// Showcase

Acceleration Sensor with Built-In Tilt Compensation

An acceleration sensor for rail traffic, measuring range 2 m/s².

01Starting point

An acceleration sensor for rail traffic, measuring range 2 m/s². That sounds harmless until you read the second boundary condition: maximum gradient 7 per cent, which is 4 degrees. At 4 degrees of tilt a gravity component of 0.68 m/s² acts along the direction of travel – 34 per cent of the measuring range, while standing still. A sensor that cannot tell a slope from an acceleration is out by a third on every ramp. And not as noise, but as a slowly drifting offset that correlates with the route.

02Approach

With four feasibility studies in eleven days, three of which end in a no – but a no with a number attached. Two inclination sensors in one plane: difference and sum signals cannot be separated, compensation impossible. Damped pendulum: at best ± 0.38 per cent residual error, and the damping is strongly temperature dependent whichever method is used. Two inclination sensors at 90 degrees: solvable on paper. And finally the variant that was built – force measurement on a preloaded standing mass. The sum of the two support forces gives the tilt, the corrected difference gives the acceleration. One arrangement, two measured quantities.

03Result

A sensor that computes its own tilt error instead of delegating it to a second instrument or to a track database. And a design that was settled before the first cut: five mechanical arrangements were compared for resolution, weight, stability and cross sensitivity, and the best of them simulated in five load cases.

What's inside

Four measuring cells at 45 degrees, water-jet cut spring beams with strain gauges, a current-driven full bridge with a programmable signal conditioner. Two channels, each 0–10 V and 4–20 mA, 24 V supply, eight-pin connector. The ± 4 degrees of tilt are mapped onto 2 to 8 V, that is 0.75 V per degree. Moving mass: 0.3 kg. Plus commissioning, training and a data sheet.

And what happened next

The most interesting decision was the step from two measuring cells to four. The reason: in simulation a lateral acceleration of 1 g produces a cell deflection of 0.7 µm, while the entire tilt range accounts for only 5.2 µm. One eighth of the quantity being corrected, arriving as a disturbance – on a rail vehicle that is not an exception, it is the curve the train is currently in. With four cells it can be compensated; with two it cannot.

All showcases

Get in touch

We are ready for new challenges

You bring the challenge. We develop the solution. From the first idea through concept and development to the validated machine, we guide your project.