Vigil
VIGIL-T · VIGIL-M
Twelve days in the dark at forty degrees north in December. Eleven minutes to erect it. One person.
Design · no towers built
Mission
The winter case is the sizing case
Hold a 360° volume of ground and low airspace, continuously, for a month at a time, with no power line, no data line and no operator on site.
VIGIL detects, classifies and tracks dismounts, vehicles and Group 1–3 unmanned aircraft, then hands a cued track to a shooter — a CAIRN node — or a commander — a PALL container — over REVENANT MESH. Every observation and every classification is hash-chained and signed on the tower before it leaves it.
The design case is not the southwest border. It is a Baltic winter treeline, a Polish logistics node, a Korean valley, an Alaskan approach: high latitude, a 26.6° December sun, snow on the glass, and four weeks between visits. A solar tower is an easy product at 32° north in August and a hard one at 40° north on 21 December, when the plane-of-array irradiance is 2.6 kWh/m²/day and the day is 9.3 hours long. We sized the December case and we published the recovery time that comes with it. The arithmetic is on this page.
VIGIL-T is the self-erecting trailer. VIGIL-M is the same sensor head on a 4.6 m vehicle mast, drawing from the host's 24 V bus, erected from the cab in ninety seconds. It also mounts on the MARROW deck, which turns an uncrewed ground vehicle into a mobile surveillance tower.
Key figures
Specification
Parameters
Design targets, not measured results, unless the line says otherwise. Ember-ticked rows are the numbers the rest of the design is subordinate to.
| Class | Persistent autonomous ground surveillance node, off-grid |
|---|---|
| Mast, deployed | 10.4 m (34 ft) sensor height, 5-section pneumatic telescope |
| Mast, nested | 3.1 m — high-wind and on-the-move sensing |
| Sensor head mass | 38 kg |
| Trailer, dry | 1,440 kg |
| GVWR | 2,000 kg — towed by a half-ton pickup, a HMMWV or a Land Cruiser |
| Footprint, deployed | 4.9 × 4.4 m over outriggers and solar wings |
| Battery | 38.4 kWh LFP, 51.2 V, 750 Ah, 245 kg — doubles as mast ballast |
| Solar array | 7.8 m², 1.68 kWp, fixed 70° tilt (45/70/90 pinnable), auto-stows above 55 kt |
| Tier 1 load | 92 W |
| Dark autonomy, Tier 1 | 12 days at end-of-warranty capacity, heater running |
| Dark autonomy, Tier 2 / 3 | 19 days / 57 days |
| Wind, operating | 65 kt at full extension |
| Wind, survival | 130 kt nested and stowed |
| Ice | 25 mm radial on the mast at 40 kt |
| Temperature | −40 °C to +55 °C operating |
| Comms | REVENANT MESH primary; LTE/5G, Ku SATCOM and 900 MHz long-haul optional |
| Autonomy | REVENANT CORE — full resident mission model, LEDGER on-node |
| Unattended interval | 30 days design; 90 days demonstrated at Tier 2 |
| Emplacement | 11 min, one person, level ground. No foundation, no permit, no concrete. |
| Stow | 14 min |
| Unit cost target | $310,000 at 200/yr from FORGE-1 |
| Mast, deployed | 4.6 m above the roof line, 3-section |
|---|---|
| Mast, on the move | 3.1 m, rated to 55 km/h |
| System mass | 210 kg all-up including mount, battery and sensor head |
| Battery | 6.8 kWh LFP; 12 A continuous from the host 24 V bus |
| Silent watch, engine off | 21 h Tier 1 / 96 h Tier 2 |
| Erect | 90 s from the cab |
| Sensors | Same head as VIGIL-T less the buried seismic string |
| On the move | Radar and RF DF operate at speed; EO/IR is stabilised but degraded above 25 km/h |
| Host platforms | JLTV, HMMWV, Bushmaster, Hilux / Land Cruiser 79, and the MARROW deck |
| Unit cost target | $186,000 at 400/yr |
| Radar | Ku-band (16.2–17.7 GHz), 4 × 90° MIMO panels, solid-state, no rotating parts. 360° × 80°, 1 Hz volume / 4 Hz on track, 640 simultaneous tracks, 34 W. |
|---|---|
| Dismount, 1 m² RCS | 3.4 km |
| Light vehicle, 10 m² RCS | 9.0 km |
| Group 1 sUAS, 0.01 m² RCS | 2.2 km |
| Group 3 UAS, 0.1 m² RCS | 4.6 km |
| Minimum detectable velocity | 0.35 m/s — a slow crawl is a real target |
| LWIR | 1280 × 1024 uncooled, 12 µm, 25–275 mm f/1.2. Human DRI 6.5 / 1.8 / 0.9 km. Vehicle 14.0 / 4.5 / 2.4 km. |
| Day channel | 4K CMOS, 30× optical, 12.5–500 mm. Human DRI 8.2 / 3.4 / 2.6 km. |
| Gimbal | 2-axis stabilised, 25 µrad RMS, 60°/s slew, 4 m geo-point accuracy at 5 km |
| RF direction finding | 70 MHz – 6 GHz, 4-element correlative interferometer, 3.0° RMS. Controller uplink 12 km, air-vehicle downlink 8 km, 400+ protocol library. 9 W. |
| Acoustic | 8-element MEMS array. Small-arms report 1.4 km, sUAS rotor 400 m, vehicle 900 m. 0.9 W. |
| Seismic / magnetic | Buried geophone and fluxgate string, 200 m of cable. Foot traffic 40 m, wheeled 250 m, tracked 600 m. 0.4 W. |
| Fusion | Multi-hypothesis tracker across radar, RF, acoustic and seismic. Track-to-classification 1.4 s including gimbal slew. |
| Tier 1 — full | 92 W · 11.8 days dark. Radar continuous, RF DF continuous, EO/IR cued, mesh live. Trigger: state of charge above 40 %. |
|---|---|
| Tier 2 — cued | 58 W · 18.8 days dark. Radar at 50 % scan duty, RF DF continuous, EO/IR on cue only, mesh receive plus burst. Trigger: 20–40 %. |
| Tier 3 — tripwire | 19 W · 57.3 days dark. Acoustic, seismic and magnetic only. Radar and EO/IR off. Mesh silent, burst on detect. Trigger: below 20 %. |
| Announcement | The tower publishes its tier to the network. A commander is told exactly what it stopped being able to see, and when. A tower that silently degrades is a liability. |
| Recovery hold | CORE holds the tower in Tier 2 until state of charge passes 60 %, which halves the December refill time |
| Top-up port | 2 kW DC input, MS3450. Four hours off a HMMWV alternator moves 6.4 kWh. |
| Optional fuel cell | 800 W methanol module, 6 kg plus a 20 L cartridge at 22 kWh, for units sited north of 55° N |
The hard claim
Twelve days, shown honestly
Twelve-day autonomy at 40° north in December is the hardest number on this page. Here is the entire arithmetic, including the part that is not flattering.
01 — Load
The claim is specific. It is twelve days of continuous Tier 1 operation, with zero solar input, on a battery at its end-of-warranty capacity, with the heater running. Not twelve days of standby. Not twelve days on a new pack.
| REVENANT CORE compute | 21 W — Orin NX-class SoC at a 20 W cap, inference gated by the tracker |
|---|---|
| Ku-band volume radar | 34 W continuous |
| EO/IR gimbal | 12 W at an 11 % slew duty — it is cued, it does not search |
| RF direction finder | 9 W continuous |
| REVENANT MESH radio | 10 W, receive continuous, transmit 4 % duty |
| Acoustic, seismic, BMS, lighting | 6 W |
| Total | 92 W → 2.349 kWh/day drawn from the pack at 0.94 conversion efficiency |
02 — Storage
| Nameplate at 25 °C | 38.4 kWh — LFP, 51.2 V, 750 Ah |
|---|---|
| Cold derate | × 0.95 → 36.5 kWh. Cells held above 0 °C by internal loss plus a 40 W thermostatic pad in the insulated ballast box. |
| Reserve floor, never spent | 8 % → 33.6 kWh. Heaters, BMS, and one burst transmission to say it is dying. |
| Discharge-path loss | × 0.94 → 31.6 kWh delivered |
| End-of-warranty fade | × 0.88 → 27.8 kWh delivered, after 3,000 cycles / 8 years |
| Tier 1 dark autonomy | 27.8 ÷ 2.349 = 11.8 days. Published as 12. |
On a new pack it is 13.5 days. We publish the end-of-life number because that is the number the tower will actually have in its fourth winter, which is the winter somebody will be relying on it.
The pack is 245 kg with its BMS, contactors and insulated enclosure. That mass is not a penalty. It sits low and forward in the ballast box, where a 10.4 m mast needs a counter-moment anyway. The battery is the ballast.
03 — Generation
Design point: 40° north, 21 December, continental United States, average cloudiness. Solar noon elevation is 90° − 40° − 23.44° = 26.6°. Day length is 9.3 hours. December mean global horizontal irradiance across that band is about 1.9 kWh/m²/day.
| Tilt | 70° fixed, south-facing, pinnable to 45 / 70 / 90. At a 26.6° sun a steep array is the correct answer, not a compromise — and it sheds snow. |
|---|---|
| Plane-of-array irradiance | 2.6 kWh/m²/day, December mean at 70° tilt, 40° N |
| Derate stack | 0.215 module × 0.97 soiling × 0.97 MPPT × 0.98 wiring × 0.97 charge acceptance × 0.88 snow allowance = 0.171 |
| Yield per m² | 2.6 × 0.171 = 0.445 kWh/m²/day |
| Break-even area | 2.349 ÷ 0.445 = 5.28 m² |
| Installed array | 7.8 m², 1.68 kWp → 3.47 kWh/day in December. Margin 1.48 ×. |
No tracker. A tracker earns most of its energy in summer, when the tower is already in surplus. December is the sizing case, and in December the sun traverses so little azimuth that a tracker returns under 9 %. It is not worth a gearbox at −30 °C.
04 — What we are not hiding
The three answers, in order of how much they actually help. One: CORE holds the tower in Tier 2 until 60 % state of charge — Tier 2 surplus is 1.99 kWh/day, which refills the pack in 14 days instead of 25 while the tower keeps its radar and its RF picture. Two: a 2 kW top-up port; four hours off a vehicle alternator moves 6.4 kWh. Three: an 800 W methanol fuel-cell module for units north of 55° N, where December solar is honestly not a strategy.
05 — Where the array goes
Why the panels are not on the mast
At 45 m/s — 100 mph — dynamic pressure is 0.613 × 45² = 1,241 Pa. Against 7.8 m² at a force coefficient of 1.3 that is 12.6 kN. Put that at the top of a 10.4 m mast and the overturning moment is 131 kN·m, which requires a foundation, which requires a concrete truck, which is the opposite of the product.
So the array does not go on the mast. Two wings fold out from the trailer deck, hinged at the frame rail, hydraulically assisted, sitting 0.9 m above grade. The overturning moment from the same wind is 11 kN·m against a 3.1 m wheel-and-outrigger track, carried by ballast alone. Above 55 kt the wings auto-stow flat in seventy seconds and the tower rides the storm out with the array horizontal and the mast nested.
Emplacement
Eleven minutes, one person
From a parked trailer to a tower on the mesh, executed by one soldier who never leaves the tongue.
| 0:00 – 1:10 | Park, chock, disconnect tow, confirm 6 m overhead clearance |
|---|---|
| 1:10 – 3:20 | Press deploy. Four electro-hydraulic screw jacks extend and auto-level to ±0.3° on a closed inclinometer loop |
| 3:20 – 5:00 | Solar wings unfold, hydraulic assist, 70° tilt, pins engage and are sensed |
| 5:00 – 6:30 | Mast rotates from its transport cradle through 90° to vertical on a single ram, pins |
| 6:30 – 9:10 | Compressor pressurises the mast to 8.5 bar and drives five sections up, collet-locking as each passes. Rate-limited to protect the head and the cable festoon. |
| 9:10 – 10:50 | Self-test: dual-antenna GNSS heading with celestial fallback, radar boresight, EO/IR autofocus and geo-registration |
| 10:50 – 11:05 | Mesh join. Ledger genesis block for this emplacement signed and broadcast. |
Concept of operations
Five phases
-
Phase 01 — Emplace
Drop a frontage in a morning
A two-vehicle team tows in and drops towers at 6–9 km spacing. Eleven minutes each. No survey, no foundation, no generator, no fuel plan. Spacing is set by the 2.2 km sUAS detection range if the threat is drones, or by the 9.0 km vehicle range if the threat is ground movement.
-
Phase 02 — Self-register
One picture, not five feeds
Each tower gyrocompasses, geo-registers its EO/IR against the terrain model, and negotiates a mesh topology with its neighbours. Any two towers in RF line of sight cross-fix an emitter to a 90 m CEP. The battery publishes one common track picture. Nobody sits at five monitors.
-
Phase 03 — Watch
Radar and RF search, EO/IR confirms
Nobody is on site. When the tracker crosses a confidence threshold set by the ROE loaded at emplacement, the tower pushes a classified track to PALL and — if it is an air threat — direct to CAIRN over MESH in 18 ms. Everything is signed as it goes.
-
Phase 04 — Degrade
The mesh is cut. The tower is not.
It keeps tracking, keeps classifying, keeps signing, and buffers the LEDGER chain to NVMe. When any link reappears — a MARROW driving past, a soldier within 400 m wearing a VEIL — the buffered chain reconciles into the battalion picture in order, signed, with its original timestamps.
-
Phase 05 — Recover or abandon
A loss, not a compromise
Fourteen minutes to stow and tow. If the position is being overrun, a signed abandon command zeroises the crypto, wipes the mission cache, and leaves the LEDGER chain sealed and unreadable. The tower becomes a $310,000 loss instead of an intelligence gift.
Engineering rationale
Why it is built this way
No foundation
Anything requiring concrete is a coordinate on an adversary's target list eight weeks before you finish pouring it. VIGIL is emplaced, moved and re-emplaced in a morning. The tradeoff: a ballasted trailer has a lower wind rating than a piled mast, which is why the operating limit is 65 kt and not 90.
The gimbal does not search
A continuously scanning EO/IR is a 34 W average load instead of 12 W, and it is a mechanism running around the clock for a year without a technician. Radar and RF search; the gimbal confirms. The tradeoff: a target that is invisible to radar and silent on RF is not found.
Uncooled LWIR, not cooled MWIR
Cooled MWIR would add 3–4 km of recognition range and 28 W of continuous cryocooler draw, which costs 3.6 days of the twelve-day claim and adds a 12,000-hour wear item to a box that is supposed to be unvisited. The tradeoff: it is a bolt-on module if the customer wants it, and autonomy drops to 8.4 days. In a humid tropical AO they should buy it.
Steep array, no tracker
The sizing case is a 26.6° sun. A steep fixed array beats a tracked flat array in December, sheds snow, and has no gearbox to seize at −30 °C. The tradeoff: we leave roughly 20 % of the summer harvest on the table. In summer the tower is in surplus anyway.
The battery is structural
245 kg of LFP in the ballast box is mass you were going to carry as dead weight to hold the mast down. The tradeoff: the pack is now a structural item, so a pack replacement is a levelled, jacked operation and not a tailgate swap.
It publishes its own degradation
A tower that quietly drops from full search to tripwire is worse than a tower that is off, because the commander still believes he is covered. VIGIL announces its tier. The tradeoff: a customer reading the tier log will see how often weather puts the tower into Tier 2, which is an uncomfortable conversation we would rather have up front.
Program candor
Open engineering risks
- The twelve-day claim depends on a 92 W load holding across four winters. Firmware grows. Every added inference model is a watt. We hold the number with a hard power cap in CORE and by refusing to ship anything that breaks it — which means saying no to customer feature requests. That is an organisational risk, not a technical one, and organisational risks are the ones that actually kill numbers like this.
- Snow occlusion is modelled at 12 % and it is the softest figure on the page. A wet snow that bonds to a 70° panel at −1 °C and then freezes will hold for days. The mitigations are the hydrophobic coating and a back-feed defrost mode that costs 0.9 kWh per clear. We have not yet run a full northern winter.
- Twenty-five-day December recovery from a full drawdown is mitigated by the Tier 2 hold and the top-up port. It is not eliminated.
- Pneumatic masts fail wet and cold. Collet locks and seals are the wear item. Design life is 1,200 cycles to first seal replacement — a long time for a fielded trailer, not long for a training unit that erects it twice a day.
- Uncooled LWIR loses to cooled MWIR in hot, humid, low-contrast conditions. We chose the power budget over the range and we will say so to a customer whose AO is maritime tropical.
Related
What it cues and what it rides on
Contact
Send us your latitude and your visit interval
Give us the site latitude, the December cloud statistics, the sensor set you need up continuously, and how often a vehicle can get to it. We will run the same arithmetic on this page against your site and tell you the tier the tower will actually sit in — including the sites where the answer is a generator.
Every figure on this page is a design target derived from first-principles analysis, not a measured result. No VIGIL tower has been built or emplaced. 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.