In 1952, Avro Canada engineer John "Jack" Frost — fresh off the CF-100 interceptor — proposed something so audacious it sounded like pulp science fiction: a circular, vertical-takeoff aircraft using a single large turbine to drive a peripheral jet, generating lift via the Coandă effect. The U.S. Army and Air Force, panicked by rumors of Soviet VTOL research, funded it under the joint designation WS-606A. Two prototypes were built at Avro's Malton, Ontario plant. The first flew on November 12, 1959.

The Avrocar was 18 feet in diameter, weighed 5,650 pounds empty, and used three Continental J69-T-9 turbojets feeding a central "turborotor" — a 124-blade fan that exhausted air through an annular nozzle around the disc's rim. By modulating the nozzle's vector via a ring of "focusing" vanes, the craft was supposed to transition from hover to forward flight at speeds up to Mach 0.5, while climbing to 10,000 feet.

It did none of those things. In flight testing at NASA Ames, the Avrocar exhibited a vicious instability called "hubcapping" — uncontrolled pitch and roll oscillations once it climbed more than about three feet off the ground, where it lost ground effect. Top speed was 35 mph. The program was canceled in December 1961 after $10 million in spending (roughly $110M in 2026 dollars). Both prototypes survive — one at the U.S. Army Transportation Museum in Fort Eustis, the other at the Smithsonian's Udvar-Hazy Center.

Why it failed: Frost's team simply didn't have the tools. The Coandă-effect lift distribution around the rim was wildly nonlinear; computational fluid dynamics didn't exist yet, and wind tunnel data couldn't capture the unsteady aerodynamics of a saucer hovering near the ground. The turborotor's mechanical complexity caused vibration at the focusing ring, and the J69s couldn't supply enough mass flow for the design lift. The control system was open-loop mechanical linkages — there was no way to damp the inherent instability fast enough.

Why it's viable now: Every single failure mode the Avrocar exhibited is a solved problem in 2026.

• Unsteady aerodynamics: Modern Large-Eddy Simulation can model rim-jet Coandă flow at full resolution. NASA's FUN3D and the open-source SU2 solver routinely handle exactly this kind of separated-flow case.
• Stability: The "hubcapping" instability is functionally identical to what quadcopters experience — and is trivially controlled today by IMU-driven flight controllers running at 1 kHz. A Pixhawk-class autopilot would have saved the Avrocar.
• Propulsion: Electric ducted fans (the kind powering Joby's S4 and Lilium's jet) deliver higher disc loading at a fraction of the J69's weight and noise. Distributed electric propulsion lets you put dozens of small fans around the rim instead of one fragile turborotor.
• Materials: The Avrocar's plenum was riveted aluminum. A carbon-fiber monocoque would cut structural weight by 40% and eliminate the vibration coupling that ate the focusing ring.

A modern Avrocar — call it an "annular eVTOL" — has real advantages over quadcopter geometries: the disc shape gives you huge usable lift area for the footprint, the rim-jet eliminates exposed rotors (a safety and noise win), and the flat profile is genuinely useful for shipboard or rooftop operations. The Pentagon's saucer wasn't a joke. It was a control system problem wearing a science-fiction costume, and we now have the controllers.