Software tool speeds up gear design

A software simulator for automotive Hall-effect sensors has dramatically reduced development times and eased interaction with customers.

A software simulator for Allegro MicroSystems' automotive Hall-effect sensors has dramatically reduced development times and eased interaction with customers.

Developed by engineers at Allegro's Europe Sensor Applications Centre in Annecy, France, the new software produces behavioural models of gear-tooth sensors based on design schematics.

This means that Allegro can optimise the chip architecture before going through expensive silicon iterations.

The simulation tool also allows for immediate and fast customer support if a sensor in the field exhibits anomalous behaviour.

All the customer has to do is to provide Allegro with the analogue differential magnetic sensor signal, which is then imported into the simulator prior to running the behavioural part model on real customer data to see if the problem can be reproduced.

'We felt the need for such a dedicated tool because traditional EDA design tools, although accurate, were far too complex and very slow for use in applications support and for doing high-level algorithm evaluations', comments Dr Andreas Friedrich, European Sensor Applications Manager.

'Allegro MicroSystems is a world leader in automotive Hall-effect gear-tooth sensors which acquire speed, position and direction information from rotating ferrous targets in cam, crank, transmission and wheel speed sensing applications.

The signals produced by the sensors are typically fed to automotive engine control units or antilock braking systems.

A key element of the Allegro sensor philosophy is the incorporation of advanced signal-processing algorithms that optimise the performance of the devices and minimise the effect of external influences such as vibration.

To develop and debug these algorithms, Allegro engineers need to simulate all possible driving scenarios, such as driving over potholes, navigating sharp corners, and reacting to ferrous materials flying in front of the sensor.

In order to develop the simulator software, Allegro's engineering team looked at a number of standard tools, evaluating ease of use, model development time, simulation time, and how well the simulation results matched the behaviour of the actual part.

They also considered cost and flexibility in tool deployment.

Based on their findings, Allegro selected Matlab and Simulink from MathWorks to design and validate system models.

Initial evaluation of high-level algorithms is done through Matlab scripts.

This leads to the shortest simulation times and allows for quick validation of new algorithms.

The limitation of this approach is set by the subsequent conversion of the Matlab script into actual device schematics.

Using Simulink, the design engineers develop and simulate device algorithms before realising the IC at the transistor level.

Later, they compare simulations between Simulink and their IC design tool to validate their models before implementation.

Allegro engineers also model existing sensors and compare the simulations with sensors in development to validate performance.

These models enable systems engineers to help customers design systems with Allegro sensors.

After the initial evaluation phase using Matlab scripts, the chip architecture is finalised by simulating and developing the behavioural models with Simulink before implementing the transistor-level models with an IC design tool.

During the initial design phase, designers simulate basic inputs to devices such as sine, triangle and square waves.

Systems engineers continue to also use the simulator models in conjunction with a laboratory mapping station to compare actual magnetic waveforms with real-world targets.

This enables the design team to account for real-world target design considerations when evaluating new algorithms.

In addition to reducing the time to market for new devices such as Allegro's next-generation engine-management sensors, the simulator software has greatly reduced the risk of algorithm flaws, while also allowing product engineers to simulate how the part will behave when configured during testing.

Back to top