Torque Converter

Take two shallow bowls and fit radial vanes inside each — straight blades running from centre to rim, 8 to 12 of them. Card, thin ply or 3D-printed vanes all work.

Mount them on separate shafts, facing each other about 10 mm apart, inside a container.

These are the impeller and the turbine. Note what you have not built: any connection between the two shafts.

Spin the input shaft with the container empty.

The output does not move. Air carries almost no momentum, so almost no torque crosses the gap.

Do this before filling it. It proves the shafts are genuinely independent, so that whatever happens next is the fluid's doing and not friction in a bearing.

Fill until both vane sets are submerged. Water shows the effect clearly; oil is what a real unit uses, because it does not corrode the parts and it carries away heat.

Spin the input again. The output now turns. Nothing was connected — the only change is a fluid.

Count input and output revolutions over 10 seconds. Compute slip = (input − output) / input × 100%.

Expect substantial slip unloaded and more as you load the output with a finger.

Slip is not a defect here, it is the mechanism. Fluid drive only transmits torque when there is a speed difference to move oil across the gap — at zero slip there is no circulation and no drive. A real coupling runs at 2–4% slip, which is also exactly why it can never be 100% efficient.

Hold the output shaft completely still and keep the input spinning for a minute.

Nothing grinds, nothing burns, nothing wears. Feel the water: it is warmer. The power going in is turning into heat, and nothing else.

Do the same to a friction clutch and you destroy it. This is why an automatic car can sit in gear at a red light with the engine running — the converter is stalled, and stalling is a normal operating state rather than a failure.

The patent. Granted 24 June 1905 to Hermann Föttinger for a hydraulic transmission using one or more driving and one or more driven turbine wheels to transfer power between adjacent shafts. Commonly cited as DE 221422 and DE 238804. ⚠ Those numbers are from secondary sources and we have not verified them against the documents; the date and the substance are well attested, the numbers are not confirmed here.

It was a shipyard problem, not a car problem. Föttinger worked for Stettiner Vulkan. The steam turbine had just made marine propulsion far more efficient and utterly unsuited to driving a propeller directly. His device let a fast turbine drive a slow screw with no gear teeth at all.

Coupling versus converter — a real distinction. Two elements give a fluid coupling: it transmits torque unchanged, minus slip. Add a stationary stator between turbine outlet and impeller inlet and the returning oil is redirected to help the impeller instead of fighting it. Now output torque can exceed input torque, typically 2–2.5× at stall. Only the three-element version is properly a torque converter, and the two terms are used interchangeably far more often than they should be.

Where it went. Marine drives, then diesel locomotives, then — decisively — the automatic gearbox. Chrysler's Fluid Drive and GM's Hydra-Matic put Föttinger's shipyard solution under millions of cars. The modern refinement is the lock-up clutch: once cruising, a clutch bolts input to output directly and takes the fluid out of the loop, because the 2–4% slip that makes the device work is pure waste once you no longer need it.