Achtung:

Sie haben Javascript deaktiviert!
Sie haben versucht eine Funktion zu nutzen, die nur mit Javascript möglich ist. Um sämtliche Funktionalitäten unserer Internetseite zu nutzen, aktivieren Sie bitte Javascript in Ihrem Browser.

Measurement Engineering Group (EMT)
Prof. Dr.-Ing. Bernd Henning
Show image information

Research topic Sensor design

Overview

When designing and optimizing measurement systems many scientifically disciplines have to be combined to find a result to fulfill the objective. Because of this, besides a sensor, also the sensor interface electronics have to be examined. For example, by controlling a transmitter cleverly it can be used as a receiver at the same time without having to modify the actual transducer.

In addition to the electronics, the used materials can cause stronger measurement effects or even enable the use of certain phenomenons. For instance the choice of adapted materials when transmitting sound through a functional layer is crucial.

Also the geometrical dimensions can be changed during the optimization process to strengthen wanted effects or suppressing unwanted ones, such as reducing identified disturbing reflections.

All these aspects play a fundamental role when generating a model and afterwards designing the sensor. They have to be seen in their entirety.

 

Sensor electronics

When implementing new systems not only the sensor construction itself but also the property of the connected electric network is a very important aspect. A part of the research done by the Measurement Engineering Group is in the area of developing a sensor interface circuit for simultaneous transmitting and receiving of an ultrasonic transducer.

In the course of development different approaches, such as a modified directional coupler and an adjustable phase shifter, are being investigated. These solutions enable us to isolate the received signal from the transmitted signal, while the transducer is still sending. So far it has been possible to absorb the transmitted signal in the received signal by -61dB, which has shown to be sufficient for a lot of applications.

Model generation

There are many ways of generating models for FEM-simulation. Depending on the starting point existing CAD-data can be used, alternatively parameters can be computed from an existing object or a new model can be generated. For this different tools are used.

Commercial pre- and postprocessor GiD

When data is already existent in any kind of CAD-format, the Measurement Engineering Group uses different commercial CAD tools to prepare the data for numeric acoustical simulations. Also the necessary discretization is done, so that both consistent structural and mesh models are available. The actual simulation is done with a commercial finite element software package, which is specialized to calculate electro-mechanical problems, even if they are nonlinear. 

Script-based model generation

Just in case of optimization the analysis of a static model often is not sufficient since the influence of single model parameters can be judged not or only on a very small scale. As a solution the Measurement Engineering Group has developed an individual toolbox for Matlab®, which makes a variable generation process for models possible. Using this toolbox, a range for each of the model's properties (material, size, stimulus etc.) can be selected which are then evaluated with the help of different optimization strategies. This way individual or multiple properties can be varied explicitly and wanted influences can be identified. This system generates a numeric simulation model, which is then calculated with the help of a commercial solver, just as in the static case.

Sensor optimization

Besides the analysis of existing sensors, when evaluating the state of an existent system, the optimization is becoming more and more important.

An optimal design point can be found at selected parameters in a predefined parameter range of several model sizes. An important role at the efficient calculation of such an optimum is the choice of a suitable optimization strategy. Methods like simulated annealing, successive approximation, artificial neural networks (KNN), principal component analysis (PCA) or genetic algorithms offer their applicability depending on wanted objectives.

The University for the Information Society