Lookout runs out of time.
Two gliders converging at a combined 200 km/h close the last kilometre in eighteen seconds. For most of that time the other aircraft is a few grey pixels against grey — and here is the cruel part: an aircraft on a genuine collision course does not move across your canopy. Constant bearing, decreasing range. It sits perfectly still in one spot of your field of view and simply grows. The human eye is a motion detector, and a threat that holds still is the one it is least likely to find.
Now add the rest of a real cockpit: a canopy frame exactly where the traffic is, eyes that drift to arm's length when there is nothing to focus on, attention on the vario and the next thermal, and a gaggle in which everybody is turning. None of that is an argument for looking out less. It is an argument for a second observer — one that never blinks, never fixates, and can see straight through the fuselage behind you.
FLARM — flight alarm, developed in Switzerland from 2004 for exactly this problem — is that observer. What makes it work is not a better sensor. It is that every aircraft cooperates.
Not radar. A conversation.
Radar and TCAS discover traffic the expensive way: they push energy out and listen for what bounces back or what answers. That costs power, weight, certification and money, which is why a 200-tonne airliner carries TCAS and a 300-kilogram glider never will.
FLARM inverts the problem. Nobody interrogates anybody. Each aircraft simply says where it is and where it is going, once or twice a second, on a low-power radio channel — and listens to everyone else saying the same thing. No central station, no ground infrastructure, no subscription, no negotiation. Two units that have never met need only share a protocol.
The price of that elegance is one strict rule: FLARM can only warn you about aircraft that are also transmitting. Cooperative means cooperative.
What actually goes out on the air.
About once a second your unit fixes its position with a GNSS receiver, reads barometric altitude, and packs a very small message: position, altitude, velocity vector, vertical speed, turn rate, aircraft type and a 24-bit identity. In Europe, Africa and Asia that goes out on 868.2 and 868.4 MHz — one to two messages per second on each channel. In the Americas, Australia and New Zealand the same system runs in the 902–928 MHz band.
Transmit power is a few tens of milliwatts — roughly a car key fob. There is no magic in the radio, and the range you get, typically three to ten kilometres, is decided almost entirely by the antenna and by what stands between it and the sky. That is a subject of its own, and it is by far the most common reason a perfectly healthy FLARM performs badly.
Transmissions are short bursts, slotted against GPS time and hopped between the channels, so that thirty gliders stacked over a start line do not simply talk over each other. The packet itself is compact and obfuscated; the motion-prediction protocol behind it, FAMP, has since been published so that other manufacturers can interoperate instead of reverse-engineer.
The prediction is the product.
Here is what separates FLARM from a moving map with other aircraft drawn on it. What goes out is not merely a position — it is the input to a flight-path prediction, and every receiving unit runs that prediction forward for every aircraft it can hear, alongside its own.
The model uses speed, acceleration, ground track, turn radius, vertical speed, wind and the configured aircraft type. A glider in a twenty-second circle is not predicted to fly straight ahead — its arc is predicted. A tug on a straight climb-out is predicted to keep climbing. Then your unit asks one question of every pair of paths: do these two futures come close enough, soon enough, to be a problem?
That is why a thermal full of gliders is quiet. Twelve aircraft two hundred metres apart in the same circle are, by any distance-based rule, in permanent violation. By trajectory they are doing precisely what they should — their arcs never meet. FLARM warns about geometry, not proximity, and a system that cried wolf in every gaggle would be switched off before the end of the first season.
Slide the traffic in and watch the level climb.
The default is a genuine collision course. Pull RANGE down and nothing about the geometry changes — only the time left does, and the level climbs 1, 2, 3 as it passes eighteen, twelve and eight seconds. Now swing INTRUDER TRACK twenty-five degrees either way at the same range: the traffic is just as close and the alarm disappears, because predicted separation is what decides. The real risk model is richer than this one — it weighs uncertainty that grows with every second it predicts, which is the widening wedge — but the question it asks is exactly this question.
Three levels, measured in seconds.
When the prediction does find a conflict, the warning is graded by time to impact — never by distance.
Seconds are the honest unit. Five hundred metres is a lifetime between two gliders drifting apart and almost nothing between two aircraft closing head-on at 400 km/h — the same distance, two entirely different emergencies. Time to impact collapses both into one number a pilot can act on.
It also means an alarm can drop a level or vanish outright as the picture updates. That is the system working, not the system glitching: somebody rolled out, somebody climbed, and the two futures no longer meet.
A warning that points somewhere.
An alarm has to answer "where", and answer it faster than you can read. Because every message carries the other aircraft's position, your unit knows the relative bearing and the relative altitude, and the display shows exactly that: an arrow or an LED at the right clock position, a plus or a minus for above or below, and a sound whose urgency matches the level.
This is the part that decides whether the seconds are any use. Two seconds spent decoding a screen at level 3 are two seconds of the eight you had. A good traffic display is read the way an altimeter is read — in a glance, from the corner of the eye, without changing what your hands are doing.
The wires it warns you about too.
The same engine solves a second problem for nothing extra. Load the obstacle database — power lines, cable cars, masts, ridge cables — and each obstacle behaves like an aircraft that never moves. Your predicted path is tested against the wires exactly as it is tested against traffic, and you get the same graded warning before you fly into a span nobody sees against a mountainside. In alpine flying that feature alone has justified the box more than once.
What else is up there: ADS-B and transponders.
Not everyone flies a FLARM. A PowerFLARM-class unit therefore carries a second receiver on 1090 MHz and listens to the transponder world as well — and the two kinds of target behave very differently.
ADS-B Out aircraft broadcast their own GNSS position, unprompted, exactly as FLARM does. They arrive complete: bearing, distance, altitude, identity, and they go through the same prediction as everything else.
Mode-S and Mode-C transponders only answer when something interrogates them — a ground radar, or a TCAS-equipped aircraft nearby. Their replies contain no position at all. Your unit can estimate distance from signal strength and read altitude out of the reply, but it cannot give you a bearing, which is why such traffic shows as a ring or a bare distance instead of an arrow. That is a property of the signal, not a shortcoming of the display. Below radar coverage they may not reply at all.
Between the two receivers, most of what shares your airspace announces itself somehow: 868 MHz covers the gliders, tugs, microlights, helicopters and drones no transponder would ever mention, and 1090 MHz covers the powered traffic that flies with one.
Where it is blind.
A safety device you trust incorrectly is worse than one you do not trust at all, so it is worth being precise about the holes.
Half of that list is installation, not electronics. An antenna in the shadow of a carbon fuselage or lying flat on its side turns a ten-kilometre system into a two-kilometre one, and the display looks entirely normal while it happens. Verify the range you actually have — the antenna article is about nothing else.
And the biggest limitation is not technical at all: FLARM does not fly the glider. It buys you seconds and it tells you where to look. The avoiding action is still yours, and the aircraft it never heard is still found the old way, with your eyes.
The second life of the data.
Every packet your glider transmits is also heard by anyone with a receiver on the ground. A volunteer network of exactly such receivers — the Open Glider Network — turns that into live tracking: the club watches its fleet on a map, the crew follows a final glide, a competition sees the whole grid at once.
The same stream is why some searches have become short. When an aircraft goes missing, a last received position with a timestamp turns a valley full of mountains into a point on a map, and rescue has repeatedly reached people in the time it used to take merely to organise the search. Nobody designed the system for that. It is what falls out when every aircraft says where it is once a second.
The trade is privacy, and it is configurable: units carry privacy and no-track flags for pilots and competitions that would rather not be followed live. Your own IGC flight log records every contact regardless — which is exactly how you check the real reception range of an installation after you have finished it.
In your cockpit.
Everything above reduces to one small radio message and one prediction. What differs between two aircraft is how well that message gets out, and how fast the pilot can read what comes back — an antenna problem and a display problem. LX builds both ends: complete FLARM and ADS-B units, traffic modules on the Navia backbone, and displays from a 57 mm radar screen to an LED bearing ring that costs almost no panel space.
If you take one thing away: the box is only worth the seconds it can give you, and those seconds should be spent looking outside, not reading. Set it up so the alarm points, then go and measure your range.
FLARM and ADS-B IN in one box, streamed to your devices over WiFi — the standard fit for club and competition gliders.
The installed module on the Navia backbone — collision avoidance feeding every display in the panel.
A dedicated 57 mm traffic display for the FLARM you already have — radar view, warnings, target info.
An LED bearing ring that needs almost no panel space and is read without looking down.