Why point-to-point wiring ran out of room.
Connect every instrument to every other instrument and the number of links grows with the square of the number of boxes. Four devices is six links; eight devices is twenty-eight. Add a second flight computer in the back seat and the loom behind the panel becomes the heaviest, least serviceable and least documented component in the aircraft. Every one of those links is also a place for a crimp to fail, and it will fail in the air rather than on the bench.
A bus replaces the whole geometry with a single shared pair. Every device listens to everything and speaks when it has something to say; adding a ninth device means one more connector, not eight more wires. The wiring stops scaling with ambition, which is precisely what lets a vario, an AHRS module, a joystick and a flap sensor coexist on one 40 cm cable that also carries their power.
What CAN actually does with two wires.
CAN was designed in the 1980s for cars, which is to say for a hot, shaking, electrically filthy environment where a lost message is not acceptable — an excellent description of an aircraft panel. It carries data differentially: the two wires always move in opposite directions, and the receiver reads only the difference between them. Interference from a radio transmission hits both wires equally and cancels out of the subtraction. That single design choice is most of why the bus survives a panel with a transponder in it.
The second clever part is how two devices that speak at once are resolved. Each message begins with an identifier, and the bus is wired so that a zero always wins over a one. Every transmitter listens to itself while sending; the moment a device sees a zero where it sent a one, it knows a higher-priority message is in progress and stops — without any data being destroyed. The important message continues uninterrupted. There is no master, no polling, and no collision recovery time.
The rest is housekeeping that matters in flight: every frame carries a checksum, a corrupted frame is retransmitted automatically, and a device that has become a nuisance takes itself off the bus rather than jamming it. From the pilot's seat this is invisible, and that is the point — the failure of one accessory does not take the panel with it.
Terminate it, or explain it later.
Switch one terminator off. Nothing dramatic happens to the picture — the edges just ring a little, and a bus with a small margin still works, on the bench, on a warm day. Switch off the second and the ringing crosses the decision threshold: frames fail their checksum, get retransmitted, and the panel develops the symptom every avionics shop knows by heart — an intermittent fault that only appears in flight. The two resistors are not optional extras. They are the reason the cable behaves like a transmission line instead of an antenna.
Power and data on the same wire.
The cable in a modern panel carries supply as well as signal, which is what makes a one-cable install possible — and which quietly imports the whole of the power article into the wiring diagram. Every device on the chain adds its current to the same conductors, and the resistance of those conductors turns that current into a voltage drop that the last device in the chain feels the most.
The tool above shows the arithmetic, and the practical conclusions are short. Feed the bus somewhere near its electrical middle rather than at one end. Keep the run as short as the airframe allows. And do not hang a genuinely hungry load — a big display, a transmitting device — off the end of a thin shared cable simply because the connector fits. A bus is a wonderfully efficient way to move information. It is still only copper when it comes to amps.
Five faults, all of them installation.
The classic. Two devices on a short bench cable work perfectly; the same pair with three metres of wing wiring throws error frames only in turbulence, only in summer.
A T-junction with long legs is not a bus. Stubs must be short — centimetres, not metres — or each one becomes its own little reflecting antenna.
Feeding the whole panel through the last device in the chain sends every amp down the whole cable. Feed near the middle, or feed heavy loads separately.
RJ45 in an aircraft is a vibration environment. A conductor that is merely touching will conduct on the ground and stop conducting at 2000 m in wave.
Two devices grounded at two points of an airframe with a current flowing between them do not agree on what zero volts means.
When a bus is no longer enough.
CAN was built to move small, urgent messages — an airspeed, a flap position, a warning — with guaranteed priority. It was never built to move a moving map. The moment a panel wants full-HD displays, live weather, camera feeds and software updates, the shared pair runs out of room, and the architecture changes shape: from a chain that everyone shares to a switch that gives every device its own gigabit path, with power delivered down the same Ethernet cable.
That is the step our NAVIA platform takes — one CAT 6 cable per device carrying 1 Gbit of data and up to 100 W of power, a hub at the centre instead of a chain, and no solder joints anywhere in the installation. The old virtue survives the transition intact: one cable per box, and the wiring stops being the hard part of owning an aeroplane.
AHRS module, LX Joy and the flaps sensor all arrive on one CAN cable that carries power and data together.
One CAN bus between the main unit and the vario indicator — the whole system upgrade with two cutouts and one cable.
Where the bus becomes a network: gigabit Ethernet and 100 W of power over a single CAT 6 cable per device.