Lankford Under the Rotor: Helicopter EM for Gold, from the Yilgarn to Mt Morgan

Engineering Essay · Helicopter and Drone EM · Gold Exploration · Rev. 4 Lankford Under the Rotor: Helicopter EM for Gold, from the Yilgarn to Mt Morgan What a mathematician’s receiving-antenna method teaches about towed EM receivers, drone-borne B-field surveys over drill targets and prospectivity maps a geologist can audit — with the numbers worked through.

The short version

Helicopter EM does not find gold. It finds the conductive plumbing around gold: sulphide bodies, altered shears, buried river channels and the clays that hide them. What limits it is signal-to-noise. Transmitter moment sets the signal, and moment is capped by what the helicopter can lift and power. Cover, ground clearance, line spacing and turn-off time all still matter, but once the moment is fixed, most of the remaining gains come from the receiver: its noise, its references and its processing.

That makes the rest of the job a receiver problem. Dallas Lankford spent three decades solving receiver problems with a mathematician’s rigour, publishing every measurement and every revision. In my reading, his habits transfer to a towed EM bird almost one for one.

This post works through what that looks like in two very different Australian settings. The first is the deeply weathered, saline Yilgarn of Western Australia, including its granite-related gold skarns, a deposit class whose pyrrhotite and magnetite EM and magnetics can actually see. The second is the sulphide-rich belt of Central Queensland around the historic Mt Morgan mine.

It builds a noise budget with real numbers, treats the transmitter’s primary field as a phasing problem, and designs the receiver coil and preamp from first principles. It closes with a staged test programme an explorer could fund, with gates that a board can actually check.

The newest part concerns ground that is already pegged for drilling. Drones now carry B-field receivers over grounded transmitters, and machine-learning prospectivity maps can now report which evidence layers drove each ranking. This post proposes using the two together to confirm, move or drop a collar before the rig arrives. None of the published work tests that step directly, and conventional ground and downhole EM remain the benchmarks it has to beat.

Revision 4, 4 November 2026. In Lankford’s habit of dating every revision: Revision 2 added drone-borne B-field EM, UAV magnetics and AI prospectivity mapping. Revision 3 added the Yilgarn gold skarns and followed a quality-assurance audit. Revision 4 follows an independent source review by four separate reviewers. It corrects the superconducting-sensor, drone-transmitter and JORC statements, adds Mt Morgan’s two-style genesis and Lankford’s own bench numbers, and draws on about seventy further sources.

Why Lankford

Dallas S. Lankford IV (1942–2020) was born in Milton, Florida, and taught mathematics at Louisiana Tech University in Ruston from 1978 until he retired in 2003. He took his PhD at the University of Texas at Austin in 1972 under W. W. Bledsoe^[1], and computer scientists still credit him with first introducing polynomial interpretations^[2] for proving that term-rewriting systems terminate. He died at his home in Ruston on 29 June 2020^[3]. He was one of the most rigorous receiving-antenna experimenters the DX hobby produced. His articles read like lab notebooks: claims stated, measured, revised and re-dated.

He never flew a survey. What he left is a method. The table below is my reading of how each of his habits carries over to a helicopter; he did not write it down for EM. Where his own practice fell short of the habit, the post says so.

The rest of this post applies those nine habits.

What the loop actually sees

Airborne EM is very good at finding conductive ore in resistive rock and much weaker beneath conductive regolith, as Worrall, Munday and Green^[15] put it in 1998. Australia hands you both cases. For EM, the Yilgarn is the hard one and the Mt Morgan belt the easier one.

The lesson is blunt. At Mt Morgan-style targets the conductor can be the mineralisation, so the receiver’s job is depth and resolution. At Mt Morgan itself, fluid and isotope work by Ulrich and colleagues^[22] separates two styles of mineralisation in one deposit: a volcanic-hosted massive-sulphide style, and a later style tied to magmatic fluids from the Mt Morgan Tonalite. In my reading, that means the most conductive part of the system need not be the richest. But barren pyrite, pyrrhotite and graphite look the same to EM, so the drill still has to tell ore from pyrite: Salty Dog hole CMD003, below, cut up to 50% sulphide carrying almost no gold.

In the Yilgarn the conductor is usually the cover, so the receiver’s job is to see past it. The gold skarns described next are the main exception. That takes low frequencies, low noise and an accurate model of the cover, because the cover’s own response can swamp a basement target however quiet the receiver.

The Angel Eyes prospect, about 11 km north-east of the Tropicana mine, adds a third case, and it is the hardest. The gold sits with disseminated sulphide that is chargeable but showed no measurable conductivity in core. The salt-lake cover above it left the regional Spectrem survey unable to map bedrock. At about 10 S/m and typically 20–40 m thick, that cover is roughly 200–400 S, four to eight times the regolith conductance at Grant’s Patch.

Gold skarns: the Yilgarn gold that conducts

Not all Yilgarn gold leaves EM nothing to see. Andreas Mueller of the Karlsruhe Institute of Technology has reviewed a class of high-temperature gold skarns tied to post-orogenic granites and pegmatites^[20] in the east-central Yilgarn. They have produced about 580 t of gold: more than 360 t from the Southern Cross greenstone belt since 1887, and 219 t from the south-west margin of the Eastern Goldfields between 1894 and 2013. Their ore minerals are the ones EM and magnetics respond to, and their geometry is the kind a drill programme has to plan around.

The supplementary data^[20] list the setting and mineralogy of every deposit. The figures below are Mueller’s compilation, using the mill-head grades he gives or estimates:

That Chalice is the old gold mine near Norseman, unrelated to Chalice Mining’s Gonneville.

What the skarns mean for EM

• A conductor that may be ore, or may not. Pyrrhotite dominates the sulphides in most of the deposits, from about 1% at Copperhead to massive at Great Victoria. Disseminated pyrrhotite is a chargeability target, like the disseminated sulphide at Angel Eyes; connected breccia and massive pyrrhotite can be an inductive conductor. Barren look-alikes sit close by. A Cornishman zone with 5–10% pyrrhotite carried only 3–55 ppb gold, Wattle Dam is centred on graphitic chert, and lenses in the quartz-sericite-pyrite halo at Spargoville carry up to 40% pyrite. Ground SAM has already separated the two in the Southern Cross belt: at Cornishman’s Triple O lode it resolved three conductors^[23], two gold-bearing and one a barren weathered shear, and at Bronco it tracked a massive pyrrhotite-pyrite horizon and its offsets.
• Magnetics maps the system, not the gold. Outer alteration turned olivine into serpentine and magnetite, giving Nevoria a magnetic signature. At Ghost Crab, three linear gold-in-soil anomalies and a circular magnetic anomaly over the southern trend made up the surface footprint^[20] of the deposit, as Miller, McLeod and Devlin’s discovery history records. By contrast, the fertile pegmatite-granite under Nevoria is magnetically quiet, because alteration destroyed most of its magnetite. At Mount Morgans in Western Australia, not to be confused with Mt Morgan in Queensland, gold sulphide replaced magnetite in banded iron formation^[24] and showed as a magnetic low, while the coincident AEM conductor was put down to weathering. Not every magnetic anomaly is gold either: near Hyden a 12 km magnetic anomaly marks magnetite-pyrite in orthogneiss, and government-sponsored drilling at nearby Mt Vernon found 13–14% iron over more than 100 m.
• The ore stops where the granite starts. Nevoria, Marvel Loch, Edna May and Yilgarn Star end 250–800 m down plunge against thick granite-pegmatite, and thick intrusions at up to 800 m depth closed three Southern Cross mines. A towed EM loop sees, at best, the top of a plunging pyrrhotite body. Following it down plunge is a job for downhole methods, with a plate model that stops where the pegmatite begins. At Marvel Loch, where the ore had no large, consistently conductive zones, borehole MMR was chosen over downhole EM^[25], with the overburden response stripped out by layered-earth models.
• Late faults move the ore. Post-skarn faults offset ore bodies and intrusions: a dextral set striking NNE to NE, and a younger sinistral set striking E to ENE that shifts the Nevoria iron formation by up to 150 m. A drone magnetic survey that tracks a magnetite-bearing iron-formation bed across those faults is a direct way to move a collar onto the displaced block.
• Two commodities in one core tray. The Earl Grey spodumene pegmatite (189 Mt at 1.5% Li₂O) dips gently north beneath gold open pits 1 km from Bounty, and its discovery triggered renewed exploration. The Mt Marion pegmatites (78 Mt at 1.37% Li₂O) lie 3 km from Ghost Crab. On Yilgarn gold ground, every pegmatite intercept deserves a lithium, caesium and tantalum assay.

Figure 1. Schematic of a granite-pegmatite-related gold skarn, after the relationships described by Mueller (2025) for the Southern Cross belt; not a section through any one deposit.

The genesis of these deposits is contested, much as Mt Morgan’s is. Mueller argues that they are a distinct intrusion-related class; others treat them as deep, “hypozonal” orogenic gold. The difference is not academic for a prospectivity map, as the section on auditing those maps shows.

Model before you fly

Western Australia already has airborne EM across the whole state at about 20 km line spacing, downloadable through the MAGIX data system. The Geological Survey of WA^[26] notes that AEM can detect conductivity variations to depths of several hundred metres, and lists separate releases for the Eastern Goldfields, East Yilgarn–Albany Fraser and Murchison. Geoscience Australia publishes the AusAEM data under CC BY 4.0^[27]; its Eastern Goldfields and East Yilgarn blocks^[28], flown with TEMPEST in 2020, cover Kalgoorlie and Southern Cross. GSWA’s petrophysics programme^[29] measures susceptibility, density, resistivity, inductive conductivity and IP on public drill core and releases the results through MAGIX, which gives host-rock properties before a hole is drilled. Company exploration reports become public after a confidentiality period, and the state’s WAMEX database^[30] lets anyone download the open-file ones free. They are the first place to look for old EM surveys and drill logs over a block. Mueller notes that mine-by-mine production statistics have not been released comprehensively since the early 1990s, so his figures are compiled from Survey bulletins, AusIMM monographs and company reports. The WA datasets stop at the border. For Mt Morgan the starting point is the Geological Survey of Queensland’s Open Data Portal^[31], with open-file company reports and geophysics, and the AusAEM Northeast Queensland^[32] survey flown in 2024 on nominal 20 km lines; check whether any of its lines cross a given tenement.

A 20 km line will never see a deposit-sized pipe. What it sees is the host: how conductive the cover is, and how deep the resistive basement lies. That is the kind of model Lankford’s EZNEC runs gave him before he strung a wire.

The pre-flight workflow follows directly:

1. Pull the nearest regional conductivity sections and open-file company surveys over the ground.
2. Build a layered host from them: cover conductance, saprolite, depth to fresh rock. Yilgarn regolith can be as much as 150 m deep^[33], but electrically it is simpler: at Lawlers, six or seven geological regolith layers resolved as only three^[34], resistive cover, conductive saprolite and clay, and resistive basement.
3. Drop a target plate into that host at a realistic depth, dip and conductance, and compute its response gate by gate.
4. Lay the modelled response over the measured noise budget (next section). The gates where the target clears the noise set the base frequency, terrain clearance and line spacing.

A Western Australian discovery shows how regional analysis and ground EM divide the work. Chalice Mining found the Gonneville palladium-platinum-nickel deposit in 2020 when its first hole, JRC001, tested a moving-loop EM conductor on the south-west margin of a magnetic anomaly^[35]. Zuo and colleagues later cited Gonneville, at 560 Mt, as a prospectivity-mapping success^[36], crediting a target from Geoscience Australia’s continental-scale analysis. Chalice’s discovery release does not mention that analysis. On either account, it was a ground EM conductor that picked the hole.

One caution belongs on the wall. Banaszczyk, Annetts and Dentith showed in synthetic models that layered-earth (1D) inversions resolve conductors dipping more steeply than about 25° poorly, and can create false dipping features[37]. A vertical sulphide pipe is an extreme case, so I would treat plate or 3D modelling as essential at Mt Morgan-style prospects.

The noise budget, with numbers

In a towed bird the receiver’s own electronics are rarely the limit. In the worked example below, even a wobble of one-thousandth of a degree sits about 39 dB above the coil’s thermal noise, comparing rms with rms.

Australian work in 1998 from the CRC for Australian Mineral Exploration Technologies (CRC AMET) set out the hierarchy (Buselli, Pik and Hwang^[38]; Lane and colleagues’ streamed QUESTEM data^[39]):

Worked example: how much wobble is too much?

A coil whose axis tilts by a small angle θ in the Earth’s field sees a changing flux. For a sinusoidal wobble of amplitude θ₀ at angular frequency ω:

Vm(t) = NA · d/dt ( BE · n̂(t) )  ≈  NA · B⊥ · θ0 · ω · cos ωt

The illustrative receiver is a 1 m diameter, 100-turn coil, so NA = 78.5 m². It is wound with 314 m of 0.5 mm² copper, giving 10.8 Ω and a thermal noise of 0.42 nV/√Hz, or 42 nV rms in a 10 kHz bandwidth.

For a vertical-axis coil, the field across the axis is the horizontal intensity, taken here as 25 µT, typical of the Yilgarn. It is nearer 30 µT at Mt Morgan, which raises every motion voltage below by about 1.6 dB, and a horizontal-axis coil sees the much larger vertical component, roughly doubling its motion noise. The total field over Australia ranges from roughly 45 µT in the far north to about 62 µT in Tasmania.

Motion at 1–5 Hz sits below a typical transmitter base frequency and is mostly filtered out. Its harmonics, and any rocking near the base frequency, land squarely in band.

This is Lankford’s MiniFlag result in a new setting. Once external noise dominates, a quieter preamp buys nothing. In my view the money goes into mechanics, references and processing, in that order.

One refinement from the drone work at Tropicana belongs here. Power networks do not deliver a single clean 50 Hz line but an envelope of frequencies around it, so the team built a notch filter whose rejection band widens to match^[21]. A suitably chosen base frequency should stack 50 Hz out anyway; the filter was added to make the system robust where ambient EM noise is high.

The primary field is the pest station

Bucking the transmitter’s primary field is phasing under another name. At a single frequency its null depth is set by the same two errors Lankford fought on medium wave: amplitude match and phase match.

If the bucking signal matches the primary to an amplitude ratio a with a phase error φ, the residual is:

Null depth (dB) = −20 log10 | 1 − a ejφ |

Amplitude match	Phase match	Null depth
1%	0.5°	37.6 dB
0.1%	0.05°	57.5 dB
0.01%	0.005°	77.5 dB

Each tenfold tightening buys 20 dB. That is why Lankford insisted on identical antennas for his phasers: a mismatch, not a lack of gain, is what fills in the null. He expected his passive phasers to reach 70 dB on groundwave and 50 dB on skywave. By the formula above, 70 dB needs about 0.02% amplitude and 0.013° phase matching, and the 20 dB lost on skywave is the drifting path: the same contrast as a bench null against one in flight.

A time-domain system is harder in two ways. The match must hold across the whole receiver bandwidth, not at one frequency, and it must survive every flex of the frame. Because the primary falls with the cube of distance, a 0.01% amplitude match needs the geometry stable to about 0.003%. In a system that measures in the off-time, bucking matters mostly to keep the receiver out of saturation during on-time. Helicopter systems show the fragility directly: Schamper, Auken and Sørensen^[40] found that frame deformations of a few centimetres bring back a cancelled coil response, so they measure it at altitude and subtract it.

In the air the null will not hold still. In the fixed-wing QUESTEM 450 system, Lane and colleagues found that the measured “primary” also contained currents induced in the aircraft’s metal frame, and that its amplitude and shape changed with transmitter-to-receiver geometry^[39]. Helicopter systems tow the loop tens of metres below the airframe, which is steadier, but flexing and temperature still move the null. It behaves like a phaser null drifting with skywave: the geometry has to be measured and the primary modelled continuously, not set once on the ground.

Protect the front end

Lankford’s R-390A filter mods were about strong-signal handling as much as selectivity. The bird’s preamp faces the same test twice per transmitter cycle:

• Overload recovery. Residual primary during on-time must not saturate the preamp, or it must recover before the first off-time gate. Recovery time, measured with the system’s own waveform, is the specification that matters.
• Linearity before subtraction. Strong carriers such as NWC and mains harmonics intermodulate inside a non-linear preamp. The products are new signals that no remote reference contains, so no subtraction can remove them afterwards. Every cancellation scheme in the next section assumes a linear front end. Lankford found that a sampling receiver’s intercepts “do not obey the usual slope 2 … and slope 3 laws”^[41], so test linearity at the real residual-primary and sferic levels rather than extrapolating an intercept from a datasheet.

A second antenna cancels what the first cannot

In Australian trials, subtracting one ground station’s record from another’s cut sferic noise by a factor of about 4. Read as an amplitude ratio, that is 12 dB. Against random noise it is worth the same as stacking sixteen times longer, which a helicopter can only do by flying sixteen times slower.

Buselli, Pik and Hwang^[38] recorded sferics at paired stations across Australia. The three field components of the noise were strongly correlated, and so was the noise at stations up to 50 km apart. High-frequency sferics, roughly 2–100 kHz, tracked the remote record, and simple time-series subtraction gave the factor-of-4 reduction. The abstract does not say whether the factor is in amplitude or power; if it is power, the gain is 6 dB, still worth a fourfold stack. VLF carriers and 50 Hz noise also yielded to referencing.

The idea is borrowed from magnetotellurics, where Gamble, Goubau and Clarke^[42] introduced the remote magnetic reference in 1979.

Why 12 dB matters so much

Stacked random noise falls as the square root of the number of stacks:

σstack = σ / √N   ⇒   σ/4 needs 16 × N

A reference station delivers the same noise floor at survey speed, which can be spent on depth or on tighter spatial resolution. The square-root rule holds only for random noise. Impulsive sferics need robust stacking that rejects outliers, and coherent VLF and 50 Hz signals do not average down as √N at all, which is why they are removed by referencing or filtering instead. Macnae, Lamontagne and West^[43] showed pruned and randomised stacking improving signal-to-noise by up to about 6:1 without more transmitter power. Lankford’s R-390A noise blanker cut pulse noise 15 to 20 dB or more^[44], but only while static was not too dense, which is the same limit on any despiking scheme.

The subtraction is a phaser in software. In signal-processing terms it is the adaptive noise cancelling^[45] that Widrow and colleagues described in 1975. A transfer function h is estimated in noise-only windows, then the filtered reference is removed:

nres(t) = nbird(t) − ( h ∗ nref )(t)

The same idea, built into the bird

A balanced pair of coaxial receiver coils rejects any noise field that is uniform over the coil spacing. A US Department of Defense (ESTCP) study by Geophex^[46] of a cart-mounted coaxial sensor for unexploded ordnance, intended for towed arrays, lists sferics, geomagnetic storms, platform motion and power-line fields among what it rejects. The price is a taller assembly and harder bucking.

That study was for targets within about a metre of the surface. At exploration depths the same balance cancels most of the target signal too: with a 1 m coil spacing and a target 100 m away, the difference signal is about 30 dB below that of a single coil. The idea transfers to noise referencing, not to the main receiver.

Running it the Lankford way

• Site the reference where the noise is the same and the geology does not matter, within a few tens of kilometres.
• Use the same sensor type and the same filter chain at both ends, so the transfer functions start nearly identical.
• Time-stamp both against GPS so the records align to the sample.
• Measure the subtraction gain on noise-only data before trusting it on target data, and publish the number.

The timing point is not academic. In the first Tropicana drone trial the IP transmitter had no GPS synchronisation, and its zero-times drifted by milliseconds along a single flight line. The team picked the zero-times by hand on every line^[21] to rescue the resistivity data. A recent review of Chinese airborne magnetics makes the parallel point for magnetic base stations: as UAV surveys spread, ground reference stations, GNSS time synchronisation and online quality control^[47] matter more, not less.

The coil and its preamp

Design the receiver for linearity and overload recovery first, and for noise second. In the worked coil below, self-resonance lands near 29 kHz and needs about 2.7 kΩ of damping to settle cleanly after each switch-off.

A time-domain receiver coil is an induction sensor: its output is proportional to the rate of change of the field through it, V = −NA·dB/dt. Electrically it is an inductor with series resistance and a parallel capacitance that includes the cable.

L ≈ μ0N2r [ ln(8r/a) − 2 ]   ·   f0 = 1 / (2π√LC)   ·   Rcrit = ½√(L/C)

What the numbers say

• Damp it critically. An undamped coil rings at f₀ after every transmitter switch-off and smears the early gates. Critical damping settles fastest without overshoot.
• Count the damping resistor’s noise. A passive 2.7 kΩ resistor across this coil is the largest noise source above about 1 kHz. Referred to the coil’s EMF it adds about 0.26 µV rms over a 10 kHz band, six times the winding’s own noise. Feedback (active) damping in the preamp avoids most of it.
• Do not chase the last nanovolt. With active damping, a preamp with 1 nV/√Hz input noise is about 2.4 times the winding’s thermal noise, so it sets the intrinsic floor: about 0.11 µV rms in 10 kHz. That floor sits about 31 dB below the smallest wobble in the table and 43–57 dB below the others, so choose the input device for overload recovery and linearity. With passive damping, the smallest wobble’s margin falls to about 23 dB.
• Lankford already chose that way. His balanced cascode loop amplifier used FETs in the output stage for better strong-signal performance, as Mark Connelly’s (WA1ION) write-up of the design^[48] records. He also tuned his J310–J271 high-impedance preamp for second-order intercept^[49], reaching an IIP2 of about +100 dBm on medium wave. His trimmed push-pull Norton amplifiers reached IIP2 above +95 dBm^[50], and he warned that leaving out the trim pots costs “at least 15 dB”^[51] of second-order intercept: balance has to be adjusted, not assumed.
• Balance the input and split the shield. A balanced input rejects common-mode electric-field pickup. The electrostatic shield must be split so it does not become a shorted turn.
• Batteries in the bird, fibre to the aircraft. That removes the common-mode path that Lankford had to choke before his MiniFlag^[52] matched a 1 m active whip. His power-supply tests found that at NDB and VLF frequencies a larger common-mode choke did not reduce the noise further^[53]. AEM bands lie lower still, which in my view argues for isolation rather than better chokes.
• Calibrate the transfer function. Sweep the coil on a vector network analyser, then drive it from a calibrated field coil. That measured transfer function is the same one the remote-reference subtraction needs. Lankford’s warning applies: “the signal generator reading is not the number of microvolts”^[54] at the receiver input, so calibrate the field at the coil, not the drive current.

When the receiver is not a coil

An induction coil measures dB/dt, so its output falls as the field changes more slowly late in the decay. A B-field sensor measures the field itself. CSIRO’s LANDTEM^[55], a ground TEM receiver built on a high-temperature SQUID, uses that advantage to detect deeper bodies. In the air the advantage has been harder to bank: high-temperature SQUIDs flown in a towed bird proved comparable with, but not superior to^[56], induction-coil systems, with bird motion the main engineering obstacle. A coil’s output can also be integrated to recover the B-field response^[57].

Tropicana’s authors cite earlier SQUID work by Du and colleagues showing B-field sensors are more sensitive than dB/dt sensors to magnetic induced polarisation^[21]. Drill core from targets chosen on their earlier ground survey then confirmed that those chargeability anomalies were magnetic induced polarisation.

For the drone version the receiver is Geometrics’ MagArrow, a micro-fabricated atomic magnetometer of about 1 kg, sampled at 1 kHz. Here the sensor’s own noise did matter, which qualifies the rule above. Moving from the standard unit at 5 pT/√Hz to a lower-noise version at 1–2 pT/√Hz, which Geometrics offered after the trial, significantly improved the late-time, low-signal data. The price was a larger heading error, so slow base frequencies of 0.5 Hz or less need winds under 15 km/h.

Both observations fit Lankford’s rule once it is stated carefully. A quieter front end buys nothing until it is the limit, and in the late off-time of a B-field system it can be.

Figure 2. Receiver signal chain: where each kind of noise enters, and the stage that removes it.

The coil sees every field at once. Each stage downstream strips out one kind of noise, and only what survives the stacked gates reaches the interpreting geophysicist.

Mechanics before electronics

Motion noise is beaten by making the motion predictable, measuring it, and subtracting it. Filtering it blindly throws signal away with it.

1. Suspend for predictability, not just isolation. A CGG patent by Miles and Marr, now held by Xcalibur, deliberately turns erratic coil motion into what it calls “well-defined sinusoidal rocking”^[58], so the induced signal is predictable and separable from the earth response.
2. Measure the attitude. An inertial unit on the coil frame records the coil axis n̂(t). The bird’s magnetometer gives the Earth’s field. Together they predict the motion voltage. Rotation-rate sensors used this way cut rotation noise by one to two orders of magnitude^[59] at low frequencies.
3. Regress and subtract. In each processing window, fit the measured output to the predicted motion voltage and its harmonics, then remove the fit:

V̂m(t) = NA · d/dt ( BE · n̂(t) )   ·   Vclean(t) = V(t) − Σk βk V̂m(k)(t)

4. Do it after the flight, on raw data. Lane and colleagues found that computer-intensive post-flight processing of streamed data suppressed coil motion noise better^[39] than the real-time methods run on board.
5. Move the noise away from the sensor. On the fixed-wing CH-4 UAV, control-surface servos produced magnetic interference peaking at 0.8 nT that fell off quadratically with distance^[60]. Mounting the sensor 4.6 m beyond the wingtip held it below 0.1 nT. With full-axis dynamic compensation, the total-field standard deviation fell from 0.126 nT to 0.022 nT.

The practical consequence is contractual. An explorer should require the raw streamed data, attitude and reference records as survey deliverables, not just the stacked gates. Without them, nobody can reprocess the survey when better methods arrive, which is exactly the kind of revision Lankford published as a matter of course.

Listen instead of transmit

Receive-only EM is pure Lankford territory: no transmitter of your own, only better antennas, better references and better processing. Two methods matter, and ground resistivity decides between them.

VLF-EM on NWC

The naval station near Exmouth, callsign NWC, transmits on 19.8 kHz^[61]. Its wavelength is about 15 km, so any survey more than a few tens of kilometres away sees it as a plane wave. A 1978 Tasmanian survey used it because NWC was the only source strong enough^[62] for the job.

VLF couples best to conductors that strike toward the transmitter: roughly north-west in the Eastern Goldfields, and east–west at Mt Morgan, some 3,700 km away. The catch, noted in surveys back to 1969, is that VLF is sensitive to conductive overburden^[63].

Airborne AFMAG

AFMAG uses natural fields, so the sferics that plague active systems become the signal. MobileMT records the magnetic field on three orthogonal airborne coils, referenced to a ground base station of grounded electric lines, in up to 30 frequency windows from 19 Hz to 26 kHz^[64]. Its developers, Prikhodko and colleagues at Expert Geophysics, claim deeper investigation than controlled-source systems, especially in conductive settings. Lankford would note that the authors built the system, and ask for an independent side-by-side. Geotech’s ZTEM is the other commercial natural-field system.

Figure 3. Plane-wave skin depth, δ = 503√(ρ/f), from 10 Hz to 100 kHz for ground resistivities of 0.1 to 1000 Ω·m. Skin depth is the 1/e attenuation depth of a plane wave, not a detection depth.

Over saline Yilgarn cover, only the low-frequency end gets through; 50 m of 1 Ω·m clay is 50 S, close to the regolith conductance measured at Grant’s Patch^[18]. At Lake Rosetta’s 0.1 Ω·m, even 30 Hz penetrates only about 29 m, less than the cover is thick.

Over resistive host rock, even 19.8 kHz reaches 36 m at 100 Ω·m and 113 m at 1000 Ω·m. That makes VLF a cheap first pass over outcropping volcanics, and nearly useless over salt lakes. Skin depth is a plane-wave attenuation depth, not a detection depth, and it does not describe a helicopter’s own loop, but it shows which way the cover pushes the design.

Drones over the drill pattern

Once ground is pegged for core drilling, a drone-borne B-field survey over the same pattern is one way to test the planned collars before the rig arrives. That use is this post’s proposal. What the published work shows is the method itself: by 2025, AngloGold Ashanti and Vedanta Zinc International (VZI) had reported 55 MagnetoArrow survey blocks^[21] across several Tropicana prospects.

Put the transmitter on the ground

Researchers comparing drone and ground surveys at the Chulbatkan gold deposit conclude that drone magnetics can replace helicopter surveys, except where electrical methods need heavy equipment^[65]. A towed helicopter TEM loop is exactly that kind of equipment.

The answer at Tropicana builds on an established ground method and on earlier semi-airborne systems. Sub-Audio Magnetics (SAM)^[66], developed by Cattach, Lee, Stanley and Boyd in the early 1990s, already paired a grounded dipole with a B-field receiver. It draws on Edwards’ magnetometric resistivity^[67] and Seigel’s magnetic induced polarisation^[68], both published in 1974. Ground-source airborne systems such as GREATEM^[69] in Japan and DESMEX^[70] in Germany had already flown a receiver over a grounded transmitter, and the DESMEX group has since flown the receiver on a multicopter^[71], recording high-quality transfer functions up to 2 km from each transmitter.

Building on VZI’s earlier drone TFEM work with an inductive source at Aggeneys in South Africa (Smit and colleagues, 2022^[72]), AngloGold Ashanti and VZI kept the grounded dipole and flew the receiver on a drone. A Geometrics MagArrow is slung underneath, and VZI’s processing code, written by Jaco Smit, recovers four products from one 1 kHz time series^[21]:

• TMI, the total magnetic intensity, as in any magnetic survey.
• TFMMR, total-field magnetometric resistivity, read in the late on-time. It maps where the current channels through less resistive structure.
• TFMMIP, total-field magnetometric induced polarisation, read in the off-time. It responds to chargeable ground such as disseminated sulphide.
• TFEM, the total-field EM decay, also read in the off-time. The authors group TFMMIP and TFEM together as EMIP.

The trick that defeats the salt lake is physical, not computational. Over flat-lying ground, the magnetic field of current injected by a grounded source does not depend on the layer conductivities, so a flat conductive cover adds no anomaly of its own and only lateral changes show. In that low-frequency limit the anomalous current depends on conductivity contrast rather than absolute conductivity, so the method sees weakly conductive and resistive structure too, within the limits of a total-field sensor, which sees only the anomaly’s component along the Earth’s field. These points come from earlier work by Edwards and Nabighian and by McNeill and colleagues, as summarised by the Tropicana authors. Edwards’ 1974 paper states the first directly: a horizontally layered earth produces no MMR anomaly. Three-dimensional modelling of SAM data by Boggs, Stanley and Cattach^[73] likewise found that a conductive overburden layer has a negligible effect on the MMR response.

Two caveats are my own. The cover still carries part of the injected current, so less of it reaches the basement. And the cover’s own induction has to die away first: by my estimate, 400 S of cover takes of the order of 0.1 s to settle at the 300–500 m scale of a survey block, comparable to the 160 ms half-cycle at 3.125 Hz. That is consistent with the Tropicana team’s advice, below, that more conductive ground may need a lower base frequency.

The Tropicana paper’s survey parameters^[21] show how the drone version compares with the ground survey it grew from:

The trial was not lossless. The 20 m height and the halved transmitter current attenuated the signal, the 1 Hz limit weakened the EMIP stacking, and one of the two MagArrows was noisier than the other. The team still judged the drone system effective.

Tropicana’s is not the only drone B-field system. Gap Geophysics, the contractor behind SAM, reports a UAVSAM system^[74] transmitting at 1–12.5 Hz, with 1 pT/√Hz sensor noise and the sensor flown 10–20 m above ground at about 10 km/h. Those are the developers’ figures, which is exactly why a side-by-side over drilled ground is worth paying for.

Tropicana reports that its transmitter and drone cost per line-kilometre is now comparable to most AEM platforms. The paper gives no figures, and its data and code are commercial in confidence, so a board should ask for a quote rather than assume a price. Line spacing changes the sum: at 40–50 m lines a drone flies 20–25 line-km per square kilometre, two to five times what a helicopter EM survey flies at 100–200 m. The method suits a drill block, not a tenement.

What the drill bit found

The drill results are reported plainly, misses included, and that is what makes them useful. The Angel Eyes targets came from the ground SAM survey; the drone later reproduced its TFMMR result over part of that block. At Salty Dog, the drone anomalies coincided with earlier Spectrem, gradient-array IP and aircore gold anomalies, and all of them shaped the targets.

• Angel Eyes, structures. Ground magnetometric conductivity data outlined an apparent structural jog whose connecting structures had not been drilled. Drilling in 2020 hit several shears west of the known mineralisation, one hosting a new parallel lode: 17.85 m at 0.89 g/t gold from 283 m.
• Angel Eyes, chargeable zones. Drilling in 2021 into an EMIP anomaly along strike returned 19 m at 1.79 g/t gold from 148 m, within a 58 m zone of more than 0.5% sulphur. Drilling of a parallel EMIP anomaly in 2022–23 returned 14.5 m at 1.38 g/t, 12 m at 1.49 g/t and 10 m at 4.44 g/t. The anomalies sat slightly offset from the drill holes, interpreted as the down-dip part of the mineralisation.
• Salty Dog, the coincident anomaly. Hole CMD001 tested coincident magnetic, IP, TFMMR and Spectrem anomalies to 252.5 m. Peak gold was 0.069 ppm.
• Salty Dog, the conductor. Hole CMD003 tested a coincident AEM and EMIP anomaly. It cut 5–50% visible sulphide from 205 to 220 m, close to the modelled Spectrem depth, with sulphur to 9.66% but peak gold of only 0.024 ppm.
• Salty Dog, the margin. Hole CMD002 tested the demagnetised rim of the intrusion and returned 2 m at 1.55 g/t gold from 73 m.

The B-field anomalies that were drilled proved to be real structures, chargeable zones or conductors. At Salty Dog the drone’s EMIP reproduced a conductor the AEM had already found, and the drill confirmed sulphide at the modelled depth. None of that promised gold: the best Salty Dog intercept came from the demagnetised rim seen in the magnetics, not from the conductor. Lankford would have published every one of these results side by side, as the authors did.

Measure the core, then believe the map

Core measurements explain why chargeability, not conductivity, tracked the gold at Angel Eyes. A handheld KT-10 meter read inductive conductivity below 0.5 S/m over the mineralised zones, while a core IP tester found chargeability above 15 ms over wide intervals. Gold grade rose with sulphide content. The KT-10’s quoted conductivity range starts at 1 S/m^[75], so that reading means “not massive sulphide” rather than a measured value.

The authors blame the conductive cover for the regional AEM’s failure there. My own reading adds a second reason: an inductive system responds to conductance, so it would struggle with this chargeable, non-conductive ore even under thin cover. That reading has published support. Comparing airborne inductive and ground galvanic IP in Tasmania, Macnae and Hine^[76] found that an inductive airborne system misses many conventional IP targets unless the mineral grains are much finer than 1 mm. In my view, measuring the core is what turns a disappointing survey into a design rule for the next one.

Field rules from four years of flying

The Tropicana team’s system development notes^[21] read like a Lankford revision list:

• In more conductive ground, a lower base frequency and slower flight may be needed.
• Wind above 25 km/h induces too much yaw, roll and pitch, and degrades the data beyond use. Fly at night where the wind drops, which also brings fewer bird attacks; in Australia that needs a CASA night approval.
• Where the wind is below 25 km/h and roughly parallel to the lines, longer-endurance drones can fly every line into the wind in a z-pattern, with faster diagonals between lines. Data quality improves about twofold.
• Allow about 50 m for the sensor to settle after a turn. Extend lines past the block and fly spline turns at survey speed.
• GPS-synchronise the transmitter. The trial’s millisecond drift cost hand-editing on every line.
• Lay extra dipole loops before flying; the authors estimate about 20% more production.
• Process each sortie in the field, so re-flies are called before the camp moves.

How low is low enough

Over designated drill areas the drone has to fly low, because the targets are deposit-sized. At Chulbatkan^[65], contractor GEPART flew a magnetometer on a DJI Matrice 600 Pro at 40–63 m, averaging 53 m, on 50 m lines over 122 km². Repeat lines agreed to ±0.93 nT.

For mapping, the drone survey agreed well with 1:10,000 ground magnetics. Over the 5.4 km² deposit itself, even doubling the drone line density did not recover the ground detail. In one overlap area, most of what was lost at height was narrow dyke anomalies of around 5 nT, some of which vanished entirely.

The authors’ recommendations are specific:

• Drones suit 1:10,000 to 1:25,000 work; at 1:5,000 and larger, ground surveys win.
• For deposit-sized objects of up to a few square kilometres, ground magnetics is still advisable.
• Below an effective magnetic susceptibility of about 50 × 10⁻⁵ SI, the risk of missing an anomaly is high.
• A magnetometer at 5–8 m is entirely acceptable and cost-effective over ground free of forest cover and complex terrain. In their forested, mountainous terrain the drone could rarely fly below 40–60 m.
• Weakly magnetic sedimentary sections are unfavourable, which calls for caution in orogenic gold settings.

Tropicana’s 15–20 m drape, limited by the tallest trees, sits between these regimes. Over a drill pattern, terrain clearance is a target-detection parameter, not a flying convenience. The published work does not yet say how deep, or how weak, a target the drone B-field method can detect; the deepest drilled confirmation is the Salty Dog sulphide at 205–220 m.

Bigger drones, cleaner platforms

Larger unmanned aircraft now carry real payloads. The performance figures below come from different measurements on different targets, so they show progress, not a ranking:

The DY-115^[77] needs only a 20 × 20 m landing area and operated in level-5 winds. Its 2019 test survey over the Bayan Obo mining area averaged 137 m above ground, over anomalies exceeding 8,000 nT. The CH-4 tests were flown at 200 m for a 1:50,000 survey, and residual servo noise of 0.1–0.2 nT above 2 Hz was not yet removed.

A review of Chinese airborne magnetics notes that a high-temperature superconducting full-tensor gradient pod^[47] reached 30 pT/m as an engineering prototype between 2016 and 2020. In that review’s account, low-temperature SQUID systems are more sensitive but remain held back by liquid-helium cooling, weight and long-term operational support. It does not report either superconducting system flying on a drone. Superconducting sensors are not new to Australian exploration, though: LANDTEM has been used as a ground TEM receiver since about 2000, though airborne trials were less conclusive, as noted above.

The three Chinese papers share a corresponding author and institute, so they are one group’s progress rather than independent confirmations. None of these heavier aircraft carries an EM transmitter in these papers. The first drone-borne transmitters are appearing elsewhere: a separate Chinese Academy of Sciences group has described a 14.94 kg drone TEM system^[78] with a 2,250 A·m² transmitter and a stated detection depth of about 50 m. It is a preprint still under review, and its moment is about two orders of magnitude below a helicopter system’s. In my view their payload headroom is the reason to watch them, though in Australia any of them would need bespoke CASA approvals.

Figure 4. Proposed loop for a designated drill area, combining drone B-field EM after Terblanche et al. (2025) with prospectivity mapping.

The loop closes at the core tray. Every hole drilled through a drone anomaly measures the property the survey claimed to see, and that measurement retunes both the next survey and the prospectivity model. The loop is this post’s proposal, assembled from the published pieces.

Prospectivity maps a geologist can audit

Mineral prospectivity mapping (MPM) turns survey layers into a ranked map of where to drill. The newest methods report which evidence layers drove each ranking, and some attach an uncertainty to it. None of the studies below tests its maps at the scale of a single collar. Hronsky and Kreuzer warn that spatial prospectivity mapping is biased towards mature, well-known areas^[79] and produces excessively large high-prospectivity areas, and they argue for “intelligence amplification” rather than artificial intelligence. This post takes the same view.

Zuo and colleagues set out the workflow that the algorithms serve, drawing on the mineral-systems and targeting-model work of Wyborn, McCuaig and others. It runs in four steps^[36]:

1. Build a robust conceptual model of the commodity and deposit type.
2. Translate it into a practical exploration targeting model.
3. Assemble a comprehensive, high-quality geodatabase, including drilling data.
4. Choose the targeting parameters and integrate them with the model.

Where the drone data goes

Magnetic layers, including drone-derived ones, can enter these models as evidence layers, although none of these studies used drone data. In the Dharwar study, regional Geological Survey of India aeromagnetic channels dominated the attribution^[81]: total magnetic intensity, its first vertical derivative, tilt derivative and analytic signal.

At Angel Eyes the Tropicana geologists did a similar integration by hand. They plotted bottom-of-hole alteration geochemistry over the ground-SAM magnetometric conductivity map and found a strong correlation between alteration strength and the anomalous features^[21]. They also noted, as Lankford would, that the result is biased by drill-hole density. That is prospectivity mapping with two evidence layers and a pencil.

Checking the scores the Lankford way

• Pick the mineral system before the layers. The Yilgarn gold skarns sit on sheared rock contacts with less than 200 m of displacement, not on crustal-scale faults, and the crustal-scale Koolyanobbing shear zone near Southern Cross contains no gold deposits^[20]. A model that rewards distance to a major fault, the orogenic habit, would rank them poorly. One built on fertile granite-pegmatite, iron formation and sheared contacts would rank them well. That is Zuo’s first step in practice. A recent mineral-systems model of the central-eastern Yilgarn gave rock-contact density the most weight and proximity to principal faults the least^[83], whereas GSWA’s regional gold targeting atlas^[84] ranked fault density and fault bends among its best criteria. Different mineral systems, different winning layers.
• Demand spatial, leakage-free validation. The Dharwar study excluded each test region from pre-training and used a 20 km buffer. Tongling’s split of 114 samples into 80 for training and 34 for testing was random, which this test would not pass. A score from randomly shuffled samples flatters any model. In the Yilgarn itself, a weights-of-evidence gold model^[85] captured 49.3% of gold points in the top 5% of area under random cross-validation, but 45.4% with 100 km spatial blocks and 42.1% with 150 km blocks.
• Read the whole curve, not the headline. In Dharwar the top 5% of ranked area captured only 15.6% of occurrences. In Fujian, capturing every known deposit took a third of the area. Zuo and colleagues cite a benchmark that MPM should cut the search area by at least an order of magnitude.
• Never compare metrics across papers. AUC and AUC-PR measure different things, and each district has its own base rate of deposits.
• Know what was not validated. The Dharwar authors call their deeper prediction heads exploratory only, unvalidated against boreholes. Their training occurrences also mix gold, iron ore, chromite and polymetallic deposits.
• Check the inherited layers too. A standard input can itself be machine-learnt. The national depth-of-regolith model^[86] was fitted to 128,033 bore depths and has a cross-validation R² of only 0.38.
• Respect how rare the positives are. Only 0.42% of the Dharwar pixels were labelled as mineralised, using 500 m buffers around point records. Most of the map is unknown, not barren.

A Lankford-style shootout

This five-phase programme spends money only after the previous phase has produced a number. Every gate is a measurement a board can read, not an opinion.

Figure 5. Proposed test programme: five phases, each closed by a measured gate.

Phases 0 to 2 need no aircraft at all. Phase 3 is where the shootout habit pays: fly the same line repeatedly over ground that has already been drilled, beside a system whose response is known. If the repeat lines do not agree within the measured noise, nothing downstream can be trusted, however good the inversion looks.

This is stricter than Lankford’s own field practice. His Quoddy Head^[87] and Grayland^[88] comparisons were mostly made on different nights and judged by ear or by log counts; his hard numbers came from the bench. The nearest he came to this standard was at Kongsfjord, with two receivers recording at once.

The standard already exists in Australian practice. Geoscience Australia’s survey contracts require flights over calibration ranges^[89], and eight surveys by seven AEM systems over one Lake Menindee line gave “a remarkably consistent interpretation”^[90] of the geology. Davis and Macnae^[91] show how a ground loop of known response quantifies an AEM system. A drone can serve as the calibration source too: George Wallner (AA7JV) measured about 25 dB front-to-back on his build of Lankford’s phased delta flag array by flying a drone transmitter around it^[92].

Near Mt Morgan, the Mt Chalmers massive-sulphide deposit, about 50 km to the north-east on another company’s ground, is candidate calibration ground: QMines reports that a reprocessed 2007 VTEM survey detected it^[93], and attributes conductance lows over some mineralised soil anomalies to IP effects.

What this means for an explorer

An explorer holding a freshly flown helicopter TEM dataset, say a SkyTEM survey over rugged Central Queensland ground with tighter infill where the terrain is steep, gets the most from it by auditing the noise before trusting the targets. Reprocessing pays: in the eastern Yilgarn, a 2013 HeliTEM survey whose review found nothing worth drilling was re-inverted for IP effects^[94], and the masked conductors it exposed led to a hole that cut 8 m at 0.64% nickel from 102 m. Interpretation and target ranking stay with the company’s geophysicist; this work makes their inputs quieter and their decisions traceable.

The coil, preamp and bench test above are a way to audit what a contractor delivers, not a proposal to build a system. What a company actually buys is a commercial system: helicopter time-domain EM such as VTEM, HeliTEM, Xcite or SkyTEM, frequency-domain RESOLVE, or a natural-field system such as ZTEM or MobileMT.

The rest of the toolkit

On designated drill ground, an airborne or drone anomaly is a hypothesis that usually meets ground and downhole methods before it meets a collar:

• Ground EM. Fixed-loop and moving-loop TEM with fluxgate or SQUID receivers are the standard way to turn a conductor into a drillable plate. Gonneville’s discovery hole tested exactly that kind of target.
• Ground IP. Gradient-array or 3D IP maps disseminated sulphide like that at Angel Eyes, where inductive methods see little. Tropicana itself began with an unexplained gold-in-soil anomaly^[95], and gradient-array IP with aircore geochemistry was then “the most cost-effective”^[96] way to target early drilling. At Salty Dog, the IP anomaly that hole CMD001 tested came from gradient-array IP.
• Downhole EM. DHEM, and downhole MMR or IP for low-conductance targets as at Marvel Loch, sees conductors the surface missed and those a hole passed beside, and follows a plunging pyrrhotite skarn down to where the granite cuts it off. It needs holes left open and cased with PVC, which the board can require in the drilling contract now.

Why a collar decision is worth paying for

Discoveries are getting dearer. MinEx Consulting’s Richard Schodde puts world exploration spending per discovery at about US$65 million in 1975–2005 and US$218 million in 2011–20^[97], while Australian discoveries fell from 294 in 2002–11 to 153 in 2012–21 as spending rose. Of 38 Australian primary gold deposits over 100,000 oz found in 2005–14, only 8% outcropped^[98], and their average depth was 64 m.

Kreuzer and colleagues show how to price an exploration decision^[99] by chaining the probabilities of each part of a mineral system. Value-of-information methods such as Mern and Caers’^[100] reach a drill decision with fewer measurements, though their tests were synthetic and they wrongly abandoned about a third of profitable multi-body cases. The drone-and-model loop in this post is a value-of-information argument: spend a little to change a big decision, and keep the gates that stop a model from overruling the drill.

Before anyone flies

• Aviation safety. Low-level helicopter EM over rugged ground like Mt Morgan’s is, in my view, the largest risk to people in an EM programme. A drone takes the crew out of low-level flight, a point in its favour that cost comparisons miss.
• Drone regulation. Commercial drone work needs a remote pilot licence (RePL), an operator’s certificate (ReOC) and, outside the standard operating conditions, a CASA flight approval^[101]. The standard conditions are daytime, within visual line of sight and below 120 m. Lines of 1.5–2.5 km imply a beyond-visual-line-of-sight approval or extended line of sight with observers, and night flying needs its own approval.
• Grounded-dipole safety. Ten to twenty-three amps through kilometres of cable, at potentially hundreds of volts, calls for live-cable procedures, step-and-touch-potential controls at the electrodes, and exclusion zones for people and stock, above all at night. In WA the work falls under the Work Health and Safety (Mines) Regulations 2022; in Queensland, under the Mining and Quarrying Safety and Health Act 1999.
• Land access and heritage. Dipole electrodes and cable runs reach kilometres beyond the drill pattern, possibly outside heritage-cleared ground. Check them against the Aboriginal Heritage Act 1972 in WA, the duty of care under Queensland’s Aboriginal Cultural Heritage Act 2003, and landholder notice under Queensland’s Land Access Code.
• Disclosure. Under the JORC Code^[102], geophysical survey results are Exploration Results, and Table 1 asks for the parameters of each geophysical tool, including make, model and calibration. ASX counts geophysical surveys as exploration^[103] and requires the Competent Person’s prior written consent. It also discourages visual estimates^[104] of mineralisation, such as “5–50% visible sulphide”, and says they should never be a proxy for laboratory analyses. A revised JORC Code is being finalised; as of August 2026 it had not been released^[105], and an ASX consultation and transition period will follow. Its draft recasts Exploration Targets as a “discussion” of exploration potential^[106], so a drone anomaly called a “target” in a release needs care under either edition.
• Ground conditions. In the Yilgarn, maghemite-rich laterite can add a superparamagnetic (magnetic viscosity) decay^[107], and it added noise to SAM data too. Airborne transients can turn negative over chargeable ground; both effects mimic or mask late-time targets. At Mt Morgan, the town, rail and power lines add cultural conductors and 50 Hz noise, and wooded, rugged terrain limits how low a drone can drape. The state-managed mine site has a large water-filled open cut^[108], upgraded high-voltage infrastructure, a water-treatment plant, a tailings-reprocessing project expected to start in 2026, and about half the site heritage-listed. Marvel Loch’s open pit was modelled as a reverse-current conductor, and a flooded pit deserves the same treatment. Mt Morgan’s genesis has swung between a volcanogenic massive-sulphide model^[109] and an intrusion-related replacement model^[110]; Ulrich and colleagues’ two-style reading combines them, and which style carries the gold changes what to model.
• Drilling co-funding. Western Australia’s Co-funded Exploration Drilling Program^[111] refunds up to half of direct drilling costs, capped at $180,000 for a multi-hole project and $220,000 for a single deep hole. It covers exploration drilling only, not geophysics or resource definition, and it opens in rounds. Ramelius Resources’ deep diamond drilling at Wattle Dam North in 2012 was co-funded this way, and its report is on WAMEX^[112].
• Season and standby. The Central Queensland wet season, WA summer heat and fire bans, and the wind limits above all cost days. Budget standby time accordingly.

Work packages

Put these in the next survey contract

The national guidelines for submitting digital exploration data^[113] ask for raw AEM data only “if supplied by the survey contractor”, and want drone data submitted in the same form as data from conventional aircraft. That condition is why the raw data has to be in the contract:

• ☐  Raw streamed data for every channel, including the transmitter current monitor
• ☐  Coil attitude and GPS at full sample rate
• ☐  Reference or base-station records covering every flight
• ☐  Waveform, system response and calibration files
• ☐  Every processing parameter, recorded per version
• ☐  Processed data in ASEG-GDF2 or ASEG-ESF, with the processing sequence documented as the national guidelines require
• ☐  A flight over a calibration line or range of known response, repeated during the survey
• ☐  For drone B-field surveys: a GPS-synchronised transmitter, its current record, and the raw 1 kHz sensor time series
• ☐  Conductivity, chargeability and magnetic susceptibility measured on representative drill core, including the pyrrhotite-rich, graphitic and magnetite-bearing intervals
• ☐  Every pegmatite intercept logged and assayed for lithium, caesium and tantalum as well as gold
• ☐  In the drilling contract: holes left open and PVC-cased for downhole EM

None of this costs flying time. All of it decides whether the survey can be revisited in five years, when the methods have moved on and the ground has not.

Further reading for a Yilgarn desk study

These are the Western Australian sources Mueller’s review draws on, for anyone building a Phase 0 desk study over Southern Cross or the south-west Kalgoorlie Terrane. Most are Geological Survey of Western Australia publications without a stable web link; search the Survey’s eBookshop or the state’s GeoVIEW and DASC portals by title.

• Keats (1991), Geology and gold mines of the Bullfinch–Parker Range region, Southern Cross Province, Western Australia, GSWA Report 28
• Doublier (2013), Geological setting of mineral deposits in the Southern Cross district: a field guide, GSWA Record 2013/11
• Swager, Griffin, Witt, Wyche, Ahmat, Hunter and McGoldrick (1995), Geology of the Archaean Kalgoorlie Terrane: an explanatory note, GSWA Report 48
• Witt (1992), Heavy-mineral characteristics, structural settings, and parental granites of pegmatites in Archaean rocks of the eastern Yilgarn Craton, GSWA Record 1992/10
• Witt (ed.) (1997), Geology and mineralization of the south-central Yilgarn Craton: a field guide, GSWA Record 1997/8, including Mason on the Bounty mine
• Matheson and Miles (1947), The mining groups of the Yilgarn goldfield north of the Great Eastern Railway, GSWA Bulletin 101
• Berkman and Mackenzie (eds) (1998), Geology of Australian and Papua New Guinean mineral deposits, AusIMM Monograph 22
• Phillips (ed.) (2017), Australian ore deposits, AusIMM Monograph 32, including Smith and Ross on Mt Marion
• Miller, McLeod and Devlin (1999), “Discovery of the Ghost Crab deposit, Kalgoorlie, Western Australia”, in New Generation Gold Mines ’99: case histories of discovery
• Bucci, Hagemann, Groves and Standing (2002), “The Archean Chalice gold deposit”, Ore Geology Reviews 19, 23–67
• Mueller and McNaughton (2000), “U-Pb ages constraining batholith emplacement, contact metamorphism, and the formation of gold and W-Mo skarns in the Southern Cross area”, Economic Geology 95, 1231–1257

Limitations

I am not a geophysicist and not a Competent Person under the JORC Code. The noise budget, coil design and time-constant estimates are worked examples with stated assumptions, not measurements on any company’s system, and the drone and prospectivity material reports what the cited authors found rather than tests of my own. Anything here that would shape a drill programme should be reviewed by the company’s geophysicist first.

Credits

• The Lankford archive. Lankford’s articles survive because others kept them. Bjarne Mjelde hosts the Dallas Lankford collection of modifications and projects^[114] at kongsfjord.no, about 90 dated items in 12 sections, and the copies cited here come from that collection, from okdxf.eu, from DXing.info and from World Radio History, which hosts the IRCA reprints. The International Radio Club of America obtained permission from Lankford’s family^[115] to reprint his articles, as Nick Hall-Patch announced in 2022, and an index of IRCA reprints^[116] lists 31 of his articles. DXer.ca also archived his articles. His designs are still being built and tested: Steve Ratzlaff (AA7U) rated his MW Phaser #2 “significantly better”^[117] than other phasers, and George Wallner (AA7JV) tested his phased delta flag array by drone.
• Antennas and amplifiers. EZNEC is Roy Lewallen’s (W7EL) modelling program. The flag antenna is Earl Cunningham’s (K6SE); the delta flag and MiniFlag are Lankford’s developments of it. The BFE-C write-up of Lankford’s balanced cascode amplifier is Mark Connelly’s (WA1ION).
• Tropicana. The B-field work is by Terblanche and colleagues of AngloGold Ashanti, with processing code by Jaco Smit, formerly of Vedanta Zinc International. AngloGold Ashanti and VZI funded the work, and the Tropicana Joint Venture partners, AngloGold Ashanti Australia and Regis Resources, part-funded it. Geometrics supplied the MagArrow and the lower-noise version. The authors acknowledge contractors including Gap Geophysics Australia, GeoRESULTS, Moombarriga, RemSense, Vortex and AirGeoX. Their paper is open access under CC BY 4.0; this post paraphrases its findings, quotes only short phrases and reproduces none of its figures.
• Yilgarn gold skarns. The deposit data, production figures and geology are Andreas Mueller’s (Karlsruhe Institute of Technology), from his 2025 review and its Electronic Supplementary Material, open access under CC BY 4.0. Much of it rests on his University of Western Australia PhD and on consulting studies of Copperhead, Ghost Crab and Chalice in 1995–1997. The Ghost Crab discovery history is Miller, McLeod and Devlin’s (1999); the Wattle Dam alteration study is Bath and colleagues’^[118] (2020). This post paraphrases and tabulates the review’s facts and reproduces none of its figures.
• Other field work and data. GEPART acquired both Chulbatkan surveys. The DY-115 was built by Jiangsu East Wing General Aviation, and the CH-4 compensation algorithm is the China Aero Geophysical Survey and Remote Sensing Center’s. The Dharwar aeromagnetics and occurrence records are the Geological Survey of India’s. Regional AEM is from the Geological Survey of Western Australia and Geoscience Australia.
• Borrowed ideas. Remote referencing comes from magnetotellurics (Gamble, Goubau and Clarke) and software subtraction from adaptive noise cancelling (Widrow and colleagues). The grounded-source B-field method rests on sub-audio magnetics (Cattach and colleagues), magnetometric resistivity (Edwards and Nabighian) and magnetic induced polarisation (Seigel).

Sources

1. Mathematics Genealogy Project, Dallas Sylvester Lankford IV: PhD, University of Texas at Austin, 1972, “Equality Atom-Term Locking”, advisor Woodrow Wilson Bledsoe
2. Nguyen, De Schreye, Giesl and Schneider-Kamp (2009), Polytool: polynomial interpretations as a basis for termination analysis of logic programs, arXiv 0912.4360
3. Ruston Daily Leader (2020), Dallas S. Lankford IV, obituary
4. Lankford, Measurements of some antennas’ signal to man-made noise ratios in the daytime MW and LW bands, II, 20 September 2006, rev. 18 February 2007; also IRCA reprint A-211
5. Lankford, Tests of new active MW antennas, DXing.info, rev. 2012
6. Lankford, Some recent results for MW antenna arrays, 4 November 2009, rev. 3 May 2010
7. Lankford, Phased delta flag arrays, 21 February 2009, rev. 23 April 2010
8. Lankford, New improved passive phasers, 2 August 2007, rev. 12 July 2008
9. Lankford, New improved passive phasers II, 2 August 2007, rev. 23 May 2008 (an earlier revision than the okdxf.eu copy)
10. Lankford, R-390A filter mods 2 and 3, December 1994 and 1 January 1995; IRCA reprint M-076, World Radio History
11. Lankford, SDR-IQ overload characteristics, 14 January 2008
12. Lankford, Half size dual active delta flag arrays, DXing.info, 31 December 2011, rev. 7 January 2012
13. Lankford, Supercharged active MiniFlag and MiniLoop antennas, rev. 8 August 2011
14. Lankford, On comparing noise output powers of amplifiers with simple equipment, draft, 11 January 2011, rev. 17 January 2011
15. Worrall, Munday and Green (1998), Beyond bump finding? Airborne electromagnetics for mineral exploration in regolith dominated terrains, Exploration Geophysics 29, 199–203
16. Huston, Dean, Messenger, Jones, Taube and White (1993), Mount Morgan gold-copper deposit: the 1992 perspective, AusIMM
17. Enterprise Metals (2012), Ponton Project: AEM survey identifies deep palaeochannel uranium targets, ASX announcement, 3 September 2012
18. Worrall, Whitaker, Lane and Meyers (2001), Exploring through cover: the integrated interpretation of high resolution aeromagnetic, airborne electromagnetic and ground gravity data from the Grant’s Patch area, Eastern Goldfields Province, Archaean Yilgarn Craton. Part A: Mapping geology using airborne data, ASEG Extended Abstracts 2001(1), 1–4
19. Stolz (2003), Direct detection of gold bearing structures at St Ives, WA: DHEM vs DHMMR, ASEG Extended Abstracts 2003(2), 1–5
20. Mueller (2025), Granite-pegmatite-related gold skarns and associated Li-Cs-Ta pegmatites in the Archean Yilgarn Craton, Western Australia, Mineralium Deposita 60, 835–867, with Electronic Supplementary Material; open access under CC BY 4.0
21. Terblanche, Brown, Darragh, Berrell, Watkins, Andrae, Mason-Apps and Smit (2025), Application of ground-based B-field (TMI, TFMMR, TFMMIP and TFEM) surveys at Tropicana Gold Mine and pioneering of a comparable drone-based acquisition system, Mineral Economics 38, 717–739, open access under CC BY 4.0
22. Ulrich, Golding, Kamber, Zaw and Taube (2002), Different mineralization styles in a volcanic-hosted ore deposit: the fluid and isotopic signatures of the Mt Morgan Au-Cu deposit, Australia, Ore Geology Reviews 22(1), 61–90
23. Jackson (2005), The use of sub-audio magnetics (SAM) in gold exploration: examples from the Yilgarn Craton, Western Australia, Exploration Geophysics 36(2), 163–169
24. Fallon and Backo (1994), Archaean BIF-hosted gold, Mount Morgans, WA: a geophysical case history, Exploration Geophysics 25(3), 166
25. Jackson and Asten (2001), Borehole MMR at Marvel Loch gold mine: solving for the conductive overburden, ASEG Extended Abstracts 2001(1), 1–4
26. Geological Survey of Western Australia, Regional airborne electromagnetic data
27. Geoscience Australia (2022), AusAEM–WA, Murchison Airborne Electromagnetic Survey Blocks: SkyTEM airborne electromagnetic data and GALEI inversion conductivity estimates
28. Geoscience Australia (2021), AusAEM (WA) 2020–21, Eastern Goldfields and East Yilgarn AEM survey blocks: TEMPEST data and GALEI conductivity estimates, CC BY 4.0
29. Geological Survey of Western Australia, EIS-funded petrophysics
30. Department of Mines, Petroleum and Exploration, WA, Access mineral exploration reports (WAMEX)
31. Geological Survey of Queensland, Open Data Portal
32. Geoscience Australia and Geological Survey of Queensland (2025), AusAEM Northeast Queensland 2024 airborne electromagnetic survey
33. Anand and Paine (2002), Regolith geology of the Yilgarn Craton, Western Australia: implications for exploration, Australian Journal of Earth Sciences 49(1), 3–162
34. Bishop, Sattel, Macnae and Munday (2001), Electrical structure of the regolith in the Lawlers District, Western Australia, Exploration Geophysics 32(1), 20–28
35. Chalice Gold Mines (2020), High-grade nickel-copper-palladium sulphide intersected at Julimar Project in WA, ASX announcement, 23 March 2020
36. Zuo, Xiong, Wang, Wang and Kreuzer (2023), A new generation of artificial intelligence algorithms for mineral prospectivity mapping, Natural Resources Research 32(5), 1859–1869
37. Banaszczyk, Annetts and Dentith (2016), Towards resolving dipping contacts undercover in the Capricorn Orogen using AEM, ASEG Extended Abstracts 2016(1), 1–8
38. Buselli, Pik and Hwang (1998), AEM noise reduction with remote referencing, Exploration Geophysics 29(1), 71–76
39. Lane, Plunkett, Price, Green and Hu (1998), Streamed data: a source of insight and improvement for time domain airborne EM, Exploration Geophysics 29, 16–23
40. Schamper, Auken and Sørensen (2014), Coil response inversion for very early time modelling of helicopter-borne time-domain electromagnetic data and mapping of near-surface geological layers, Geophysical Prospecting 62(3), 658–674
41. Lankford, Perseus review, 2 February 2008, rev. 22 December 2008
42. Gamble, Goubau and Clarke (1979), Magnetotellurics with a remote magnetic reference, Geophysics 44(1), 53–68
43. Macnae, Lamontagne and West (1984), Noise processing techniques for time-domain EM systems, Geophysics 49(7), 934–948
44. Lankford, R-390A noise blanker, 14 July 1995, rev. 10 February 2005
45. Widrow, Glover, McCool et al. (1975), Adaptive noise cancelling: principles and applications, Proceedings of the IEEE 63(12), 1692–1716
46. SanFilipo (2006), Coaxial EMI sensor for UXO detection and discrimination, Geophex Ltd, ESTCP project UX-200418, DTIC ADA608474
47. Hou, Guo, Luo, Huang, Li and Yu (2026), The development and future prospects of multi-parameter aeromagnetic survey technology in China, Applied Geophysics 23(4), 1–10
48. Connelly (WA1ION), BFE-C balanced loop amplifier card, after Lankford’s 1991 design
49. Lankford, J310–J271 high-Z preamp IIP2 optimization, 6 April 2011, rev. 25 April 2011
50. Lankford, Common base transformer feedback Norton amplifiers, 8 June 1994, rev. 21 May 2007; also IRCA reprint M-079
51. Lankford, High reverse isolation push-pull Norton-Rohde feedback amplifiers, 6 November 2009, rev. 7 November 2009
52. Lankford, MiniFlag and MiniLoop antennas, rev. 7 March 2011
53. Lankford, Low noise active antennas AC/DC power supplies II, 26 November 2006, rev. 17 February 2007
54. Lankford, R-390A 10 dB S+N/N AM sensitivity, 7 July 2002, rev. 13 July 2002
55. CSIROpedia, LANDTEM: a transient electromagnetic method of mineral prospecting
56. Lee, Turner, Downey, Maddever, Dart, Foley, Binks, Lewis, Murray, Panjkovic and Asten (2001), Experience with SQUID magnetometers in airborne TEM surveying, Exploration Geophysics 32(1), 9–13
57. Smith and Annan (2000), Using an induction coil sensor to indirectly measure the B-field response in the bandwidth of the transient electromagnetic method, Geophysics 65(5), 1489–1494
58. Miles and Marr, US9817147B2: Airborne electromagnetic system with large suspension coil assembly, CGG Data Services AG, granted 2017
59. Kratzer and Macnae (2014), Prediction and removal of rotation noise in airborne EM systems, Exploration Geophysics 45(3), 147–153
60. Guo, Huang, Xu, Li, Li, Luo and Yu (2026), Principle of dynamic noise compensation in UAV aeromagnetic gradiometer systems, Applied Geophysics 23(1), 18–27
61. Ring masts at Naval Communication Station Harold E. Holt, Structurae
62. Mudge (1978), Geophysical surveys, Mariner 1 and 3, Black Bluff, Tasmania (EL 10/74), Geopeko Ltd; Mineral Resources Tasmania open file 78_1294
63. Kirwan (1969), Radem VLF electromagnetic survey, Kim claim group, Kimberley, BC, Imperial Oil Enterprises Ltd; BC assessment report 02071
64. Prikhodko, Bagrianski, Kuzmin and Sirohey (2022), Natural field airborne electromagnetics: history of development and current exploration capabilities, Minerals 12(5), 583
65. Nosyrev, Yurchuk, Gilmanova and Peskov (2026), Potential of large-scale magnetic survey using unmanned aerial vehicles in gold prospecting: the case of the Chulbatkan ore field, Russian Journal of Pacific Geology 20 (Suppl. 2), S181–S196
66. Cattach, Lee, Stanley and Boyd (1993), Sub-Audio Magnetics (SAM): a high resolution technique for simultaneously mapping electrical and magnetic properties, Exploration Geophysics 24(3), 387–399
67. Edwards (1974), The magnetometric resistivity method and its application to the mapping of a fault, Canadian Journal of Earth Sciences 11(8), 1136–1156
68. Seigel (1974), The magnetic induced polarization (MIP) method, Geophysics 39(3), 321–339
69. Mogi, Tanaka, Kusunoki, Morikawa and Jomori (1998), Development of grounded electrical source airborne transient EM (GREATEM), Exploration Geophysics 29(2), 61–64
70. Becken, Nittinger, Smirnova and colleagues (2020), DESMEX: a novel system development for semi-airborne electromagnetic exploration, Geophysics 85(6), E253–E267
71. Kotowski, Becken, Thiede, Schmidt, Schmalzl, Ueding and Klingen (2022), Evaluation of a semi-airborne electromagnetic survey based on a multicopter aircraft system, Geosciences 12(1), 26
72. Smit, Stettler, Price, Schaefer, Schodlok and Zhang (2022), Use of drones in acquiring B-field total-field electromagnetic data for mineral exploration, Mineral Economics 35, 455–465
73. Boggs, Stanley and Cattach (1999), Three-dimensional numerical modelling of sub-audio magnetic data, Exploration Geophysics 30(4), 147–156
74. Eremenco, Griffin and Cattach (2023), New developments with the HeliSAM and UAVSAM systems, 4th Australasian Exploration Geoscience Conference
75. Terraplus KT-10 susceptibility and conductivity meter, specifications, Geomatrix
76. Macnae and Hine (2016), Comparing induced polarization responses from airborne inductive and galvanic ground systems: Tasmania, Geophysics 81(6), E471–E479
77. Huang, Guo, Han and Huang (2025), Development and application of an aeromagnetic survey system for a large load rotary-wing UAV, Applied Geophysics 22(2), 546–553
78. Guo, Huang, Li, Fang, Liu and Liu (2026), Development of a UAV-based fully-airborne transient electromagnetic system, EGUsphere preprint, under review
79. Hronsky and Kreuzer (2019), Applying spatial prospectivity mapping to exploration targeting: fundamental practical issues and suggested solutions for the future, Ore Geology Reviews 107, 647–653
80. Zhu, Gu, Zhang, Zhang, Jia and He (2026), Mineral prospectivity mapping via interpretable machine learning techniques: a case study in the Tongling ore district, China, Natural Resources Research 35(2), 751–769
81. Christa, Suhari and Mane (2026), Self-supervised multi-modal fusion for mineral prospectivity mapping using vision transformers integrating multi-scale geological maps with aeromagnetic data, Natural Resources Research, online 2026
82. Zuo (2025), Key technology for intelligent mineral prospectivity mapping: challenges and solutions, Science China Earth Sciences 68(9), 2976–2991
83. Kreuzer, Roshanravan, Buckingham, Core, Konecke, McDwyer and Mustard (2025), Controls, expressions and discovery potential of gold mineralization in the central-eastern Yilgarn Craton, Minerals 15(12), 1255
84. Witt, Ford, Hanrahan and Mamuse (2014), Regional-scale targeting for gold in the Yilgarn Craton: part 1 of the Yilgarn Gold Exploration Targeting Atlas, GSWA Report 125
85. Luo, Zou, Wang, Song and Tang (2026), Stratified Monte Carlo sampled weights-of-evidence for gold prospectivity mapping in the Yilgarn Craton, Western Australia, Minerals 16(6), 629
86. Wilford, Searle, Thomas and Grundy (2015), Depth of regolith, TERN, CC BY 4.0
87. Lankford, Quoddy Head DXpedition, 19–26 October 2008, rev. 21 December 2008
88. Lankford, Grayland quad delta flag array report, April 2009, rev. 23 April 2010
89. Poudjom Djomani and Goodwin (2021), Towards standard technical deeds for (airborne) geophysical surveys in Australia, 3rd Australasian Exploration Geoscience Conference
90. Ray, McPherson, Brodie, Ley-Cooper, Tan and colleagues (2026), A decade of airborne electromagnetic surveying Lake Menindee (Australia) under varying water levels, arXiv 2603.19933, preprint
91. Davis and Macnae (2008), Quantifying AEM system characteristics using a ground loop, Geophysics 73(4), F179–F188
92. Wallner (AA7JV) (13 March 2020), Phased delta flag array test, Topband reflector
93. QMines (26 April 2023), Geophysical survey identifies 34 new targets, ASX announcement
94. Neroni, Blundell and Viezzoli (2024), Interpreting airborne electromagnetic data unburdened from induced polarisation effects: an unconventional mineral discovery case study from the eastern Yilgarn region of Western Australia, 1st ASEG DISCOVER Symposium
95. Tropicana Joint Venture, History
96. Martin (2015), Geophysics of the Tropicana gold deposit: minerals keynote paper, ASEG Extended Abstracts 2015(1)
97. Schodde (31 October 2023), Exploration: Australia vs the world, MinEx Consulting, IMARC
98. Schodde (November 2015), Australia discovery performance, MinEx Consulting, IMARC
99. Kreuzer, Etheridge, Guj, McMahon and Holden (2008), Linking mineral deposit models to quantitative risk analysis and decision-making in exploration, Economic Geology 103(4), 829–850
100. Mern and Caers (2023), The Intelligent Prospector v1.0: geoscientific model development and prediction by sequential data acquisition planning with application to mineral exploration, Geoscientific Model Development 16, 289–313
101. Civil Aviation Safety Authority, Flight approvals and permissions for drones
102. Joint Ore Reserves Committee, The JORC Code, 2012 edition
103. ASX, Guidance Note 31: Reporting on mining activities
104. ASX (21 April 2023), Listed@ASX Compliance Update 04/23
105. Joint Ore Reserves Committee, JORC Code update, status page, latest entry August 2026
106. Joint Ore Reserves Committee (10 September 2025), Draft JORC Code: summary of feedback and changes 2024–2025
107. Buselli (1982), The effect of near-surface superparamagnetic material on electromagnetic measurements, Geophysics 47(9), 1315–1324
108. Queensland Government, Mount Morgan remediation project
109. Taube (1986), The Mount Morgan gold-copper mine and environment, Queensland: a volcanogenic massive sulfide deposit associated with penecontemporaneous faulting, Economic Geology 81(6), 1322–1340
110. Arnold and Sillitoe (1989), Mount Morgan gold-copper deposit, Queensland, Australia: evidence for an intrusion-related replacement origin, Economic Geology 84(7), 1805–1816
111. business.gov.au, Co-funded Exploration Drilling Program (WA)
112. Ramelius Resources (2012), Deep diamond drilling at the Wattle Dam North project, EIS co-funded drilling report, WAMEX report A96373, as cited by Mueller (2025)
113. Government Geoscience Information Committee (March 2026), Australian guidelines for the submission of digital mineral exploration data, version 5.0
114. Mjelde (host), Dallas Lankford collection of modifications and projects, kongsfjord.no
115. SWLing Post (20 February 2022), International Radio Club reprints collection of 900 articles
116. DXer.ca (2025), IRCA reprint index
117. McDonald (VE7SL) (6 March 2015), Phasers on medium wave, relaying Steve Ratzlaff (AA7U)
118. Bath, Walshe, Cloutier, Rotherham, Hutchison, Hollebeck, Verrall and Schmitt (2020), Alteration patterns linked to high-grade gold mineralization at the Wattle Dam deposit, Western Australia, Ore Geology Reviews 125, 103471

About the author

Mike Peace (VK6ADA) designs receiving antennas and low-noise RF front ends. His published work on vk6ada.com.au includes active loop antennas with full KiCad fabrication packages, ferrite-core engineering, and an eight-JFET loop amplifier under development against the best commercial benchmarks.

He spent 11 years in Australia as an enterprise systems consultant to Western Australian mining, utilities and public-sector organisations. He also works as an expert witness on failed enterprise system implementations, where every conclusion has to survive cross-examination.

vk6ada.com.au is building a preservation mirror of the Dallas Lankford Files, first hosted by Bjarne Mjelde at kongsfjord.no. Contact: [email protected]

---

Mike Peace VK6ADA · r-390a.net and foxtango.org Administrator