Kestrel

Every gram on the gimbal is a minute off the flight.

The client's existing payload worked well and weighed too much. On their most popular airframe it cut flight time by a third and needed two people and a hex key to swap. Operators were choosing lower-quality integrated cameras instead, and the client was losing the sale.

They asked for the same five-band imager and stabilisation performance at half the mass, a swap that one person could do in the field in under a minute, and a thermal design that would hold sensor calibration from a cold morning start through a midday flight.

Took the structure out of the housing and put it in the electronics.

The old design was a camera block bolted inside a machined shell. We inverted it: the main PCB is a rigid-flex that also forms the structural spine, the sensor modules mount directly to it, and the housing is a thin magnesium shell that carries no load. That removed most of the fasteners, all of the internal harnesses, and about forty percent of the mass on its own.

The gimbal uses the client's proven motors with a new carbon yoke and a redesigned slip ring. We modelled the moment of inertia through every packaging iteration, because the stabilisation loop could only be as good as the mass distribution it was fighting.

The quick-release is a quarter-turn bayonet with blind-mate power and data, keyed so it cannot go on wrong. One hand, no tools, and it clicks. We tested it through five thousand cycles with grit in the joint before we were satisfied.

Thermally, the imagers sit on a common heat spreader tied to the magnesium shell, with a small heater that brings the sensors to their calibrated temperature before take-off and holds them there. Firmware handles the warm-up, reports readiness to the flight controller, and streams five bands with per-frame metadata to the client's existing processing pipeline.

• The rigid-flex spine is the structure, the harness and the ground plane. Three parts became one.
• The bayonet ring is anodised in a colour that matches nothing else on the aircraft, so it is the obvious thing to grab.
• The heater's power budget was traded against flight time explicitly. Ninety seconds of warm-up on the ground beat a drifting calibration in the air.
• We designed three mounts up front for the client's three most common airframes, so the same payload sells into all of them.
• The vibration test used spectra logged on real flights, not a standard profile. The standard profile would have passed a design that failed in the field.

Kestrel launched on three airframes and returned flight time on the client's most popular platform to within a few minutes of an unloaded aircraft. The design files, test procedures and fixtures are in the client's vault, and their team has since added a sixth band without us.