 3D underground mine. Bring your own mine — import centreline geometry from DXF, CSV, KML/GeoJSON, or draw drifts, declines, and stopes directly in the scene — auto-route the cable or draw your own runs and drop power splitters by hand, run a duplex channel plan that evaluates coverage on the worst traffic (voice) carrier, place inline amplifiers, and read a floor-level coverage heatmap with live ITU-R UTD corner diffraction and corridor decay into the non-cabled drifts. Downlink, uplink, and worst-of-link modes at a chosen reliability percentile, per-level slicing, a three-level hover, and per-tier coverage statistics, all in the browser. Currently in Beta.

Open tool → Create free account
Built-in multi-level mine model (L1 through L7) with the auto-routed leaky feeder cable.
Built-in multi-level mine model (L1 through L7) with the auto-routed leaky feeder cable.
Manual cable design. Draw runs, drop 2/3/4-way splitters that break the run into loss-accounted segments, and set the headend.
Manual cable design. Draw runs, drop 2/3/4-way splitters that break the run into loss-accounted segments, and set the headend.
Place amp mode with the position preview, plus one-click auto-spacing along every run.
Place amp mode with the position preview, plus one-click auto-spacing along every run.
Downlink voice-carrier heatmap with the per-tier coverage statistics at the chosen reliability percentile.
Downlink voice-carrier heatmap with the per-tier coverage statistics at the chosen reliability percentile.
Uplink and worst-of-link modes showing the talk-back picture per point.
Uplink and worst-of-link modes showing the talk-back picture per point.

Overview

Radiating cable, more commonly called leaky feeder, is the dominant radio coverage method underground. The cable runs the length of the drifts and leaks RF energy continuously, acting as a long distributed antenna so a handheld has coverage wherever the cable goes. The hard part of the design is everything off the cable: how far the signal diffracts into a stope access stub or an ore-zone crosscut that carries no cable, where the along-cable loss has accumulated enough to need another inline amplifier, and whether a worker at the worst-case point can not only hear the base station but reach it back. A single-run 2D cascade cannot express the level-by-level loss accumulation, the splitter dilution, and the stub shadowing of a real multi-level mine.

The noIM₃ 3D Leaky Feeder Coverage Predictor is a browser-based 3D tool for exactly that picture. The geometry is yours: import a real mine’s centreline strings from a 3D DXF export (Surpac, Vulcan, Deswik, AutoCAD), waypoint CSV, KML/GeoJSON, or the native JSON format; or draw drifts on a level plane, generate spiral and straight declines, raises, shafts, and ore passes between levels, and place stopes by footprint directly in the scene; or start from the built-in seven-level demo mine. Every geometry path runs through the same validation pipeline — junction welding, cable-direction re-orientation, level derivation, connectivity and loop checks, survey-grid recentring — with a full report before anything is accepted.

The radiating cable is yours to lay. Keep the auto-route through the primary access drift and each level main (respecting the no-cable annotations on raises, ore passes, shafts, and production stopes), or switch to manual mode and draw your own runs, drop 2/3/4-way power splitters that break each run into loss-accounted segments, and set the headend where the feed connects. Coverage runs off a duplex channel plan — a +0 dBm AGC pilot plus your downlink and uplink traffic carriers — evaluated on the worst channel per direction at the voice-carrier level (the level a worker actually talks on, backed off below the pilot the amplifiers hold). Place inline amplifiers from a real vendor catalogue by hand or fill every run in one click with auto-spacing, and the AGC cascade adapts through every splitter. Coverage renders as a floor-level heatmap ribbon along every drift — cabled drifts combine the along-cable cascade with the radial coupling loss and operator body loss, while non-cabled drifts are fed by live ITU-R P.526-16 UTD corner diffraction plus ITU-R P.1238-13 corridor decay — classified at a chosen reliability percentile. Downlink, uplink, and worst-of-link modes, per-level slicing, a three-level hover (DL voice, UL voice, pilot), and live per-tier statistics complete the picture.

The tool is in Beta. The RF maths — the along-cable cascade, the AGC amplifier model, the splitter port-loss accounting, the uplink walk-back and talk-back budget, the UTD corner diffraction, the corridor decay, operator body loss, and the sensitivity-anchored tier classification — is exercised by a unit-test suite against closed-form expectations and the ITU-R recommendations. Known limitations (a tree cable model with loop disclosure, coupling-loss clamping above a cable’s characterised band, and no full vendor splitter/tap insertion-loss data yet) are stated plainly below. For the 2D installation cascade with full vendor data and a bill of materials, use the companion 2D Leaky Feeder Designer.

Capabilities

Bring your own mine — import, draw, or start from the demo

Import a real mine’s centreline geometry through a validating wizard: 3D DXF strings (LINE / LWPOLYLINE / POLYLINE-VERTEX as Surpac, Vulcan, Deswik, and Micromine export, with $INSUNITS detection and a per-layer mapping step assigning drift types, no-cable flags, and stope roles), waypoint CSV with header auto-detection, KML and GeoJSON LineStrings (altitudes used when present), and the native noim3-mine JSON. The pipeline welds junctions to the router tolerance, inserts tees, re-orients and splits drifts so the cable cascade always runs the right way, derives the level table, checks connectivity and loops, and recentres survey-grid coordinates — and shows the full report before anything is accepted. After import, draw drifts on a level plane with waypoint snapping, drag/insert/delete waypoints, edit per-drift properties, generate spiral and straight declines, vertical raises, shafts, and ore passes between levels, and place stopes by dragged footprint — all with undo. Or start from the built-in immutable seven-level demo mine (editing it forks a library copy).

Design the cable — auto-route or draw your own runs and splitters

The radiating cable auto-routes through the primary access drift, the conveyor decline, and each level main on placement, respecting the level connections and the no-cable annotations on raises, ore passes, ventilation shafts, and production stopes. Or switch to manual (blank-canvas) mode and lay the cable yourself: Draw cable clicks waypoints along the runs (start on an existing run’s endpoint to extend it), Set start (headend) marks where the feed / base station connects, and Place splitter drops a power splitter — 2-way by default, set 2/3/4-way — that breaks the run into input and branch segments so the 10·log10(N) + insertion loss is actually applied downstream. Draw branch runs snapped to the splitter to add outputs; drawn cable snaps to splitters and amps; per-section delete, per-splitter removal, and clear-all keep it tidy. A Cable network inspector lists every routed segment, splitter, and unconnected drift, and the drawn cable drives the coverage directly.

Channel plan, AGC pilot, and the voice carrier

Coverage runs off a duplex channel plan, not a single frequency. The plan carries a downlink-only +0 dBm AGC pilot plus your traffic channels — each a downlink and an uplink carrier — and coverage is evaluated on the worst (highest-frequency) channel per direction, shown in the Eval @ chip, so a VHF voice channel and a UHF data channel show their different coverage. The AGC holds the pilot flat and restores it at every amplifier, while the traffic carriers ride a back-off below it (default −7 dB, editable per channel to protect amplifier headroom). The heatmap, the per-tier statistics, and the off-air hover levels are all the voice carrier — the level a worker actually talks on — while the pilot is reported separately as the AGC reference. The same plan feeds a composite-power and intermodulation read-out for amplifier-overload checks.

Live ITU-R stub physics — P.526 UTD corner diffraction and P.1238 corridor decay

Non-cabled drifts (stope access stubs, dead ends, ore-zone crosscuts) get their coverage from corner diffraction computed live at the channel-plan frequency. Per junction, the engine extracts the actual corner geometry from the drift waypoint headings and evaluates the ITU-R P.526-16 §6 UTD wedge-diffraction kernel with hard-rock material defaults, power-summing both diffracting corners at T-junctions and four-way intersections. Propagation along the stub past the corner follows the ITU-R P.1238-13 §3.1 corridor log-distance model, switching from the line-of-sight to the non-line-of-sight coefficient set past an internal bend. Both models are under unit test against the recommendations.

Floor-level voice-coverage heatmap — cascade, coupling, and body loss

Coverage renders as a heatmap ribbon along every drift floor at the worst-case receiver position. On cabled drifts each station combines the along-cable cascade (per-100 m loss times the cable run distance, through every splitter and amplifier) with the cable’s measured radial coupling loss from the ceiling-mounted cable to the chest-height receiver, the operator body loss, and the traffic-carrier back-off — then classifies it at the chosen reliability percentile. On non-cabled drifts the ribbon is diffraction-fed from every junction shared with a cabled drift. It is a true per-drift picture that captures the level-by-level loss accumulation, the splitter dilution, and the stub shadowing a single-run 2D cascade cannot show.

Downlink, uplink, and worst-of-link coverage modes

Downlink mode renders the voice level a handheld hears at the worst-case floor position — for gap identification and amplifier-spacing decisions. Uplink mode renders the level reaching the first upstream amplifier from a portable transmitting at each point — the limiting factor for talk-back, with the full budget to the headend and the cascaded uplink noise figure carried in the hover. Worst-of-link renders the minimum of the two, the single honest "can a worker use the radio here" view that finds holes where a worker can hear the base station but cannot reach it back. Diffraction-fed stubs carry no cable to inject into, so uplink and worst-of-link cover cabled drifts only — the statistics card discloses it.

Level slicing, reliability percentile, and the three-level hover

The level slice selector in the status bar (Off, L1 through L7) restricts the heatmap to one level’s band and dims the geometry above and below, so each level reads without overdraw — the way you find the worst-case run per level. The reliability percentile in Design settings (50 % median, 90 %, 95 % design-of-record, 99 % critical) applies the lognormal shadow-fading margin, with the cable’s cl50→cl95 coupling-loss spread setting the σ so 95 % reproduces cl95 with no double count. Hover any heat point for the three levels — DL voice (off-air at chest and on the cable), UL voice (injected back onto the cable and the level at base), and the pilot (the AGC reference and its droop) — or hover an amplifier marker for its input, AGC gain, output, and saturation state. The coverage statistics overlay doubles as the legend with live per-tier percentages, and splitter labels toggle on for an installation-document review.

Cable library fitted to manufacturer datasheets

A datasheet-fitted radiating-cable library: RFS RADIAFLEX (RLK114 VHF and the RLKU broadband series) and ZCG / Benelec 1/2" corrugated 50 Ω cables, plus DD-LFC-350, RNG-350, and Rojone CC-3529 75 Ω VHF/UHF cables. Each carries an along-cable loss curve as a least-squares K₁·√f + K₂·f fit against the manufacturer datasheet tables plus the per-cable cl50 / cl95 radial coupling-loss curve at the reference distance — with provenance recorded and any estimated field flagged. The channel-plan frequency drives the cable loss, the coupling loss, the UTD corner diffraction, and the corridor decay together, and the radio system selector (analog FM, P25, DMR, TETRA, NXDN, LTE MC) sets the tier thresholds from the ITU-R M.1808 reference sensitivities.

Mine library, self-contained project files, and autosave

Mines live in a named browser-storage library alongside the built-in demo — select, rename, duplicate, delete, and export to a .mine.json file. Projects autosave to local storage, and a project exports to a self-contained .lf3d.json from the Projects picker — cable, channel plan, radio system, amplifiers, your drawn cable runs and splitters, and the full mine geometry embedded, so a shared file re-imports later without the original mine.

Honest Beta scope

The tool is in Beta and says so. The RF maths is under unit test against the ITU-R recommendations — including operator body loss and the traffic-carrier back-off — but the model has stated limits: the cable model is a tree, so loops in the drift network are cabled along the first-found path (disclosed per junction — mark a redundant leg no-cable to choose explicitly); DXF arcs, splines, and block inserts are skipped with a disclosed count; coupling loss above a cable’s characterised band clamps; and the engine does not yet model frequency-dependent cable bend loss, full vendor splitter/tap insertion loss, polarisation, or the leakage-vs-radiating-mode regime. For the 2D installation cascade with full vendor splitter and tap data and a bill of materials, use the 2D Leaky Feeder Designer.

Standards & methodology

  • ITU-R P.526-16 §6 — uniform theory of diffraction (UTD) wedge diffraction, used for the corner loss into non-cabled drifts
  • ITU-R P.1238-13 §3.1 — indoor / corridor log-distance propagation, used for the decay along non-cabled drifts (LoS/NLoS bend switching; shadow-fading margins per Table 2 surfaced via the reliability percentile)
  • ITU-R M.1808-1 — reference sensitivities for land mobile systems, anchoring the Excellent through No Signal tier thresholds
  • ETSI (TETRA), TIA (P25), and 3GPP (LTE) air-interface sensitivity figures (conservative end) for the radio-system selector
  • IEC 61196-4 — radiating-cable coupling-loss measurement method (the cl50 / cl95 figures and 2 m reference distance for the cable library)
  • Vendor radiating cable datasheets (RFS RADIAFLEX, ZCG, Benelec, DD, RNG, Rojone) for the along-cable loss and radial coupling-loss curves, with provenance recorded per cable

When to use this tool

  • Importing a real mine’s centreline strings from Surpac, Vulcan, Deswik, or AutoCAD DXF and predicting leaky feeder coverage on the actual geometry
  • Laying the radiating cable by hand — drawing your own runs, dropping power splitters, and setting the headend — instead of accepting the auto-route
  • Running a duplex channel plan to compare VHF voice and UHF data coverage on the same cable, evaluated on the worst carrier per direction
  • Sketching a proposed extension (new drift, decline, or stope) onto the imported mine and seeing the coverage impact before the cable order
  • Sizing the inline amplifier spacing along the cable network (primary access, level mains, crosscuts) to hold downlink voice coverage across the mine
  • Identifying uplink coverage holes where a worker can hear the base station but cannot reach the system back
  • Using worst-of-link mode to find every point where either direction of the radio link fails
  • Designing at a 95 % reliability percentile so the heatmap reflects the cl95 coupling-loss spread rather than the median
  • Inspecting level slices (L1 through L7) to confirm the production stopes and ore-zone crosscuts are within coverage
  • Checking the composite power and intermodulation off the channel plan for amplifier overload before committing the carrier line-up
  • Sense-checking a vendor- or consultant-submitted leaky feeder coverage prediction against an independent calculation
  • Comparing two cable products, or two frequencies on one cable, to see the along-cable cascade and coupling/diffraction trade-offs
  • Producing a 3D visualisation appendix for an engineering report, a customer presentation, or a mine safety-case where the emergency-network coverage is documented

Is this the right tool for you?

Reach for the 3D Leaky Feeder Coverage in any of the following situations.

  • You have a mine survey export (DXF strings, or KML from the GIS) and want leaky feeder coverage predicted on the real drift geometry rather than a stylised single tunnel.
  • You want to lay the cable the way it will actually be installed — your own runs, power splitters at the level junctions, and the headend at the surface portal — and see coverage update as you draw.
  • You are extending an existing underground network into a new level or stope and need to see where coverage falls off — and where the next inline amplifier has to go — before committing the cable order.
  • You are checking whether handheld talk-back works at the worst-case point on a long run, where a worker can hear the base station on downlink but the uplink to the first upstream amplifier is the limiting direction.
  • You need to know how far coverage diffracts into the non-cabled stope accesses and ore-zone crosscuts off a cabled level main, computed honestly rather than assumed.
  • You are reviewing a contractor’s leaky feeder coverage submission for an underground emergency network and want an independent 3D prediction on the same geometry, at a 95 % reliability percentile, for the safety case.

Frequently asked questions

What does the tool actually compute, and is the physics real?

On cabled drifts it computes the along-cable cascade (per-100 m loss times the cable run distance, through every splitter and inline amplifier with the AGC restore) plus the cable’s measured radial coupling loss to a chest-height receiver, the operator body loss, and the traffic-carrier back-off below the AGC pilot. On non-cabled drifts it computes corner diffraction live at the channel-plan frequency per the ITU-R P.526-16 §6 UTD wedge kernel (power-summing both corners at T- and four-way junctions) and the decay along the stub per the ITU-R P.1238-13 §3.1 corridor model. Tier thresholds (Excellent through No Signal) are anchored to the selected radio’s ITU-R M.1808 reference sensitivity, classified at a reliability percentile you choose (50–99 %). The along-cable cascade, AGC model, splitter accounting, uplink walk-back and talk-back budget, UTD diffraction, corridor decay, body loss, and tier classification are all exercised by a unit-test suite against closed-form expectations and the ITU-R recommendations.

Do I have to use the auto-routed cable, or can I lay my own?

Either. The radiating cable auto-routes through the primary access and level mains on placement, but you can switch to manual (blank-canvas) mode and lay the cable yourself: draw your own runs (extend an existing run by starting on its endpoint), drop 2/3/4-way power splitters that break a run into loss-accounted segments, set the headend where the feed connects, and delete runs or splitters as needed. The drawn cable snaps to splitters and amplifiers and drives the coverage directly, replacing the auto-router — so the prediction reflects the cable the way it will actually be installed.

What is the channel plan, and why is coverage the "voice carrier"?

Underground leaky feeder systems combine a downlink pilot tone with the traffic carriers. The channel plan holds a +0 dBm AGC pilot plus your duplex traffic channels (a downlink and an uplink carrier each). The AGC amplifiers hold the pilot flat and restore it at every stage, but the traffic carriers are backed off below the pilot (a typical −7 dB) to protect amplifier headroom against intermodulation. The heatmap and statistics show that traffic (voice) carrier — the level a worker actually talks on — not the pilot, and coverage is evaluated on the worst (highest-frequency) channel per direction so adding a UHF channel reveals its extra loss versus VHF. The hover reports the pilot separately as the AGC reference.

What mine formats can I import, and what if I have no mine file?

The import wizard reads 3D DXF centreline strings (LINE / LWPOLYLINE / POLYLINE-VERTEX, the Surpac / Vulcan / Deswik / AutoCAD export format, with unit detection and a per-layer mapping step), waypoint CSV (one row per waypoint, header auto-detected), KML and GeoJSON LineStrings (altitudes used when present), and the native noim3-mine JSON. Every import runs through a validation pipeline (junction welding, cable-direction re-orientation, level derivation, connectivity and loop checks, survey-grid recentring) and shows a full report before anything is accepted. If you have no file, draw the mine in the scene (drifts, declines, raises, shafts, ore passes, stopes — all with undo) or start from the built-in seven-level demo mine.

What is the difference between downlink, uplink, and worst-of-link?

Downlink is the voice level a handheld hears at the worst-case floor position — the base-to-mobile direction. Uplink is the level reaching the first upstream amplifier from a portable transmitting at each point — the mobile-to-base direction, usually the limiting one for talk-back, with the full budget to the headend and the cascaded noise figure in the hover. Worst-of-link is the minimum of the two at every point: the single view that answers "can a worker use the radio here", and the one that finds holes where a worker can hear the base station but cannot reach it back. Diffraction-fed non-cabled stubs have no cable to inject into, so uplink and worst-of-link cover cabled drifts only (the statistics card discloses it).

How is this different from the 2D Leaky Feeder Designer?

The 3D Coverage Predictor answers "where is there coverage across the whole mine" — it renders a floor-level heatmap over real 3D geometry, including diffraction into non-cabled drifts, and is built around importing or drawing a mine and laying the cable on it. The 2D Leaky Feeder Designer is the bench design workflow: a schematic cascade with full vendor splitter and tap insertion-loss data, multi-carrier IM3 with duplexer passband filtering, a DC power feed solver, bidirectional uplink/downlink link budgets, and a bill of materials. Use the 3D tool for the spatial coverage picture and the 2D tool for the detailed installation cascade and documentation.

Is the tool production-ready?

It is in Beta. The RF maths is under unit test against the ITU-R recommendations, but there are stated limits: the cable model is a tree, so loops in the drift network are cabled along the first-found path (disclosed per junction — mark a redundant leg no-cable to choose explicitly); DXF arcs, splines, and block inserts are skipped with a disclosed count (re-export those drives as strings); coupling loss above a cable’s characterised band clamps; and the engine does not yet model frequency-dependent cable bend loss, full vendor splitter/tap insertion loss, polarisation, or the leakage-vs-radiating-mode regime. Coverage is classified at a reliability percentile you choose (50–99 %). The tool informs the design; certification and regulatory sign-off remain a Principal RF Engineer responsibility.

What radiating cables and radio systems are included?

Radiating cables: RFS RADIAFLEX (RLK114 VHF and the RLKU broadband series) and ZCG / Benelec 1/2" corrugated 50 Ω cables, plus DD-LFC-350, RNG-350, and Rojone CC-3529 75 Ω VHF/UHF cables — each carrying an along-cable loss curve least-squares fitted to the manufacturer datasheet and the cl50 / cl95 radial coupling-loss curve at the reference distance, with provenance recorded and any estimated field flagged. Radio systems: analog FM, P25 Phase 1 and Phase 2, DMR, TETRA, NXDN, and LTE MC, with reference sensitivities per ITU-R M.1808 and the relevant ETSI / TIA / 3GPP air-interface standards (conservative end) setting the coverage tier thresholds.