On the bench
MEASURED · JULY 2026A Rev B board, tested against a written acceptance plan, with the raw data kept beside each report.
Two identical lanes with hardware arbitration between them, and heading from precise magnetics as well as GNSS.
We are building it from the sensors up. The AGS-M1 magnetometer board is built and field-tested, the AGS-G1 GNSS module is in design, and the flight control lanes come next. We are starting them by looking for an autopilot partner.
Two identical flight control lanes run side by side. The nodes both lanes need sit on both lanes' buses, with a guarded port for each, so a fault in one lane or on one bus is meant to stay there. It is in design, and nothing has flown.
THE FLIGHT CONTROL SYSTEM →Two identical flight control lanes. One commands and the other stands by. A lane that fails takes itself out of command in hardware, and the other takes over.
Fluxgate magnetometers, mounted away from the aircraft's own currents and read the way the AGS-M1 reads them, give a heading that needs neither GNSS nor a radio link.
A pair of AGS-G1 modules gives a second, independent heading, and each lane checks one source against the other. The GNSS heading needs both modules; without it, the lanes use magnetic heading.
Each lane checks every heading it is given and falls back on its own. The companion computer never switches a lane's heading source.
The flight control lanes are next, and we want to build them with a team that brings flight control software of its own. Tell us about your autopilot: king@airandgravity.com
The magnetometer the flight control system is built around, and a product in its own right: a three-axis fluxgate magnetometer board, built to go inside your design. On our bench its acquisition chain measures 0.035 nT/√Hz, with the inputs shorted at the sensor connector and the converter at 503.8 S/s; the board page shows how it was calculated. One board sits between the fluxgate and your host. Each row says where its value comes from, and anything not yet measured or specified is marked OPEN rather than left out.
FULL SPECIFICATIONS →
| Supply | 9 V nominal | DESIGN |
|---|---|---|
| Supply range | Not yet specified | OPEN |
| Host link | RS-485 half-duplex at 100,000 baud, or 3.3 V TTL serial at 921,600 baud; one link per power-up, set by a switch | DESIGN |
| Clock pin | J3 pin 5 carries the board's own 8 MHz as shipped; do not drive a clock into it on the shipping firmware | DESIGN |
| External clock | Runs from your 8 MHz master clock on J3 pin 5, and stops cleanly if it is lost, on the clock-following firmware we send on request (one board, 4 of 4 tests, July 2026) | MEASURED |
| Recording | ≈35,000 samples on board; every sample checksummed; resumable, verified readback | DESIGN |
| Accelerometer | Three-axis, on board; a coarse tilt witness | DESIGN |
| Connectors | Molex PicoBlade, 1.25 mm pitch, on the top side. J4 power, 0530470410: 1–2 DC in, 3–4 ground. J3 host link, 0530470610: 1 TTL serial TX, 2 RS-485 A, 3 RS-485 B, 4 TTL serial RX, 5 8 MHz clock (the board's own as shipped; yours with the clock-following firmware), 6 sync (no function in the shipping firmware); the serial, clock and sync pins are 3.3 V logic. J3 has no ground pin: its signal ground is J4's. J2 sensor, 0530470810: 1–3 X, Y, Z in, 4 analog return, 5 supply return, 6 +12.5 V, 7 −12.5 V, 8 self-test. | DESIGN |
| Mounting | Four non-plated Ø3.175 mm holes, each 2.667 mm in from both edges at a corner: a 37.846 × 89.916 mm rectangle, for M3 or #4 screws. Keep-outs: both faces are free of copper to Ø5.3 mm and of parts to Ø6.0 mm around each hole, and copper and parts stay at least 0.63 mm in from every edge. | DESIGN |
| Operating temperature | Not yet specified | OPEN |
The evidence log names the board and the conditions behind each milestone, and lists what is still open beside what is done.
FULL EVIDENCE LOG →A Rev B board, tested against a written acceptance plan, with the raw data kept beside each report.
A dual-port DroneCAN RTK GNSS module. For integrators of redundant UAS, and of survey aircraft that carry magnetometers.
THE AGS-G1 MODULE →GNSS on two CAN FD ports. Each has its own power stage, ground and transceiver, so a fault on one bus stays on that bus, and the module runs from either port's power.
A MUTE input on each port lets the autopilot on that bus silence the module's port in hardware, without relying on the module's firmware.
Two modules, strapped base and rover in the harness, report heading over DroneCAN without a magnetometer.
Designed from low-magnetic parts. Every unit will be measured on the fluxgate bench we use for the AGS-M1, and no figure is published until the first one is.
The simulator and the test vectors let your host code run before a board is on your bench. Both are working and early, and we send them on request.
INTEGRATION GUIDE →Everything your host needs, in one numbered document. In preparation.
Poll and response over RS-485 or TTL serial, with CRC-32 on every frame and a fixed register map.
The board's behaviour, with fault injection, for testing your host without hardware.
A Windows application for live streaming, spectrum and noise analysis, and verified recordings.
There is no cart and no price list. Every integration starts with an email to the person who designed the board: king@airandgravity.com
The interface control document is the one to start with.
ALL DOCUMENTS →| DOCUMENT | TITLE | REV | STATUS |
|---|---|---|---|
| AGS-ICD-002 | AGS-M1: interface control document | A | IN PREPARATION |
| AGS-DS-001 | AGS-M1: datasheet | A | IN PREPARATION |
| AGS-OM-003 | AGS-M1: operation manual | A | IN PREPARATION |
Air & Gravity Systems is a Canadian company. We design and build magnetic field measurement hardware, and we are designing the GNSS module and the redundant flight control system that fly with it.
Nothing here is claimed above its evidence.
We do not build weapons or anything that drops a payload.