Cairn
Ground counter-UAS effector node
The cheapest effector that can service the target is the one that fires. The node decides in under four hundred milliseconds, and signs why.
Design
Mission
Counter-UAS is an arithmetic problem, not a hit-probability problem
Defeat unmanned aircraft over a fixed or semi-fixed asset — an airfield, a logistics node, a command post, a bridge, a pier — at a cost per kill that does not bankrupt the defender.
Everybody's probability of kill against a quadcopter is fine. That is not where this fight is lost. It is lost on magazine depth and cost exchange. A $600 airframe that costs $118,000 to shoot down wins even when it loses, and forty of them arriving in ninety seconds beat any node with four rails on it regardless of how good those four shots are.
CAIRN's answer is a ladder of five effectors in one emplacement, on one power plant, in one fire-control chain. Rung selection is resolved by REVENANT CORE on the node in under 400 ms, with no network, and every selection — including the rungs that were rejected and why — is written to REVENANT LEDGER.
Key figures
The argument
The effector ladder
Rungs 0 and 1 are automatic with a human on the loop. Rungs 2 to 4 require a human authorisation unless the ROE signed at emplacement explicitly delegates.
| Rung 0 — RF defeat | $0.04 per engagement · 2.5 km · unlimited magazine, power only. Protocol-specific link takedown and bounded-volume GNSS denial against Group 1 commercial airframes on a known protocol. The real cost of this rung is spectrum fratricide, not money. |
|---|---|
| Rung 1 — High-power microwave | $0.60 per burst including amplifier-module life amortised over 20,000 shots · 900 m in a 25° cone · 8 s recharge · unlimited magazine. Eight to fourteen airframes per composite burst. This is the rung that answers mass. |
| Rung 2 — 30 kW fibre laser | $95 per engagement honestly amortised — $0.15 if you only count electrons · 2,200 m in clear air · ~14 s dwell · one target at a time · unlimited while fuelled. Weather-limited; see the risks. |
| Rung 3 — 30 × 113 mm gun | $1,330 per engagement — a 14-round burst at $95 a round · 200 rpm · 240 rounds ready · 1,800 m against Group 1–2, 2,400 m against Group 3. All weather. This rung exists because fog is not an edge case in northern Europe; it is November. |
| Rung 4 — BANSHEE-M | $0 if it does not engage — it returns, refuels and re-arms itself in nine minutes unattended — and ~$118,000 if it does · 12 km+ · 4 on the node's rail, more from the battery. Cued from CAIRN, not organic to it. Group 3–5, cruise missiles, anything past the gun. |
| Interceptor node, 4 rails | Engages 4, regenerates once, engages 4 more. 32 leakers. $944,000 spent and the asset is hit. |
|---|---|
| Interceptors, unlimited rails | 40 kills for $4,720,000 |
| CAIRN | 3 HPM composite bursts service 8–14 each inside the cone, the gun mops up 6 outside it, RF defeat takes 4 still on a factory protocol. $7,982 for 40 kills — $200 each. |
| Cost exchange | 590 × in the defender's favour. That is the entire reason the node carries four effectors instead of one good one. |
Architecture
CAIRN is a shooter. It is not a sensor.
No search radar
CAIRN carries no volume-search radar. It takes search from a VIGIL tower, a BANSHEE picket or a divisional sensor over REVENANT MESH, and carries only what it needs to shoot.
This is an architectural decision, not a cost saving. A node that searches is a node that emits continuously, and therefore a node an adversary can find and hit. CAIRN emits only when it is about to shoot. In EMCON it sits on the mesh with its fire-control radar cold, cued by a tower eight kilometres away.
If the mesh is cut, CAIRN falls back to its own fire-control radar in a 90° staring sector plus the RF direction finder and the tracking gimbal — degraded to roughly 30 % of its cued engagement volume. It reports exactly that to the network before the network disappears.
| Ka-band fire-control radar | 6 km, 0.2 mrad angular, 0.1 m range resolution. Track, range and hit assessment. Cold until commit. |
|---|---|
| MWIR + day gimbal | Cooled MWIR 1280 × 1024, 45 µrad, boresighted to the beam director within 0.05 mrad. Aimpoint refinement and visual ID for the ROE. |
| RF direction finder | 70 MHz – 6 GHz, 3° RMS. Rung 0 protocol identification, and the target's control link as an independent cue. |
| Search | Off-board. VIGIL, BANSHEE picket, or a divisional sensor over REVENANT MESH — 18 ms cue latency. |
| The problem | A 30 kW laser draws 86 kW electrical during dwell. You cannot step-load a diesel generator by 86 kW. It droops, the frequency sags, and on the third engagement it trips. |
|---|---|
| The answer | A 48 kWh LFP buffer between the generator and the effectors. The generator sees a smooth 5–15 kW charging load; the battery serves the pulses. |
| Also serves | The HPM pulse-forming network, which is a far worse load — hundreds of kilowatts for milliseconds. |
| Prime power | 125 kW tactical generator or a vehicle power take-off |
| Generator off | 143 laser engagements, or 37 h of silent watch at the 1.3 kW watch load. Thermally cold, acoustically silent, full directed-energy magazine. |
Specification
Parameters
Design targets, not measured results, unless the line says otherwise.
| Emplacement footprint | One 20 ft ISO footprint for the turret and effectors, plus a 4.5 m power skid |
|---|---|
| Mass, effector module | 6,080 kg — turret 1,850; laser head and chiller 1,100; HPM array 640; gun and 240 rounds 780; buffer battery 310; structure 1,400 |
| Transport | CH-47F sling, HEMTT-LHS flatrack, PLS, commercial chassis |
| Turret | 2-axis, 360° continuous azimuth, −12° to +82° elevation |
| Slew | 120°/s azimuth, 90°/s elevation, 220°/s² acceleration |
| Pointing accuracy | 0.15 mrad |
| Levelling | Four jacks, five minutes, then gyrocompass and geo-register |
| Environmental | −40 °C to +55 °C. The laser derates in fog, rain and dust — see the risks. |
| Unit cost target | $4.8M at 120/yr from FORGE-1 |
| Cue over MESH | 18 ms from a VIGIL tower |
|---|---|
| Correlation and rung selection | 390 ms — on the node, no network, ROE checked |
| Turret slew, worst case | 1.5 s from 180° off |
| Fire-control lock | 0.4 s |
| Human authorisation | 1.5–4 s when the ROE requires it |
| Cue to effect, automatic | 2.3 s |
| Cue to effect, human on the loop | 4.5 s. The human is the long pole, and we are not going to pretend otherwise. |
| Escalation after a failed engagement | CORE steps one rung and re-engages without a new authorisation, because the authorisation was against the target, not the effector. Delegated only by explicit signed ROE, and time-boxed. |
| Mass-raid mode | Above a raid-size threshold, per-target selection switches to sector allocation: HPM takes an azimuth sector on composite, the gun takes leakers outside the cone, rung 4 releases only against tracks classified above Group 2. |
| Record | Track history, classification, model version hash, full rung evaluation with the reason each cheaper rung was rejected, the authorising identity, and the hit assessment — one signed LEDGER block per engagement |
Concept of operations
Six phases
-
Phase 01 — Emplace
Rung lockouts are part of the record
Dropped by flatrack at the asset. Four jacks, five minutes, gyrocompass, geo-register, join the VIGIL battery on MESH. The ROE for this emplacement — which rungs are delegated, which are human-authorised, which are locked out entirely by overflight and collateral constraints — is loaded and signed at emplacement. It is not a runtime setting somebody can quietly change.
-
Phase 02 — Cued watch
Fire control cold
VIGIL searches. CAIRN listens. No emission, no thermal plume from a generator that does not need to run, 37 hours of watch on the buffer.
-
Phase 03 — Cue
Eighteen milliseconds and a decision
A track crosses threshold at a tower 8 km away. CORE correlates it against its own RF direction finder, resolves the classification, checks the ROE, and selects the cheapest rung whose envelope and confidence permit the engagement. 390 ms.
-
Phase 04 — Engage
Slew, lock, authorise, effect
Turret slews, fire-control radar acquires, and either the automatic rung fires or an authorisation is requested and granted. Hit assessment on the radar and the gimbal. A failed rung escalates one step — under signed, time-boxed delegation only.
-
Phase 05 — Sign
The artefact that survives the investigation
Not just "we shot it." The signed block says: RF defeat evaluated and rejected, protocol unknown; HPM evaluated and rejected, friendly quadcopter inside the cone; gun fired; here is the chain. That is what a board of inquiry asks for, and it is the difference between keeping the capability and losing it.
-
Phase 06 — Mass raid
A human authorising forty times in ninety seconds is a denial of service
Above the raid threshold, CORE switches to sector allocation and rung 0–1 authority is pre-delegated for the duration — signed, time-boxed, and closed automatically when the raid ends.
Engineering rationale
Why it is built this way
Four effectors, one emplacement
Assembling the same capability from four separate boxes gives you four networks, four generators, four maintenance chains and a fire-control deconfliction problem that nobody owns. The tradeoff: a single node is a single point of failure and a fat target, which is why siting and the cued architecture matter as much as the effectors.
Cheapest-rung-first is an architecture
It has to resolve inside the fire-control loop in under 400 ms, at the node, with no network. Make it a doctrine document instead and the operator reaches for the effector he trusts, which is always the expensive one. The tradeoff: a machine choosing the weapon is exactly the automation a customer will scrutinise hardest, which is why the rejected rungs are recorded too.
The gun is on the node because weather is real
A directed-energy-only node is a fair-weather node. The tradeoff: 780 kg of gun and ammunition, a resupply chain, and fragments that land somewhere — which is why the rung is often locked out over populated ground.
No search radar
A node that emits is a node that dies. Search comes from a tower that costs $310,000 and is cheap to replace. The tradeoff: a mesh-isolated CAIRN covers about 30 % of its cued engagement volume, and it says so.
The buffer, not a bigger generator
A 48 kWh battery lets a 125 kW generator serve an 86 kW pulsed load, and it buys 37 hours of silent watch as a side effect. The tradeoff: 310 kg and a fire-safety design problem on a vehicle that also carries ammunition.
The ladder records what it did not fire
The defensible artefact is not the shot. It is the evaluation. Every rung considered, every rejection reason, the model version, the authorising identity. The tradeoff: a dedicated append-only appliance and the discipline to make the fire-control chain explain itself in 390 ms.
Program candor
Open engineering risks
- High-power microwave is not selective. Inside the cone it will drop friendly unmanned aircraft and upset unshielded electronics, including civilian ones. CORE geofences the cone against the friendly track file — but that file is only as good as its last update, and in a mesh-denied fight it is stale. HPM in a cluttered or urban environment is a spectrum and fratricide decision before it is a kinetic one, and it is the rung most likely to be locked out.
- The laser is weather-limited and we will not hide it. Fog, rain, blowing dust and marine haze cut effective range by 50–80 %. In a Gulf summer the 2,200 m laser is a 900 m laser. In Baltic fog it is a 600 m laser.
- The gun's fragments land somewhere. The rounds self-destruct, but an engagement at 1,800 m puts a fragment pattern into somebody's ground. Over a populated area the gun rung is commonly locked out, and when it is, cost per kill rises to the laser or the interceptor. The cost-exchange argument is weaker in cities and a customer should hear that from us, not discover it.
- A 125 kW generator is a thermal and acoustic beacon. Silent watch is 37 h on the buffer. After that CAIRN is audible and visible in infrared. The mitigation is siting and a hybrid duty cycle, not a technical fix.
- Rung escalation after a failed engagement is the highest-risk automation in the entire land portfolio. A machine that decides on its own to step from a jammer to a gun, against a target a human authorised two seconds earlier, is exactly the behaviour that ends a program if it is wrong once. It is delegated only by explicit signed ROE, it is time-boxed, and it is the most heavily instrumented path in LEDGER.
- Integration risk dominates schedule risk. Four effectors, three sensors, one turret and one 86 kW pulsed power system is a systems-engineering problem, not a physics problem — and systems-engineering problems are the ones that slip.
Related
What cues it and what it reaches for
Contact
Bring us the raid and the ROE
Send us the threat profile, the raid size, the collateral constraints on your site and the rungs your rules of engagement will actually let you use. We will model the ladder against it and give you a cost per kill — including the sites where the ROE locks out enough rungs that the arithmetic stops working in our favour.
Every figure on this page is a design target derived from first-principles analysis, not a measured result. No CAIRN has been built and the selection logic has not been exercised against a real effector inventory. Where a number here is later measured and comes out worse, we will publish the measured number. No export-controlled information is published on this site.